Extends Ghost to include an abstraction for OpenXR, which I refer to as
Ghost-XR. Such an API is the base for the following commit, which introduces VR
support to Blender.
Main features:
* Simple and high-level interface for Blender specific code to call.
* Extensible for muliple graphics backends, currently OpenGL and a DirectX
compatibility layer are supported.
* Carefully designed error handling strategy allowing Blender to handle errors
gracefully and with useful error messages.
* OpenXR extension and API-layer management.
* OpenXR session management.
* Basic OpenXR event management.
* Debug utilities for Ghost-XR and OpenXR
For more information on this API, check
https://wiki.blender.org/wiki/Source/Interface/XR.
Reviewed by: Brecht Van Lommel
Differential Revision: https://developer.blender.org/D6188
The initial code from earlier from today didn't really work reliable
since it is not possible to apply virtual modifiers but not the real
multires one (in a situation like mesh with shapekeys and multires).
New code uses less memory and has better performance for the case
when there are actual modifiers leading the multires. The case when
there is only multires will not be as performant as possible at this
moment.
The parenting was using the old logic, but with new engine the draw is done using eval data.
Fixed the depsgraph relationship missing with bones to get an update when the bone is transformed.
Also fixed Snap cursor to Selected
BKE_ocean.h uses the bool type without including stdbool.h
counting on someone else including that before it.
With D6811 enabling automatic sorting of the includes
this can no longer be counted on. This changes includes
stdbool.h in BKE_ocean.h so it can build without being
depended on others including the right headers before it.
Fix crash when the operator search is used while no active object exists. The cause of the issue is an attempt to dereference `ob` when it is `NULL`. Therefore this patch checks the return value of `SCULPT_mode_poll()` first, to ensure that `ob` isn't `NULL`.
Reviewed By: pablodp606
Maniphest Tasks: T74838
Differential Revision: https://developer.blender.org/D7156
The code would have break the first (deform only) modifiers
once the index is reached, but it will not prevent second
loop (over remaining modifiers) from run.
This was applying deform modifier twice in some conditions:
having single deform modifier and calculating deformed mesh
up to the first modifier (index=0).
Use full argument name.
Also order arguments in the generosity order: from depsgraph
(which has everything) to object (which contains multires)
specific multires modifier.
The feature is hidden behind an experimental option, you'll have to
enable it in the preferences to try it.
This feature is not yet considered fully stable, crashes may happen, as
well as .blend file corruptions (very unlikely, but still possible).
In a nutshell, the ideas behind this code are to:
* Detect unchanged IDs across an undo step.
* Reuse as much as possible existing IDs memory, even when its content
did change.
* Re-use existing depsgraphs instead of building new ones from scratch.
* Store accumulated recalc flags, to avoid needless re-compute of things
that did not change, when the ID itself is detected as modified.
See T60695 and D6580 for more technical details.
Unreported Crash. When hidden faces are active (retopology) the depth
test could fail as the default framebuffers aren't set. This patch will
check if we are rendering a depth only and skip the clearing of the
buffer.
With this change baking jobs will be aborted faster. The user will not have to wait for the current frame to finish baking. The bake job will exit early and discard the incomplete frame.
This option existed already and was just hidden in the UI. With the new fluids system though, it will only be used for rendering - and not to optimize the cache.
Extract will steal all depsgraphs currently stored in given bmain, and
restore will put them back in place, using scene and viewlayers as keys.
Preliminary work for undo-speedup.
Part of T60695/D6580.
Mask extract modifies the topology when adding the boundary loop, so
previous face sets may not be correct in the new mesh. Remove the face
sets datalayer and let sculpt mode rebuild it when entering sculpt mode
in the new object.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7129
The previous behaivour didn't make sense as sculpt mode was still active
when switching to the new object, so it was rendering inconrrectly.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7130
The face set color variable needs to be declared inside of the loop in
order to reset it per iteration.
Reviewed By: jbakker
Maniphest Tasks: T74626
Differential Revision: https://developer.blender.org/D7096
Small UX fix. When hidding everything but the active face set with H,
the last stroke lotation center can be in a part of the model that is
hidden, so viewport navigation becomes confusing until you start a new
stroke on the visible face set. Now the viewport navigation rotation
center is updated to the active vertex when using a visibility operation
that uses it, so it always rotates using the visible face set as the
origin.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7137
`ANIM_animdata_update()` did not sort grease pencil frames. A
pre-existing comment stated this wouldn't be necessary as
`posttrans_gpd_clean()` already does this. However, this is only
applicable when the change is performed via the transform system. The
mirror operator doesn't call `posttrans_gpd_clean()`, invalidating the
assumption in the comment.
I moved the sorting code into `BKE_gpencil_layer_frames_sort()`, which
is now called from both `ANIM_animdata_update()` and
`posttrans_gpd_clean()`.
This commit adds 7 themes submitted by the community on Devtalk. These themes both serve specific purposes, provide a greater variety in look & feel, and serve as welcoming homes for users coming on board from other packages. This is the initial commit, but these themes can be continuously updated over time to fix issues and keep them up to date with changes.
Thanks to all contributors, and in particular the makes of the picked themes: Pierre Schiller, Edward Agwi, Vojtěch Lacina, Michail Soluyanov, Jason van Gumster, Mr Wax Police & Jonathan Lampel.
An overview is here: https://developer.blender.org/T74360
This patch fixes various problems of alignment and element backdrops for
the animation channels drawing, mainly in the Graph editor but also for
grease pencil and mask layers in the Dope Sheet.
Reviewed By: billreynish, sybren
Differential Revision: https://developer.blender.org/D5204
This fix jitter of overlay and GPencil. But I'm not sure this should
be the responsibility of the subsequent draw engines or the responsibility
of the current engine to reset the view.
Test specs are read from strings, and there was a comma instead
of a decimal point, and then an extra decimal point in the Quad0 test.
This test has been flaky on Windows buildbot. Perhaps this is why.
- Remove Shift for drawing poly-lines, just as in the default keymap
- Use consistent hotkeys for size and strength radial controls
- Fix some bugs/missing items from the GP merge
selections
Previously this was only supported in single click selections, doing an
island selection with box/circle/lasso would just select individual
vertices instead. Now selects islands properly.
This also unifys some logic between box/circle/lasso:
- use early selection test from lasso [makes things faster] in box/
circle
- circle wasnt checking visible face
- Add icons for Sculpt Cloth, Clay Thumb and Draw Face Sets, as well as GP Tint, Replace and Transform Fill tools
- Tweak icons for Sculpt Rotate, Pinch, Multiplane Scrape, Inflate, Blob, Draw Sharp, based on feedback on Devtalk
Fixed memory leak that showed up after the original issue (crash) had been fixed in 93ac4709eb. The fix ensures that light cache bakes free up GPU smoke textures and the smoke domain list correctly.
This commit also removes the workaround (f3a33a9298) that disabled light cache bakes for fluid objects.
The `check_is_animated()` function will be used by the upcoming Alembic
exporter as well. There is nothing USD-specific in the function.
No functional changes.
Owner of filelisting job was changed, without proper update of all
access/usages of that owner to reach the job, leading to failure of
timer removal from the WM, and attempt to double-free the job...
Caused by rB2c4dfbb00246ff.
The issue was that we were creating temporary mesh copies and storing
them in bmain and then later using BKE_mesh_nomain_to_mesh which would
add them to bmain once more (duplicates).
This would lead to crashes later as the custom data of the mesh could be
trashed quite easily.
Was happening when object does not have CD_MDISPS allocated yet.
Need to make sure totdisp and level is specified on CD_MDISPS data
prior to loading (as the load expects them to be properly set).
Part of the fix was to get gputexture to use an array to accomodate each
eye. This takes care of viewports showing individual Left or Right
views.
For the combined view the fix was in overlay_image.c:camera_background_images_stereo_setup.
Note 1: Referece images are still not supporting stereo.
Note 2: For painting, and getting image bindcode I'm hardcording a
single-view experience.
Note 3: Without D6922 stereo is too broken to even test this patch.
With D6922 + this patch the fullscreen modes work (anaglyph/interlace
not yet).
Differential Revision: D7143
This change fixes artifacts produced by these operations.
On a technical aspect this is done by porting all of the operations
to the new subdivision surface implementation which ensures that
tangent space used to evaluate modifier and those operations is
exactly the same (before modifier will use new code and the operations
will still use an old one).
The next step is to get sculpting on a non-top level to work, and
that actually requires fixes in the undo system.
Makes it work better "out of the box" for irregular topology like
Suzanne mesh.
There might be some performance impact on non-regular meshes, but
those are not very common usecase for multires and for those its
always possible to lower the quality if needed.
Loose vertices and vertices of loose edges callback was not working
correct if some of other callbacks were set to NULL.
Was caused by missing bitmask set in the callbacks which were set
to NULL.
Allow to mark individual vertices as infinitely sharp even if there is
no full topology and no access to edges: infinite sharp vertices do not
need connectivity information.
This allows fast access to various arrays in the Python API.
Most notably, `image.pixels` can be accessed much more efficiently now.
**Benchmark**
Below are the results of a benchmark that compares different ways to
set/get all pixel values. I do the tests on 2048x2048 rgba images.
The benchmark tests the following dimensions:
- Byte vs. float per color channel
- Python list vs. numpy array containing floats
- `foreach_set` (new) vs. `image.pixels = ...` (old)
```
Pixel amount: 2048 * 2048 = 4.194.304
Byte buffer size: 16.8 mb
Float buffer size: 67.1 mb
Set pixel colors:
byte - new - list: 271 ms
byte - new - buffer: 29 ms
byte - old - list: 350 ms
byte - old - buffer: 2900 ms
float - new - list: 249 ms
float - new - buffer: 8 ms
float - old - list: 330 ms
float - old - buffer: 2880 ms
Get pixel colors:
byte - list: 128 ms
byte - buffer: 9 ms
float - list: 125 ms
float - buffer: 8 ms
```
**Observations**
The best set and get speed can be achieved with buffers and a float image,
at the cost of higher memory consumption. Furthermore, using buffers when
using `pixels = ...` is incredibly slow, because it is not optimized.
Optimizing this is possible, but might not be trivial (there were multiple
attempts afaik).
Float images are faster due to overhead introduced by the api for byte images.
If I profiled it correctly, a lot of time is spend in the `[0, 1] -> {0, ..., 255}`
conversion. The functions doing that conversion is `unit_float_to_uchar_clamp`.
While I have an idea on how it can be optimized, I do not know if it can be done
without changing its functionality slightly. Performance wise the best solution
would be to not do this conversion at all and accept byte input from the api
user directly, but that seems to be a more involved task as well.
Differential Revision: https://developer.blender.org/D7053
Reviewers: JacquesLucke, mont29
The code was copied from the Alembic exporter, and some of the options are
no longer used.
Not updating the Alembic exporter itself, as this will be done in a much
larger rewrite.
Both are doing almost the same and can be merged. This reduce complexity for user and less code to maintain.
Reviewed By: mendio, pepeland, fclem
Differential Revision: https://developer.blender.org/D7134
The Deform modifiers was reallocating buffers that only fit the vertices
of the inner loop. This patch first counts the maximum needed buffer and
allocates one.
When using the daily dweebs animation file the playback performance went
from 0.66 fps to 0.93 fps.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D7132
If one of the faces connected to a vertex is hidden in the face sets, we
can assume that the vertex is part of a boundary edge, so it should be
cosidered like that in all automasking and edge detection functions.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7126
The idea of the visibility system is that tools should behave like
hidden vertices do not exist, so the flood fill operation should ignore
hidden vertices for all operators.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7125
The face set ID is sequential, so implementing this was straightforward.
Suggested by Jeroen Bakker
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7123
The face_set_set function which sets a face sets given a vertex index
can ignore all modifications to hidden face sets, so we can skip all
vertex visibility checks outside that function. This makes the code
faster, simpler and fixes multiple bugs.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7122
The previous solution was also working fine as the white face set has no
meaning, but now it is a little bit more random. Also, bigger face sets
have more chance of getting the white color.
Reviewed By: jbakker
Maniphest Tasks: T74646
Differential Revision: https://developer.blender.org/D7111
The mesh provided in the report has 0 area faces and overlapping
vertices, causing the relax code to fail when calculating the plane to
constraint the vertex movement. Now it works fine both in the brush and
in the mesh filter.
Reviewed By: jbakker
Maniphest Tasks: T74648
Differential Revision: https://developer.blender.org/D7109
If the relax mesh filter was used on a non manifold mesh with open
boundaries, all the vertices were relaxed and the mesh was shrinking.
This was an unintended behavior that was making the filter unusable with
these meshes.
The mesh filter is now initializing an automasking buffer using the same
boundary automasking function from the brush code. Now edges are
preserved and the relax filter works as it should.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7097
Embedded data should always be considered as outside of Main database
here.
Note that it's a bit of an edge case to decide whether those should
always have their `LIB_TAG_NOMAIN` set too, or not? For now, let's keep
things as they are here.
The behaviour of GP layers is the same as annotation layers: they show
in the dope sheet regardless of whether they have frames or not. This is
easily resolved by adding some extra filtering.
Part of the function was following an "if-ok: do-this" pattern, and then
mid-function switched to a "if-bad: skip" pattern. The function now just
uses the latter.
No functional changes.
Some draw state bits are mutual exclusive. This patch will free some
draw state bits by packing the mutual exclusive bits in a mask.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D7088
- Use typed enum for line_data callback.
- Pass in 'const' arguments where possible.
- Use 'r_' prefix for return arguments.
- Remove unused return value from line_get callback.
- Remove redundant casts.
This extends the ocean modifier to add new spectra
(Pierson-Moskowitz, Jonswap, TMA).
These models are very different to the Phillips spectrum.
They are intended for more established,
large area, oceans and/or shallow water situations.
Bullet currently generates the majority of the warnings
on windows all of them are silly. This patch disables
all warns from bullet for now.
We should revisit this if/when we update bullet
to a newer version.
Reviewed By: sergey brecht
Differential Revision: https://developer.blender.org/D7118
`google::LogMessageFatal::~LogMessageFatal` calls `abort`
which MSVC correctly identifies as 'not returning'
and warns about a potential memory leak.
Given this is intended behaviour and glog is not overly
concerned with shutting down the process nicely, we
can safely ignore this warning.
Rather than passing around void pointers, various Blender image sources now
subclass this. OIIO is also just another type of image loader.
Also fixes T67718: Cycles viewport render crash editing point density settings
This is mostly straightforward, but required some refactoring to ensure that
the default volume material does not always turn on the volume feature for GPU
rendering.
printf is called for a size_t (64 bit on x64) type
but the formatter is `%lu` (32 bit) leading to a
warning with MSVC.
`%zu` is the appropriate formatter.
Was happening when having compositor open with Viewer node attached
directly to Render Layers output.
There were two things involved here:
1. The code which was storing CoW-ed versions of IDs was checking all
IDs for whether they are expanded or not. This was causing access
of freed memory for deleted IDs which do not need CoW (such as IM).
Simple fix: store ID type as a scalar and use early check before
doing more elaborate check based on accessing fields of id_cow.
2. The code which was ensuring view layer pointer is doing CoW for
scene. This isn't an issue on its own, but scene might have an
embedded ID such as compositor which was actually traversed by the
ID remap routines. This was causing remapping procedure to go into
non-updated copy of compositor, accessing freed Viewer image ID.
Solved by not recursing into embedded IDs for datablocks as those
are supposed to have own copy-on-write operations which takes care
of re-mapping.
Reported my Bastien, and also pair-coded with him.
For debug builds we link the against the release mode libs
for C based libraries, which are technically linked against
a different CRT, which the linker will implicitly try to link.
Which results in a linker warning about mixing the debug/release CRT.
This patch prevents the implicit linking of the release
CRT in debug configurations for sub projects that had issues
with it.
We implement cubemap array support for EEVEE's lightcache reflection probes.
This removes stretched texels and bottom hemisphere seams artifacts caused
by the octahedral projection previously used.
This introduce versioning code for the lightcache which will discard any
lightcache version that is not compatible.
Differential Revision: https://developer.blender.org/D7066
This patch is (almost) a complete rewrite of workbench engine.
The features remain unchanged but the code quality is greatly improved.
Hair shading is brighter but also more correct.
This also introduce the concept of `DRWShaderLibrary` to make a simple
include system inside the GLSL files.
Differential Revision: https://developer.blender.org/D7060
Casting a 64 bit pointer to a 32 bit DWORD gave 2 warnings.
Solved by storing the actual DWORD in the registry table.
Would have preferred to use a union, but C++ doesn't let you
initialize anything other than the first field, and C99 style
initializers are not supported until C++20, so this solution
will have to do until then.
Update Dopesheet header to include missing buttons, remove Scene Active only buttton and also removed duplicated search box.
The removed options come from old 2.7x version and they are not required now.
Reviewed By: mendio, pepeland
Differential Revision: https://developer.blender.org/D7107
Select Similar Group and Select Similar Shape had this issue since they
were added. Basically it assumes there is pose data which in some cases
it does not.
This has no user visible impact yet since smoke volumes only support a fixed
set of attributes, but will become important with the new volume object.
For GPU shader compilation, volume grids are now handled separately from
image textures. They are somewhere between a vertex attribute and an image
texture, basically an attribute that is stored as a texture.
Differential Revision: https://developer.blender.org/D6952
This is more in line with standard grids and means we don't have to make
many special exceptions in the upcoming change for arbitrary number of volume
grids support in Eevee.
The workbench shader was also changed to fix bugs where squared density was
used, and the smoke color would affect the density so that black smoke would
be invisible. This can change the look of smoke in workbench significantly.
When using the color grid when smoke has a constant color, the color grid
will no longer be premultiplied by the density. If the color is constant
we want to be able not to store a grid at all. This breaks one test for
Cycles and Eevee, but the setup in that test using a color without density
does not make sense. It suffers from artifacts since the unpremultiplied
color grid by itself will not have smooth boundaries.
Differential Revision: https://developer.blender.org/D6951
Issue revealed by own recent cleanup in rB8820ab4, and moticed by
@brecht, thanks.
Note that am not 100% sure whether we should allow call on lib_query
without a proper valid owner_id, for embedded data-blocks. But this can
be investifated later, so far things have been working like that.
'Private' can be a rather confusing term, especially when considering
its meaning in programming languages.
So now root node trees and master collections are 'embedded' IDs
instead.
The former is always a real, in-Main data-block, while the later, when
different, should be one of those embedded 'private' IDs (like root node
ree or master collection).
- Move gpuPush/Pop from GPU_draw.h into GPU_state.h
as this is for pushing/popping state.
- Add 'GPU_STANDALONE' define, to bypass use of user-preferences
for theme colors and pixelsize, as well as pbvh init/free functions.
Needed to get GHOST tests working again.
I've added a very minimal mesh validation before the Alembic mesh is actually
converted to a Blender mesh. This prevents a specific crash with an example
file attached to T74200.
A default face set color was not being set in previously saved meshes,
so it will always render the default face set with a random color until
the colors were recalculated.
Bump subversion to 283.8
Reviewed By: dfelinto
Maniphest Tasks: T74613
Differential Revision: https://developer.blender.org/D7094
Good to be explicit about the fact that we may still use the pixel
property sub-type when DPI will be applied.
See comments in https://developer.blender.org/D7077.
The 3D view grid size property is a multiplier, not the size of the grid itself.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7085
The Extrapolation Mode menu in the graph editor channel list was
incorrectly using the operator for the Action/Dopesheet editor. The
operator was even missing in the generic dopesheet hotkeys, so
{key Shift E} was listed as hotkey but didn't work. This is now all
fixed.
EasingType was implemented rBdaccaa713b6e for the GraphEditor (but never
made it to the Dopesheet). If you can select Easing Mode in the
DopeSheet, then you should also be able to select the associated Easing
Type.
Thanks @lichtwerk for the initial implementation.
Maniphest Tasks: T65076
Differential Revision: https://developer.blender.org/D6094
Caused by the introduction of UDIM (rBc30d6571bb47).
We need to make sure the tiles ImageUser is set up correctly [especially
the framenr], otherwise BKE_image_acquire_ibuf() and friends will fail
to find the correct ImBuf.
Also instead of initializing a minimal BKE_imageuser_default, now use
an appropriate ImageUser if avaliable and pass this around (instead of
just the tile_number). 2D painting can reuse the Image Editor ImageUser,
for 3D painting we still rely on a default ImageUser in most places, but
at least set the framenr correctly].
This also fixes crashes when doing image operations such as inverting or
resizing on images in a sequence in the Image Editor.
This also fixes color sampling (S) from the 3DView going wrong for image
sequences (would fallback to OpenGL sampling because an ImBuf could not
be found).
Maniphest Tasks: T74425
Differential Revision: https://developer.blender.org/D7022
This started happening after changing filter ID to 64 bit in rB2841b2be3949,
however there was a pre-existing error here in the comparison to detect updates
to filter flags.
When scaling the root bone of a rig to apply a global scale, the
corrective smooth modifier results in wrong deformation due to incorrect
scaling. The delta calculations are not taking into account any scale
value.
To fix it, a scale property is added to the modifier, allowing to set
manually the scale value for the deltas by simply multiplying the
vectors by this value. There is a similar implementation in Maya's Delta
Mush deformer. This property can be for example driven by the scale of
the root bone of the rig, to dynamically update when the animator scale
this bone.
Reviewed By: brecht, sybren
Differential Revision: https://developer.blender.org/D6622
When user used a custom background color, this color was also visible in
material preview mode, when the world opacity was less than 1. This
patch will draw the theme color as it was used to.
It seems like OSX drivers are using standard attributes for passing
gl_VertexID and gl_InstanceID to the vertex shader, and count them in the
limit of MAX_VERTEX_ATTRIBS.
This patch make sure to never use more than 13 attributes by packing some
attributes together.
I could not reproduce the issue, but it looks like it was produced by
this division by 0. In any case, the code here was wrong.
Reviewed By: jbakker
Maniphest Tasks: T74354
Differential Revision: https://developer.blender.org/D6987
This automasking option protects the open boundary edges of the mesh from the brush deformation. This is needed to sculpt cloths and it works nicely with the cloth brush.
It has a Propagation Steps property that controls the falloff of the mask from the edge.
Limitations:
- The automask is recalculated at the beginning of each stroke, creating a little bit of lag in high poly meshes, but it is not necessary. This can be fixed in the future by caching the edge distances, increasing a little bit the complexity of the code.
- The boundary vertex detection in meshes is not ideal and it fails with triangulated geometry, but it is the same as in the smooth brush. After fixing this, we should refactor the smooth brush to use the API and let the automasking option manually control the affected vertices.
- It does not work in Multires (it needs to be implemented in the API). The smooth brush in Multires is also not making boundary vertices.
- The falloff has a visible line artifact on grid patterns. We can smooth the final automasking factors several iterations, but it will make the initialization much slower. This can also be added in the future if we decided to cache the distances.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6705
This only applies to the case where preferences are opened as an editor in
a workspace, not with Edit > Preferences in a new window.
Differential Revision: https://developer.blender.org/D7001
This introduces a variable to store a face set ID which is going to be
rendered white. When initializing a mesh or randomizing the colors, this
variable gets updated to always render a white face set. This way the
face set overlay can be enabled without adding colors to the mesh if
face sets are not in use. After creating the first face set, new colors
are generated randomly like usual.
The face set stored as default does not have any special meaning for
tools or brushes, it just affects the rendering color.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7035
This enables a relax operation that works only on face sets boundaries,
which smooths the jagged edges that are produced when painting or
expanding face sets by sliding the topology without affecting the shape
of the mesh. This has many uses in hard surface sculpting for things
like sculpting panels or smoothing surfaces. It can also help when
working with remeshed topology as it makes the face sets looks better
and more organized if needed.
The operation is implemented as an Shift smooth in the Draw Face sets
tool, similar to the Slide/Relax tool.
The same operation is also available in the mesh filter to smooth all
the face sets boundaries uniformly or to smooth the face set under the
cursor with the Use Face Sets option.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7034
The cloth brush was not using the automasking values when calculating
the mask value on each vertex.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7083
Was crashing SculptSession data will not longer be valid if the total
number of polys is modified when rendering the mesh again.
This deletes all face sets in the mesh when slicing the mask. I'll try
to add code to generate a new face set in with faces that are created
when filling the holes, but for now this avoids the crash.
Reviewed By: brecht
Maniphest Tasks: T74500
Differential Revision: https://developer.blender.org/D7049
Create face sets by visibility needs to check if all face sets of a
vertex are visible to set the new face set. I renamed the functions to
make this more cleare in the API.
I also added a visibility check when creating by mask to avoid modifying
hidden areas.
Reviewed By: brecht
Maniphest Tasks: T74499
Differential Revision: https://developer.blender.org/D7048
Hardness is now a property implemented for all brushes, so this is no
longer needed.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7078
The operator was resetting the mask data when cancelling instead of the
face set data, so it was crashing because mask data was not available
when starting the operator in expand face set mode.
Reviewed By: brecht
Maniphest Tasks: T74492
Differential Revision: https://developer.blender.org/D7043
This way we can change the color generation easily if we want to improve
it in the future. I also added more values to randomize a little bit the
saturation and value of the colors, as previously it was too easy to get
similar colors when creating new faces, forcing you to use the randomize
colors more than necessary.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D7042
The problem happens because, in wireframe mode, `bool use_wire` is
always `true`, so the function that draws all edges is the called.
The solution is set `use_wire` as `false` when the mesh has no edges.
This matches the behavior of blender 2.79.
Reviewed By: fclem, brecht
Differential Revision: https://developer.blender.org/D7041
This has been long standing TODO...
Note that remaining usages of BKE_xxx_delete should all be carefully
checked for and utilmately nuked in favor of `BKE_id_delete()`, think we
still have quiet a few bugs hidden in those (code seems to usually
assume those functions do a full ID deletion, which is not the case).
Only covers direct usages of new callbacks from IDTypeInfo.
We still have a lot of those switch/case, many can probably go away
with minimal refactor now, but that will be for later.
In rB7c5a44c71f13 I changed the way transform matrices are loaded from
Alembic. Instead of having the Alembic importer convert matrices from
local (in the Alembic file) to World (to pass to the constraint handling
the animation of transforms), I set the constraint space to
`CONSTRAINT_SPACE_LOCAL`.
This worked thanks to rB7728bfd4c45c. However, that commit was reverted,
which meant that for parentless objects `CONSTRAINT_SPACE_LOCAL` no
longer means "local space".
The situation is resolved by setting the constraint to world space
again, and computing the world matrix in the Alembic importer.
Steps to reproduce were:
* Disable tool settings region in 3D View (View > Tool Settings)
* Split the 3D View and drag all the way down
The removed code doesn't seem to be needed anymore. Tested this on hiDPI
too, seems fine.
These kind of fixes are always tricky, so I wouldn't be surprised if
there are any issues caused by this.
The issue of T72253 was that the density threshold (RNA adapt_threshold) was considering cells as empty cells too early and thus also shrinking the domain too early. The fix for this is to use smaller threshold values for the adaptive domain. This fix gives more flexibility in the UI to do just that.
This commit is a full refactor of the grease pencil modules including Draw Engine, Modifiers, VFX, depsgraph update, improvements in operators and conversion of Sculpt and Weight paint tools to real brushes.
Also, a huge code cleanup has been done at all levels.
Thanks to @fclem for his work and yo @pepeland and @mendio for the testing and help in the development.
Differential Revision: https://developer.blender.org/D6293
Unused no-op operation nodes are not bound to a callback function, and
have no outgoing relations. Incoming relations of such nodes are removed
since ff60dd8b18. However, this was done
too broadly, causing too many relations to be lost and indirectly linked
objects to be unevaluated.
This commit introduces a `DEPSOP_FLAG_FAKE_USER` flag for operation
nodes, which indicates they are not to be removed, even when they appear
to be unused.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D7074
Apparently this happened when the object is in a flat view and has
customdata `CD_SCULPT_FACE_SETS`
Differential Revision: https://developer.blender.org/D7073
While it might be handy to have type-less functionality which is
similar to how C++ math is implemented it can not be easily achieved
with just preprocessor in a way which does not have side-effects on
wrong usage.
There macros where often used on a non-trivial expression, and there
was at least one usage where it was causing an actual side effect/bug
on Windows (see change around square_f(sh[index++]) in studiolight.c).
For such cases it is handy to have a function which is guaranteed to
have zero side-effects. The motivation behind actually removing the
macros is that there is already a way to do similar calculation. Also,
not having such macros is a way to guarantee that its usage is not
changed in a way which have side-effects and that it's not used as an
inspiration for cases where it should not be used.
Differential Revision: https://developer.blender.org/D7051
Currently, this change does not bring functional changes.
But it is necessary to extend the use of the snap system for gizmos,
since, after a Undo, the `depsgraph` pointed by the `snap_context`
has its memory invalidated.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D7013
This reverts commit 7728bfd4c4.
Although this brings back an inconsistency in the behaviour of
constraints on objects and bones, people were relying on the old
behaviour, and the new behaviour broke their files.
It is still desired to remove this inconsistency, but it will happen
more gradually.
This allows deleting both collections and objects in the outliner
selection at the same time. This only works using the keyboard shortcuts
(X or Delete).
While this works, a more robust solution should be implemented later to
allow deleting the whole selection from the context menu as well.
Missing changes from one of the renamings of the initial face sets
patch.
Reviewed By: brecht
Maniphest Tasks: T74513
Differential Revision: https://developer.blender.org/D7054
The initial iteration for all symmetry areas is always 0. We were using
1 for the main stroke, so it was 1 step behind.
This was broken for expanding masks and face sets, but with face sets it
is more noticeable.
Reviewed By: brecht
Maniphest Tasks: T74501
Differential Revision: https://developer.blender.org/D7050
By using PBVH_ITER_UNIQUE hidden vertices are skipped, like in the rest of
the brushes and tools.
Reviewed By: brecht
Maniphest Tasks: T74498
Differential Revision: https://developer.blender.org/D7047
This was particularly strange because we had a _free_data() function.
But still the one I replaced was of course the _free() one.
And we should rename the _free_data_ function later to avoid confusions.
Remove Windows special folder FOLDERID_Objects3D. Requires newer SDK.
Differential Revision: https://developer.blender.org/D7014
Reviewed by Brecht Van Lommel
The old convention was easy to confuse with ScrArea.
Part of https://developer.blender.org/T74432.
This is mostly a batch rename with some manual fixing. Only single word
variable names are changed, no prefixed/suffixed names.
Brecht van Lommel and Campbell Barton both gave me a green light for
this convention change.
Also ran clan clang format on affected files.
Part of https://developer.blender.org/T74429.
There's a chance that this causes some issues becaue in some cases we
change from getting the window from context to getting it from somewhere
else.
This moves the `alembic`, `avi`, `collada`, and `usd` modules into a common
`io` directory.
This also cleans up some `#include "../../{somedir}/{somefile}.h"` by
adding `../../io/{somedir}` to `CMakeLists.txt` and then just using
`#include "{somefile}.h"`.
No functional changes.
This is the companion of D7031. That patch adds a new DIMENSIONS node to
the depsgraph for each object that has geometry. However, this node is
only necessary when there are drivers using an object's dimensions as
variable. Since this is rare, it's easiest to remove these nodes after
they turn out to be unnecessary. This is what (almost) happens in this
patch.
Removing nodes from the depsgraph is hard, and there are no functions to
do this yet. Instead, this patch recursively removes all the incoming
relations from unused no-op nodes (i.e. no-op operation nodes without
outgoing connections). Actually removing the nodes will be left as a
future improvement.
I've tested this on a Spring file [1]. Here are there results of blender
--debug-depsgraph-time spring_02_055_A.eevee.blend and letting it run
for a while to stabilise the reported FPS:
master: 11.7 FPS
Just D7031: 11.7 FPS
Just D7033: 11.8 FPS
Both D7031 + D7033: 12.3 FPS
[1] https://cloud.blender.org/p/spring/5d30a1076249366fa1939cf1
Differential Revision: https://developer.blender.org/D7033
`(size_t)(int * int)` will actually cast overflown integer to size_t,
which isn't what was intended here. Correct thing would be to cast
in the following manner `(size_t)int * int`.
In this particular case can as well use function which is designed to
allocate an array of memory without overflow.
Since hair collisions were integrated with the cloth solver
(rBd42a7bbd6ea5), there are a couple of relevant settings which were not
exposed to the User:
- Collision Quality
- Minimum Distance (this was reported in T73842, default of 0.015m was
still limiting in certain scenarios - this can now be made smaller)
- Impulse clamping
- Collision collection
This will add a 'Collisions' panel to Hair Dynamics with those settings
Note: in contrast to 'real' cloth, self-collisions are not supported for
hair
Maniphest Tasks: T73842
Differential Revision: https://developer.blender.org/D7032
Make it obvious that this relates to being in sculpt mode.
ref T74489
Maniphest Tasks: T74489
Differential Revision: https://developer.blender.org/D7044
The docstring was likely copy-pasted, and the return type was not adjusted.
Since the `bpy.msgbus.subscribe_rna()` returns nothing, I just removed the
return type documentation.
Previously, `writedata` was used to store `bNodeSocket->default_value`.
There are a couple of issues with that:
* Breaks if someone tries to load the file on a big endian system (afaik).
* There is a `/* do not use for structs */` comment on `writedata`.
* Depends on `MEM_allocN_len` which should be avoided in my opinion.
* Now it is more apparent, that this should be handled by callbacks as well.
The part in `readfile.c` should work just fine still. I could also do a case
distinction there, but the code would be the same for every case for now.
Just `sock->default_value = newdataadr(fd, sock->default_value);`.
This might change, if we want to store more complex socket type specific data
that does not fit into a single struct.
Differential Revision: https://developer.blender.org/D7039
Reviewers: brecht
Currently the link limit of sockets is stored in bNodeSocket->limit.
This allows for a lot of flexibility, but is also very redundant.
In every case I've had to deal with so far, it would have "more correct"
to set the link limit per socket type and not per socket. I did not enforce
this constraint yet, because the link limit is exposed in the Python API,
which I did not want to break here.
In the future it might even make sense to only support only three kinds of link limits:
a) no links, b) at most one link, c) an arbitrary number links links. The other link
limits usually don't work well with tools (e.g. which link should be removed when a new
one is connected?) and is not used in practice. However, that is for another day.
Eventually, I would like to get rid of bNodeSocket->limit completely and replace it
either with fixed link limits or a callback in bNodeSocketType.
This patch consists of three parts:
**1. Support defining link limit in socket type**
This introduces a new `nodeSocketLinkLimit` function that serves as an indirection to
hide where the link limit of a socket is defined.
**2. Define link limits for builtin sockets on socket type**
Data sockets: one input, many outputs
Virtual sockets: one input, one output
Undefined sockets: many inputs, many outputs (to avoid that links are removed when the type of the socket is not known)
**3. Remove `bNodeSocketTemplate->limit`**
This wasn't used anymore after the second commit. Removing it simplifies socket definitions
in hundreds of places and removes a lot of redundancy.
Differential Revision: https://developer.blender.org/D7038
Reviewers: brecht
We would previously not store the transforms that were added to the
group transform node. This would lead to pointer to allocated memory
being lost and not freed.
This patch will make use of the eDrawType where it was used as a
variable or parameter name. The eObjectDrawType was renamed to eDrawType
as it is also used by `View3DShading.type`.
Motivation the functions get 3 different kind of flag parameters (ImBuf,
DrawType, OffscreenRendering) the naming of the flags were not clear,
leading to mistakes and unnecessary time spend debugging.
This was not set anywhere, code relies on MLOOPUV_VERTSEL everywhere.
BMLoopUV.select_edge was never returning True and wasnt updating uv
select state when set manually.
Afaict, there were no official Addons using this from python.
Resolves T65836
Maniphest Tasks: T65836
Differential Revision: https://developer.blender.org/D6772
There are two ways in which adding/deleting keyframes can fail, but only
one of those was handled. The other would be shown as a successful
add/delete of -2 keyframes.
The `ANIM_apply_keyingset()` returns a value that indicates the number of
changed channels (if nonnegative) or an error state (negative). In the
places where the return value was actually used, this value was stored in
a badly named variable.
This fixes an issue where drivers using `object.dimension` only add a dependency on `GEOMETRY` to the depsgraph, whereas they should also depend on `TRANSFORM`.
This patch adds a new no-op operation that depends on the geometry and transform components to the Parameters component.
An alternative implementation would be to have `RNANodeQuery::construct_node_identifier` return multiple node identifiers. However, this would spread throughout the depsgraph code and unnecessarily force many other functions to either return or handle multiple nodes where in 99.999% of the time a single node would suffice.
The new `DIMENSIONS` node is added for each object. An upcoming patch will go over all no-op operation nodes and remove them from the depsgraph. Since this is a more dangerous operation, it'll be reviewed separately.
Differential Revision: https://developer.blender.org/D7031
Since the bounding boxes are now also being used for effector objects, there needs to be a better name for them. Instead of calling them EmissionMap, which caters only to emission objects, they will now be called FluidObjectBB.
- Keymap items now have 'repeat' boolean which can be set
to make keymap items respond to key repeat events or not.
- Support for X11 & WIN32 (not macOS currently).
This allows for the possibility to perform actions while a key is held
and finish the action upon release.
Thanks to @Severin for review and WIN32 support.
- Use 'int' for counters instead of short.
- Use 'bool' instead of a counter when only a change is being detected.
- Use typed enum for keying set flags.
- Include in comments when a negate error code may be returned.
- Split 'verify_fcurve' into two functions:
ED_action_fcurve_ensure which adds the f-curve if needed.
ED_action_fcurve_find which returns NULL when not found.
Callers of ED_action_fcurve_find had unused 'group'
argument which has been removed.
- Rename verify_adt_action to ED_id_action_ensure
It had an argument to add data which was always true,
remove this instead of splitting in into a separate function.
- Use 'BKE_object_defgroup' prefix for object functions.
- Rename 'defvert_verify_index' to 'defvert_ensure_index'
since this adds the group if it isn't found.
Use only one font icon in File Browser for all platforms. Correct mistake to allow this icon to be used.
Differential Revision: https://developer.blender.org/D7037
Reviewed by Brecht Van Lommel
Face Sets are the new system to control the visibility state of the mesh in sculpt and paint modes. They are designed to work in modes where brushes are the primary way of interaction and they provide much more control when working with meshes with complex shapes and overlapping surfaces.
This initial commit includes:
- Sculpt Face Sets data structures and PBVH rendering.
- Face Set overlay and opacity controls.
- Sculpt Undo support.
- Remesher reprojection support. The visibility state of the mesh is also preserved when remeshing.
- Automasking and Mesh filter support.
- Mask expand operator mode to expand Face Sets (Shift + W) and flood fill areas by connectivity (press Ctrl while expanding).
- Sculpt Mode Face Sets and Visibility API.
- Sculpt Face Sets creation and visibility management operators.
- Operator to randomize the Face Sets colors.
- Draw Face Sets brush tool to create and edit the Face Sets. Drawing on the mesh creates a new Face Set. Pressing Ctrl before drawing modifies the Face Set under the brush at the beginning of the stroke.
- Updated keymap and menu to work with Face Sets from Sculpt Mode (H to toggle visibility, Alt + H to show all, Shit + H to hide).
- Pie menu on the W key with Face common Sets operations.
Know limitations:
- Multires support. The Face Sets and Visibility API needs to be implemented for Multires.
Reviewed By: jbakker, #user_interface, Severin
Differential Revision: https://developer.blender.org/D6070
Adding more Windows special folder locations, used when browsing or bookmarking.
Differential Revision: https://developer.blender.org/D7014
Reviewed by Brecht Van Lommel
Nicer icons for File Browser System and Favorites lists on Mac.
Differential Revision: https://developer.blender.org/D6398
Reviewed by Brecht Van Lommel
Adds a minimal DirectX 11 Ghost context, plus some shared DirectX-OpenGL
resource interface using the NV_DX_interop2 WGL extension. From what I
know, this should be available on modern GPUs. If not, it should fail
gracefully.
There should be no user visible changes at this point.
Needed for DirectX-only OpenXR platforms (e.g. Windows Mixed Reality). I
heard there are other use-cases as well though.
It's known that this currently fails on some AMD systems, but that seems
to be fixable.
Most of this comes from the 2019 GSoC project, "Core Support of Virtual
Reality Headsets through OpenXR"
(https://wiki.blender.org/wiki/User:Severin/GSoC-2019/).
Reviewed by: Jeroen Bakker, Ray Molenkam, Brecht van Lommel
Differential Revision: https://developer.blender.org/D6190
Since `std::deque` is used in a few areas of the Depsgraph, and an
upcoming patch adds one more, it's time it's considered as "commonly
used type".
No functional changes.
"session-wise" here mean while editing a same .blend file. So creating
or opening a new one will reset the uuid counter. This should avoid any
overflow in practice.
Only IDs added to Main database get an uuid, runtime-only ones are not
affected.
This is intended to provide undo with a way to find IDs across several
'memory realms' (undo speedup project).
No behavior change is expected from this commit itself.
Part of T60695.
Differential Revision: https://developer.blender.org/D7007
Enabling viewport denoising causes Cycles to use a multi-device, which always returned NULL when
asked for OSL memory and would subsequently crash. This fixes that by returning the correct OSL
memory pointer from the CPU device in the special viewport denoising multi-device.
Selecting certain child datablocks also selects the parent (e.g.
selecting a pose bone selects the armature). The base was selected, but
the outliner tree element was left unselected. The subsequent selection
sync would then deselect the parent base because it was not flagged as
selected in the outliner.
This led to issues like T74332 where selecting a pose bone in the
outliner did not show drivers in the driver editor unless the armature
was explicitly added to the selection afterwards.
The solution is to also flag the outliner elements as selected when
selecting parent bases.
Differential Revision: https://developer.blender.org/D7029
This feature takes some inspiration from
"RenderMan: An Advanced Path Tracing Architecture for Movie Rendering" and
"A Hierarchical Automatic Stopping Condition for Monte Carlo Global Illumination"
The basic principle is as follows:
While samples are being added to a pixel, the adaptive sampler writes half
of the samples to a separate buffer. This gives it two separate estimates
of the same pixel, and by comparing their difference it estimates convergence.
Once convergence drops below a given threshold, the pixel is considered done.
When a pixel has not converged yet and needs more samples than the minimum,
its immediate neighbors are also set to take more samples. This is done in order
to more reliably detect sharp features such as caustics. A 3x3 box filter that
is run periodically over the tile buffer is used for that purpose.
After a tile has finished rendering, the values of all passes are scaled as if
they were rendered with the full number of samples. This way, any code operating
on these buffers, for example the denoiser, does not need to be changed for
per-pixel sample counts.
Reviewed By: brecht, #cycles
Differential Revision: https://developer.blender.org/D4686
Introduce new IDTypeInfo structure.
Each ID type will have its own, with some minimal basic common info,
and ID management callbacks.
This patch only does it for Object type, for demo/testing purpose.
Moving all existing IDs is a goal of next "cleanup Friday".
Note that BKE_idcode features should then be merged back into BKE_idtype -
but this will have to be done later, once all ID types have been properly
converted to the new system.
Another later TODO might be to try and add callbacks for file read/write,
and lib_query ID usages looper.
This is part of T73719.
Thanks to @brecht for initial idea, and reviewing the patch.
Differential Revision: https://developer.blender.org/D6966
This was raised in T74017, the issue being that we point to `/dev`
version of the manual for the addons when we want to point to a specific
version instead.
Instead of manually updating the URL every release we can do this.
The `bl_info` for addons will need to be updated in the format of
`'doc_url':
{BLENDER_MANUAL_URL}/addons/import_export/scene_gltf2.html",`
Differential Revision: https://developer.blender.org/D6995
When running with debug enabled ('-d' argument),
warnings are printed for add-ons which are not yet updated.
Reminder to name things based on what they do,
not the technologies they use :)
- Use 'BKE_object_material_*', 'BKE_id_material_*' prefix
for functions that operate on Object and ID types.
- Use '_len' suffix for length (matching BLI naming).
- Use '_p' suffix for functions that return a pointer to values
where the value would typically be returned.
Functions renamed:
- BKE_object_material_resize was BKE_material_resize_object
- BKE_object_material_remap was BKE_material_remap_object
- BKE_object_material_remap_calc was BKE_material_remap_object_calc
- BKE_object_material_array_p was BKE_object_material_array
- BKE_object_material_len_p was BKE_object_material_num
- BKE_id_material_array_p was BKE_id_material_array
- BKE_id_material_len_p was BKE_id_material_num
- BKE_id_material_resize was BKE_material_resize_id
- BKE_id_material_append was BKE_material_append_id
- BKE_id_material_pop was BKE_material_pop_id
- BKE_id_material_clear was BKE_material_clear_id
This commit adds bounding box support for emission objects - similarly to flow objects. Before, each effector object had to iterate over the entire domain. Bake times of scenes with multiple obstacles improved significantly with this optimization.
Other improvements that were implemented alongside the bbox feature:
- Option for subframe sampling for effector objects
- Option to enable / disable effectors (can be animated)
- Optimization for static objects. If a flow or effector object does not move and the adaptive domain is not in use, the bake time will be optimized further by reusing the flow / effector grids from the previous frame (no recalculation).
not work for unlocking
Special case for when we do this on "lock":
- locked sequences are not in "selected_editable_sequences"
- use "selected_sequences" in that case
Maniphest Tasks: T74434
Differential Revision: https://developer.blender.org/D7023
The OpenXR-SDK contains utilities for using the OpenXR standard
(https://www.khronos.org/openxr/). Namely C-headers and a so called
"loader" to manage runtime linking to OpenXR platforms ("runtimes")
installed on the user's system.
The WITH_XR_OPENXR build option is disabled by default for now, as there
is no code using it yet. On macOS it will remain disabled for now, it's
untested and there's no OpenXR runtime in sight for it.
Some points on the OpenXR-SDK dependency:
* The repository is located at
https://github.com/KhronosGroup/OpenXR-SDK (Apache 2).
* Notes on updating the dependency:
https://wiki.blender.org/wiki/Source/OpenXR_SDK_Dependency
* It contains a bunch of generated files, for which the sources are in a
separate repository
(https://github.com/KhronosGroup/OpenXR-SDK-Source).
* We could use that other repo by default, but I'd rather go with the
simpler solution and allow people to opt in if they want advanced dev
features.
* We currently use the OpenXR loader lib from it and the headers.
* To use the injected OpenXR API-layers from the SDK (e.g. API
validation layers), the SDK needs to be compiled from this other
repository.
The extra "XR_" prefix in the build option is to avoid mix-ups of OpenXR
with OpenEXR.
Most of this comes from the 2019 GSoC project, "Core Support of Virtual
Reality Headsets through OpenXR"
(https://wiki.blender.org/wiki/User:Severin/GSoC-2019/).
Differential Revision: https://developer.blender.org/D6188
Reviewed by: Campbell Barton, Sergey Sharybin, Bastien Montagne, Ray
Molenkamp
After switching tools, the active vertex can be the same and the cursor
won't update the previews, so the pose_ik_chain_preview will be null.
This often happens in low poly meshes where chances of hovering the same
vertex are high.
Reviewed By: sergey
Maniphest Tasks: T74426
Differential Revision: https://developer.blender.org/D7021
Instead of using anonymous booleans flags, also allows to keep the same
behavior in all cases, without needing special handling from calling
code for our beloved oddballs object proxies...
fails
Exposed by rB50d5c03e2d14.
This was only a problem when _only_ 'Object Data' was made local.
If also e.g. 'Object' or 'Materials' was checked, these were already
making sure realations were updated [DEG_relations_tag_update(bmain)].
Now also call DEG_relations_tag_update(bmain) for the 'Object Data'
case.
I assume the underlying issue is that there is some ID_NEW_REMAP
happening in 'single_obdata_users()' -- including that for 'me-
>texcomesh', which might be responsible for the glitches in selection/
editing? Also not entirely sure why this wasnt a problem prior to
rB50d5c03e2d14.(I assume this was somewhat hidden by the fact this was
always called with a default action being nothing, the user would always
need to access the redo panel. So this might have been hidden by an Undo
step involved?)
Maniphest Tasks: T73797
Differential Revision: https://developer.blender.org/D7020
This would happen when done from editmode, on a mesh with any modifier,
after adding/removing geometry in editmode.
Similar to rBba0870713b9b (which did this for weightpaint and sculpt
already), ensure an evaluated depsgraph, otherwise
'runtime.mesh_deform_eval' would not be up to date causing problems
later.
Maniphest Tasks: T72028
Differential Revision: https://developer.blender.org/D7011
This does not affect the RNA access API, since how the boolean is read from
DNA abstracted away in the API.
Differential Revision: https://developer.blender.org/D7002
USD is being included before the blender headers.
USD includes TBB, which includes the windows headers
which define rad2 as a constant conflicting with a field
with that exact name in the MetaElem dna struct.
Added `-DWIN32_LEAN_AND_MEAN` to keep the windows headers
from defining rad2.
It was missing this local space conversion before calculating the rake
angle.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D7008
Metaballs are exported as evaluated polygonal mesh.
USDExporterContext::depsgraph was made non-const to allow calling
`BKE_mesh_new_from_object()`. The alternative was to make the depsgraph
parameter of that function const, but that would be too intrusive as
that would require even more functions to accept a const depsgraph
pointer.
Made an explicit call to the exec function be a blocking call and a call
from invoke (or otherwise) be nonblocking.
Reviewed By: Bastien
Differential Revision: http://developer.blender.org/D7006
Both partial derivatives were evaluated to 0 for a special vertex on
Suzanne model: this is happening on a vertex where two adjacent faces
with 2 common edges are connected (in the nose of Suzanne).
This was breaking multires in this point since tangent matrix is all
zeroes, and hence no displacement can be applied in that vertex.
into account for duplicating actions
Previously actions remained linked after duplication, now this is based
on the User Preferences (PreferencesEdit.use_duplicate_action).
note: default is ON here, so this changes default behavior of separate
operators.
First intuition was to respect _all_ preferences here (e.g. also
duplicating materials if chosen in the User Preferences) but after
consideration this is probably not what the User would expect from such
'modeling' opertions (e.g. separate by loose parts resulting in possibly
many duplicate materials)
Fixes T71038
Maniphest Tasks: T71038
Differential Revision: https://developer.blender.org/D6120
Object Scale was not taken into account.
This lead to reports like T74247 where the user scaled the envelope
distance and radii to the supposedly right values inthe viewport, but
these were actually 'wrong' under the hood. Assigning weights from bone
envelopes seemed like it would fail, but this code would actually take
the armature scaling into account when checking envelope distance and
weight.
ref T74247
Maniphest Tasks: T74247
Differential Revision: https://developer.blender.org/D6964
Recent refactor external dependencies handling (D6642)
improperly linked all library dependencies with public
linkage rather than interface linkage. Causing excessive
(re)builds of subprojects when not needed.
This patch restores the interface linkage.
Reviewed By: brecht sergey
Differential Revision: https://developer.blender.org/D6983
This sampling pattern is particularly suited to adaptive sampling, and will
be used for that upcoming feature.
Based on "Progressive Multi-Jittered Sample Sequences" by Per Christensen,
Andrew Kensler and Charlie Kilpatrick.
Ref D4686
This patch refactors arrow keys navigation to move properties and enum
to generic ED_select_utils.h and property to
WM_operator_properties_select_walk_direction()
No functional change
Reviewed By: Julian Eisel
Differential Revision: https://developer.blender.org/D4771
gizmos are too big
Caused by rB9cac5fa681c5 which fixed the handles for the camera border
gizmo (being much too small)-- which in turn was broken by
rB98e4fbc7cc9e.
Camera border gizmo sets up its matrix_space a bit differently [this is
why we are getting very small values in gizmo_calc_rect_view_scale --
needs more investigation], but instead of breaking it for Light, Image
and Compositor Backdrop, lets just compensate by scaling the Camera
Border Gizmo for now.
Maniphest Tasks: T74278
Differential Revision: https://developer.blender.org/D6991
As explained in T65568 by @LucaRood, the self collision system should exclude triangles that are connected by sewing springs.
Differential Revision: https://developer.blender.org/D6911
Added check to see if the mesh has any polygons at all.
If there are no polygons, the bvh tree data will consist of null
pointers, so skip internal springs if this is the case.
When using Material Previews not all uniform blocks were filled. This
patch will add the renderpass_block when drawing the background.
Note that I wasn't able to reproduce the issue on my system, but
according the the backtrace it most likely solves the issue. I let the
reporter test.
Clay Thumb was using a custom hardness implementation and value to make
it functional. Hardness is now a property of brushes and it is
implemented for all brushes, so this is no longer needed.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6981
Color management currently happens after the python handlers. This means
that all common drawing operations needs to happen in SRS. The
PyGPUOffscreen requested that the color management was applied, that
resulted into applying colormanagement twice.
This patch makes sure that PyGPUOffscreen.draw_view3d renders in SRS.
There are more code paths that needs to be checked that the correct
space is being requested.
When beveling architectural objects like baseboards or crown mouldings that
may consist of multiple islands, it's useful if the orientation is at least
conistent.
This changes the arbitrary decision of how the orientation should start at a
chain beginning to use the highest side of the profile in the Z direction.
Reviewed By: howardt
Differential Revision: https://developer.blender.org/D6946
While animating sound strips volume within the graph editor, it is possible to set keyframes to negative values.
The drawing code of waveforms wasn't clamping these values to zero and was instead drawing an "inverted" curve
Author: a.monti
Reviewed By: iss
Differential Revision: https://developer.blender.org/D6971
The hardness property moves the brush falloff towards the edges, making
the brush sharper.
This should be the intended way to control the brush falloff instead of
tweaking the falloff curve manually, as it can be mapped to pressure to
make the falloff variable during the stroke.
It is also a good idea to show in the UI that the custom curves is an
advance features and it should almost never be modified when sculpting/
painting unless you want to create some advanced effects. By modifying
the curves freely it is really easy to break the brushes and make them
produce artifacts. This needs to be done in a later after merging the
pending projects to reorganize all the brush properties accordingly.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6902
At least a brush under each tool needs to exist by default in the blend
file to reset its properties. If it does not exist, the user needs to
reset it manually.
Reviewed By: brecht
Maniphest Tasks: T74307
Differential Revision: https://developer.blender.org/D6972
When sculpting on multi user data normals needs to be recalculated to
avoid artifacts. I refactored that line to make it more readable.
Reviewed By: brecht
Maniphest Tasks: T73707
Differential Revision: https://developer.blender.org/D6957
When processing a timer event WM_event_tablet_data returns 0 instead of
the last valid pressure value from the tablet. This always stores the
last pressure value and uses it in case a timer event is being
processed.
Reviewed By: brecht
Maniphest Tasks: T74074
Differential Revision: https://developer.blender.org/D6950
This includes the following changes:
- Use always the angle stored in the StrokeCache when deforming
- Interpolate between the previous and the new sampled angles
- Calculate the cursor matrix only on the 0 radial symmetry iteration
Reviewed By: brecht
Maniphest Tasks: T73947
Differential Revision: https://developer.blender.org/D6901
Now only the hue is different between different strip types, with the
saturation the same for all.
Differential Revision: https://developer.blender.org/D6775
The 'random' unit tests and some examples from the new boolean code
triggered asserts and crashes. This fixes those.
There is a new flag in the input that optionally disables a pass
over input to snap segment edges to other segments.
Removing the 'Large Cursors' option as it is no longer applicable or useful on any platform.
Differential Revision: https://developer.blender.org/D6958
Reviewed by Brecht Van Lommel
This brush has a simple physics solver that helps when sculpting cloth.
- The mass and the damping properties of the simulation are properties of the brush.
- It has two additional radius control to limit the influence and falloff of the simulation.
- Masked vertices are pinned in the simulation, and it applies the sculpt gravity directly in the solver.
- The Cloth Brush has 7 deformation modes with 2 falloff types (radial and plane).
The brush can create the constraints only on the required PBVH nodes, so the simulation is isolated on high poly meshes. As long
as the brush size is not too big it should be possible to keep it real time.
Known issues:
- The way constraints are created is extremely basic and it creates repeated constraints. Maybe there is another way to create fewer constraints while keeping the simulation quality decent. This part can also be multithreaded. (As it is it works ok, but it could be better)
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6715
The coordinates should be read from the PBVH when using deform modifiers.
This is needed for the cloth brush to work with subdivisions, as it reads the
vertex coordinates using this function when building and updating the
constraints.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6967
Updates include:
- std::move() cleanup in rcmatrix.h
- Enabled parallelization for fluid guiding (fairly noticeable speed improvement).
- More flexible flags setter function with control over boundary width.
This fixes denoising being delayed until after all rendering has finished. Instead, tile-based
denoising is now part of the "RENDER" task again, so that it is all in one task and does not
cause issues with dedicated task pools where tasks are serialized.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6940
This makes the memory allocation for the denoiser state use the memory allocator in Cycles, which
will evict textures to host memory when there is not enough space on the device. This means the
allocation for the denoiser state won't just fail if there is no more space and instead more space is
made for it to work. Also simplifies code somewhat.
Fix for fix (rB8bf7ca9fe31c).
The fix was not actually working [neither for multi-armature-editing as
intended nor for single armatures] and it broke single-armature bone
separation completely.
More or less indentation issue, move relevant code out of bone loop, do
after all bones have been checked, guess this was intended anyways.
Differential Revision: https://developer.blender.org/D6949
Use BASE_VISIBLE_VIEWLAYER (rather than BASE_VISIBLE_DEPSGRAPH -- which
is not including the 'Hide in Viewport')
Note: the is one case where this still 'fails':
- 'Show only Hidden' plus excluding the parent collection (would be nice
to show those -- but contents of excluded excluded collections dont get
show in any case... this would be more work and for another report...)
Maniphest Tasks: T71611
Differential Revision: https://developer.blender.org/D6953
Useful for cases when topology does not need to have any
crease or UV layers. Now instead of assigning callbacks
which returns zero data is possible to simply assign the
callback itself to NULL.
This is in preparation of new object types. This only changes mesh_eval, we
may do the same for mesh_deform_eval and other areas in the future if there is
a need for it.
This previously caused a bug in T74283, that should be fixed now.
Differential Revision: https://developer.blender.org/D6695
This reverts commit f2b95b9eae.
Fix T74283: modifier display lost when moving object in edit mode.
The cause is not immediately obvious so better to revert and look at this
carefully.
Under the special circumstance of double clicking and then opening a
panel with sub-panels again, the parent panel's offset in y-direction
was added twice to the child panels: once inside "get_panel_real_ofsy"
and once within "align_sub_panels".
This introduced a visible glitch.
Reviewed by: Julian Eisel
Differential Revision: https://developer.blender.org/D6942
Was affecting quad faces. where 0 was always passed for inner
loops and a wrong one for boundary ones.
In the current code this change shouldn't cause any difference
since the corner index is ignored in the actual callback, but
it is required to have his fixed for an upcoming changes.
The apparent off-by-one error was caused by a few factors:
- The 'blend base' colour was green for the two frames directly
surrounding the current frame, but black for the current frame itself.
- For the frames before the current one, the 'blend base' was mixed with
black, making the green stand out clearly, but fading to black again
for the current frame. This looks like an off-by-one, even though it
was just bad mixing.
- For the frames after the current one, the 'blend base' was mixed with
cyan, which already has a strong green component, so mixing it there
was much less visible, making the entire thing look like an off-by-one
error where it actually wasn't.
I have simplified the code, so now it only chooses green as the 'blend
base' for the current frame, and simplified the mixing for the current
frame.
This further separates requested attributes and textures from the actual
node graph, that can be retained after the graph has been compiled and
freed. It makes it easier to add volume grids as a native concept, which
sits somewhere between an attribute and a texture.
It also adds explicit link types for UDIM tile mapping, rather than
relying on fairly hidden logic.
System GLEW often is not new enough, which gives error on startup. The build
correctness should not be affected by using lite vs. full, so better to leave
this out than save compiling one extra file.
Cycles recently fixed this issue, EEVEE needed to be adapted to output
similar results in the light passes.
This patch implements cycles `safe_divide_even_color` function to a GLSL
function that will be used when extracting the light passes.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6948
Shadow could penetrate occluded geometry. This patch adds a check to see
if the light is in the right location to light the pixel.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6918
This is in preparation of new object types. This only changes mesh_eval, we
may do the same for mesh_deform_eval and other areas in the future if there is
a need for it.
Differential Revision: https://developer.blender.org/D6695
The dope sheet, NLA editor, and other similar animation editors were
ignoring the temporariy visibility flag of collections. As a result, an
editor that's supposed to show animation data of visible objects only was
still showing such data of objects that were hidden by hiding their
collection.
This was observed while fixing T71743.
Cycles uses the display_space_shader it gets from Blender. When
requested the shader used to be a GPU_SHADER_2D_IMAGE_COLOR, which
set the `color` uniform incorrectly. This resulted in a yellowish tint
when used.
To fix this the `GPU_SHADER_2D_IMAGE` shader will be used.
Reviewed By: brecht, fclem
Maniphest Tasks: T74119
Differential Revision: https://developer.blender.org/D6921
Bug was introduced by the render passes. We had to tweak the bloom
shader a bit so we could reuse it. After that tweaking the original
alpha was ignored.
This patch will read and store the correct alpha channel.
Don't use -ffast-math functions for libc compatibility implementation, since
then the function ends up calling itself. This may not be ideal for performance
but should be insignificant in practice.
When disabling AO or BLOOM in the render tab, when the pass is shown in
a 3d viewport the pass wasn't reset. This resulted in showing a black
texture and a not filled UI render pass in the shading popover.
This patch will by default reset to the combined pass. It is intended
that the render_pass in the 3d shading struct isn't set to combined as
people could have disabled AO/bloom by mistake and it could reset
viewports that aren't visible.
When disabling AO or BLOOM in the render tab, when the pass is shown in
a 3d viewport the pass wasn't reset. This resulted in showing a black
texture and a not filled UI render pass in the shading popover.
This patch will by default reset to the combined pass. It is intended
that the render_pass in the 3d shading struct isn't set to combined as
people could have disabled AO/bloom by mistake and it could reset
viewports that aren't visible.
This clarifies logic in ED_object_add_generic_get_opts (see comments in
code).
Also following issues are adressed:
- do not enforce setting the op UI to WORLD when rotation is set (e.g.
on redo). This is not wrong (happens in world space), but is more
confusing than helpful (was added in own rB3b8a14a3c072).
- respect setting the User Preference to WORLD properly (could fail when
set to something else, then back to WORLD)
Maniphest Tasks: T69576
Differential Revision: https://developer.blender.org/D6943
This fixes {T70269}.
Before this commit there was complicated code to try and compute the
correct parent inverse matrix for the 'Child Of' and 'Object Solver'
constraints outside the constraint evaluation. This was done mostly
correctly, but did have some issues. The Set Inverse operator now defers
this computation to be performed during constraint evaluation by just
setting a flag. If the constraint is disabled, and thus tagging it for
update in the depsgraph is not enough to trigger immediate evaluation,
evaluation is forced by temporarily enabling it.
This fix changes the way how the inverse matrix works when some of the
channels of the constraint are disabled. Before this commit, the channel
flags were used to filter both the parent and the inverse matrix. This
meant that it was impossible to make an inverse matrix that would
actually fully neutralize the effect of the constraint. Now only the
parent matrix is filtered, while inverse is applied fully. As a result,
pressing the 'Set Inverse' matrix produces the same transformation as
disabling the constraint. This is also reflected in the changed values
in the 'Child Of' unit test.
This change is not backward compatible, but it should be OK because the
old way was effectively unusable, so it is unlikely anybody relied on
it.
The change in matrix for the Object Solver constraint is due to a
different method of computing it, which caused a slightly different
floating point error that was slightly bigger than allowed by the test,
so I updated the matrix values there as well.
This patch was original written by @angavrilov and subsequently updated
by me.
Differential Revision: https://developer.blender.org/D6091
While one issue of the report was fixed already (rB27adc4aa3347), still a
couple of clarifications could be made:
- correct/clarify which modes these apply to
- clarify that "Face Mask" is a paint thing
Maniphest Tasks: T54323
Differential Revision: https://developer.blender.org/D6797
UI
Main issue is that the UI doesn not respect the Unified Weight/Strength
setting (but instead alwasy shows the brushes weight/strength)
Was working before, but somewhat missed in rBfb74dcc5d69d.
See discussion in T74025 on why the Gradient Tool in its current state
is not very user friendly in general, what I think we should do though
is at least have the values in the UI respect Unified Weight / Unified
Strength since this is what the tool will use internally...
ref T74025
Maniphest Tasks: T74025
Differential Revision: https://developer.blender.org/D6900
Change to false color are based on artist feedback:
* Even 2.5 EV differences down to -10 EV.
* Even 2.167 EV differences up to +6.5 EV.
* Slightly below middle grey biased toward cyan.
* Slightly above middle grey biased toward yellow.
* Converts to nonlinear luma for more accuracy.
On Linux, precompiled libraries may be made with a glibc version that is
incompatible with the system libraries that Blender is built on. To solve
this we add a few -ffast-math symbols that can be missing.
Differential Revision: https://developer.blender.org/D6930
Based on work by Nathan Craddock, with further changes to apply it to all
precompiled libraries.
Differential Revision: https://developer.blender.org/D6929
If the current frame was not in range, the ImageTile's 'ok' was stuck at
0. Similar to the ImageUser being reset to 'ok' in
BKE_image_user_frame_calc, we now do for ImageTiles as well.
note: the crasher part of T74225 was fixed in rB5c490d437743.
Maniphest Tasks: T74225
Differential Revision: https://developer.blender.org/D6939
The slashed were used for comparing bookmarks to the current directory.
Add trailing slashes in 'fsmenu_insert_entry',
which avoids having to duplicate strings just to add a slash
before passing to this function.
Currently, there its a function that sets manually the fully_hidden flag
of the nodes from the visibility operators in paint_hide.c. The undo
code was not updating the flag, so the visibility state of the nodes was
incorrect after preforming undo operations. This sometimes was drawing
fully hidden nodes with empty buffers, causing artifacts in the
geometry.
I added a function to mark nodes which visibility state changed (similar
as we are updating the mask flags and the nodes bounding boxes). This
way, the tools, operators and undo code don't have to update the
visibility flags, making everything much simpler to understand and
maintain.
I did not remove the flag update code from the current visibility
operators in this patch, but after reimplementing them (and all the new
ones) in the new visibility system, all visibility updates should be
done using this method and the BKE_pbvh_node_fully_hidden_set function
should be removed.
Reviewed By: jbakker
Maniphest Tasks: T72721
Differential Revision: https://developer.blender.org/D6767
This un-indents the `childof_evaluate()` function by one level, making it
easier to add new functionality in an upcoming commit.
No functional changes.
The 'Clear Inverse' operator didn't properly update the constraint, so
it didn't do anything until the entire depsgraph was updated. It's now
properly tagged for update.
In the collections unit test file developers can now disable layer
collections and declutter the 3D Viewport while working in
`constraints.blend`, without influencing the actual unit tests themselves.
Lots of fixes and cleanups, mainly addressing:
* OpenEXR building was fully broken.
* Missing dependencies of Alembic to Boost and openEXR.
* OSL had changed its CMake parameters for custom OpenEXR install path.
* Dependencies between libs were not properly handles when switching a
lib from own build to system package.
SSS buffers are lazy initialized when needed. When shaders recompile the
SSS buffers could be incorrectly drawn. During the render passes project
we tried to fix this, but that resulted in incorrect result of the first
sample after a shader was compiled.
We revert this change knowing that we know the issue, but haven't found
a proper solution for it.
This patch adds a new user-configurable option to change at which sample viewport
denoising should kick in. Setting it to zero retains previous behavior (start immediately), while
other values will defer denoising until the particular sample has been reached. Default is now
at one, to avoid the weirdness that is AI denoising at small resolutions.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6906
This patch adds the ability to set colors and alpha of dashed line of
Time Markers. That way themes can avoid contrast issues and communicate
selection better. See screenshots in D6877.
Bumps subversion.
Reviewed By: William Reynish, Julian Eisel
Differential Revision: https://developer.blender.org/D6877
This adds the `Half Float Precision` option in the image property panel.
This option is only available on float textures and is enabled by default.
Adding a flag inside the imbuf (IB_halffloat) on load is done for EXR and PSD formats that can store half floating point (16bits/channels).
The option is then not displayed in this case and forced.
Related task T73086
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6891
This takes the idea from the aaOcean library to link the RNG seed to the
surface point, so that changing resolution only adds/remove surface detail.
Differential Revision: https://developer.blender.org/D6871
Currently this only tests the Child Of constraint. My aim is to cover
constraints with tests before they are refactored/altered.
No functional changes.
This is using the GGX probe as background. This has the drawback of
having the resolution choosed in the indirect lighting setting.
The blurring is not really high-quality.
The pros is that it has a simple implementation and is fast to evaluate.
This patch also fades the background alpha to make overlay engine draw the
default background color in the correct color space. Removing one colorspace
hack.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6895
This simplifies compositors setups and will be consistent with Eevee render
passes from D6331. There's a continuum between these passes and it's not clear
there is much advantage to having them available separately.
Differential Revision: https://developer.blender.org/D6848
Put DefaultFramebufferList and DefaultTextureList into an own header
file. These structs were private, but also used by the window manager
and gpu package. The previous implementation we needed to keep these
structs in sync.
The OpenCOLLADA package contains a mix of files with unix and dos line endings.
Now we mark the diff as a binary file so that the patch also contains a mix of
line endings that matches the package.
With the OptiX viewport denoiser active, an OptiX device was added to the device list even when
rendering and denoising on different devices (e.g. CPU or CUDA) in background rendering. This
fixes it, so the OptiX device is only added when actually needed, as in only when OptiX denoising is
actually active in background rendering.
Also cleaned up code there, making a proper poll function for the apply
modifier operator, that way button is properly disabled in UI itself in
most invalid situations.
The collection property `loop_triangles` was given the RNA type
`MeshLoopTriangle` (the type of the collection's items), instead
of `MeshLoopTriangles` (the actual expected RNA collection/array type).
The cutest, tiniest typo mistake in RNA code, leading to some complete
non-sense... It's fairly amazing that this did not cause more severe
issues actually.
The current code allocates and transfers a lot of memory to the GPU,
but only a small portion of this memory is actually used.
In addition, the code calls many costly gl operations during the
caching process.
This commit significantly reduce the amount of memory by allocating
and transferring a flat array without pads to the GPU.
It also calls as little as possible the gl operations during the cache.
This code also simulate a billinear filter `GL_LINEAR` using a 1D texture.
**Average drawing time:**
|before:|0.00003184 sec
|now:|0.00001943 sec
|fac:|1.6385156675048407
**5 worst times:**
|before:|[0.001075, 0.001433, 0.002143, 0.002915, 0.003242]
|now:|[0.00094, 0.000993, 0.001502, 0.002284, 0.002328]
Differential Revision: https://developer.blender.org/D6886
Since this control is set as you drag in the viewport anyway, this amount value is effectively useless.
It was only recently added by mistake with the Bevel GSOC.
- Move gizmo cursor check into ED_region_cursor_set so the result of
calling this function is the same as flagging for cursor update.
- Use tagging in ui_popup_block_remove which avoids adding a
mouse-move event in case the cursor needs to be changed again.
- Fixed uninitialized result used in DynamicMusic::seek().
The comment to this function says false is returned if the handle
is invalid, while in practice non-initialized value will be returned.
- Spelling typos in comment.
- Silence -Wdelete-non-abstract-non-virtual-dtor warning.
Differential Revision: https://developer.blender.org/D6896
File Browser using Windows file attributes for decorating and hiding items.
Differential Revision: https://developer.blender.org/D6816
Reviewed by Campbell Barton
This fixes a threading issue (T73593) between drivers that write to the
same memory address. Driver nodes in the depsgraph now get relations to
each other in order to ensure serialisation.
These relations are only added between drivers that target the same
struct in RNA, which is determined by removing everything after the last
period. For example, a driver with data path
`pose.bones["Arm_L"].rotation_euler[2]` will be grouped with all other
drivers on that datablock with a data path that starts with
`pose.bones["Arm_L"]` to form a 'driver group'.
To find a suitable relation within such a driver group, say the relation
(from → to), a depth-first search is performed (turned out to be
marginally faster than a breadth-first in my test case) to see whether
this will create a cycle, and to see whether there already is such a
connection (direct or transitive). This is done by recursively
inspecting the incoming connections of the 'to' node and thereby walking
from it towards the 'from' node. This is an order of magnitde faster
than inspecting the outgoing connections of the 'from' node.
This approach generalises the special case for array properties, so the
code to support that special case has been removed from
`DepsgraphRelationBuilder::build_animdata_drivers()`.
A test on the Spring rig [1] shows that this process adds approximately
8% to the build time of the dependency graph. In my test case, it takes
28 ms for this process on a total 329 ms construction time. However,
since it also made some code obsolete, it only adds 24 ms (=8%) to the
construction time. I have experimented with a simple cache to keep track
of known-connected (from, to) node pairs, but this did not significantly
improve the timing.
Note that animation data and drivers are already connected by a
relation, which means that animating a field and also changing it with a
driver will not cause conflicts.
[1] https://cloud.blender.org/p/spring/5d30a1076249366fa1939cf1
Differential Revision: https://developer.blender.org/D6905
Reviewed By: sergey, mont29
Updates include:
- A fix from Jacques that changed the loop order in the mesh creation function (the fix speeds up the function significantly due to fewer cache misses).
- Some of the grid copy helper functions are now multithreaded.
- A fix for Windows file IO. Now it possible to load files with non ASCII characters on Windows too.
bone constraints settings
'UI_context_copy_to_selected_list()' was working fine for constraints on
objects (would properly get the path and list of selected objects) and
also for bone settings (would properly get the path and list of bones).
There was no special case for constraints on bones though, so code would
work on selected objects (not selected bones).
So we now get the right bones and resolve the path to the constraint
setting _from the bones_.
Maniphest Tasks: T74006
Differential Revision: https://developer.blender.org/D6897
This partially fixes T73593.
The `add_relation(driver_key, property_entry_key, ...);` call can fail
in the following situation:
- A collection is linked, and instanced into the scene by an Empty.
- The collection contains an object with a driver on its `hide_render` or
`hide_viewport` property.
As the object doesn't exist as a real object in the scene, it's added with
`base_index=-1` to the depsgraph (see `DepsgraphNodeBuilder::build_collection()`).
As a result the node for syncing the restrictflags back to the base
isn't present in the depsgraph, and the `add_relation()` call failed.
This commit fixes the warning, simply by not attempting to add the
offending relation.
This patch adds new render passes to EEVEE. These passes include:
* Emission
* Diffuse Light
* Diffuse Color
* Glossy Light
* Glossy Color
* Environment
* Volume Scattering
* Volume Transmission
* Bloom
* Shadow
With these passes it will be possible to use EEVEE effectively for
compositing. During development we kept a close eye on how to get similar
results compared to cycles render passes there are some differences that
are related to how EEVEE works. For EEVEE we combined the passes to
`Diffuse` and `Specular`. There are no transmittance or sss passes anymore.
Cycles will be changed accordingly.
Cycles volume transmittance is added to multiple surface col passes. For
EEVEE we left the volume transmittance as a separate pass.
Known Limitations
* All materials that use alpha blending will not be rendered in the render
passes. Other transparency modes are supported.
* More GPU memory is required to store the render passes. When rendering
a HD image with all render passes enabled at max extra 570MB GPU memory is
required.
Implementation Details
An overview of render passes have been described in
https://wiki.blender.org/wiki/Source/Render/EEVEE/RenderPasses
Future Developments
* In this implementation the materials are re-rendered for Diffuse/Glossy
and Emission passes. We could use multi target rendering to improve the
render speed.
* Other passes can be added later
* Don't render material based passes when only requesting AO or Shadow.
* Add more passes to the system. These could include Cryptomatte, AOV's, Vector,
ObjectID, MaterialID, UV.
Reviewed By: Clément Foucault
Differential Revision: https://developer.blender.org/D6331
editmode
This was already supported in "Select" & "Deselect" but not in "Assign".
So similar to rB6b39dc7672eb, we now check if the material corresponding
to the currently selected material slot is found in other objects
materials and assign this (instead of always assigning their 'own'
actcol).
Maniphest Tasks: T73871
Differential Revision: https://developer.blender.org/D6869
PAINT_TEXTURE)
This fails because some tool contexts define their tools with functions
[see the following list for context that fail]:
- PARTICLE (_defs_particle.generate_from_brushes)
- SCULPT (_defs_sculpt.generate_from_brushes)
- PAINT_TEXTURE (_defs_texture_paint.generate_from_brushes)
- PAINT_VERTEX (_defs_vertex_paint.generate_from_brushes)
- PAINT_WEIGHT (_defs_weight_paint.generate_from_brushes)
- PAINT_GPENCIL (_defs_gpencil_paint.generate_from_brushes)
- SCULPT_GPENCIL (_defs_gpencil_sculpt.generate_from_brushes)
- WEIGHT_GPENCIL (_defs_gpencil_weight.generate_from_brushes)
ToolSelectPanelHelper._tools_flatten() is usually called with
cls.tools_from_context(context) [that already yields from the function].
But when registering a tool, _tools_flatten() will still give back this
function, not a ToolDef - and we cannot get a bl_idname from that.
Now check for this and yield None in that case.
Also share logic across all tool_flatten functions:
- _tools_flatten
- _tools_flatten_with_tool_index
- _tools_flatten_with_keymap
Maniphest Tasks: T63892
Differential Revision: https://developer.blender.org/D6763
The latest versions of Embree support Catmull-Rom splines
which use less memory than the previously used Hermite splines.
The representation is also much closer to Cycles own data structures
and can hopefully be unified in the future for more memory savings.
Memory savings using Victor benchmark scene:
Compared to previous Embree: ~400MB
Compared to Cycles' native BVH: ~1GB
Consequences of own refactor of usercount handling in readfile.c
(rB367ecff15d74).
Not super happy to have to call that function twice, but that should be
OK (not a real overhead here anyway).
The functionality of the mask modifier remains unchanged.
This patch updates the mask modifier so that it uses C++.
The manual memory management has been replaced with proper containers.
The large `applyModifier` function has been splitup into multiple smaller functions.
A large speedup is achieved by using simple arrays instead of hash tables in multiple places.
In my performance test file the playback speed increased from 1.1 to 5.1 fps on my laptop.
Reviewers: campbellbarton, brecht
Differential Revision: https://developer.blender.org/D6779
The problem is that Custom Shape Bones can also have a custom size.
So the pchan->disp_mat doesn't always consider the actual length of the bone.
The proposed solution is to calculate the axes matrix at the drawing pass.
Ref T65640
Reviewed By: fclem
Differential Revision: http://developer.blender.org/D5049
We detect the case where we need to invert the facing directly inside the
DRWView update and do the appropriate GL calls at draw time.
Fix T63047 Camera with negative scale works only in Cycles Rendered view
Fix T71352 Negative scale camera causes BLI_assert
This commit replaces the "Use Gradient" checkbox theme option with an
enum and implements a radial background.
Whith this change, it should be easier to implemet other types of more
complex background types, like a world space oriented gradient.
Reviewed By: billreynish, fclem, brecht
Differential Revision: https://developer.blender.org/D6825
It does not make sense to read those values when loading a file and they can crash the cursor if they contain invalid coordinates.
Reviewed By: brecht
Maniphest Tasks: T54270
Differential Revision: https://developer.blender.org/D6754
By pinching this way, we can fix some artifacts when sculpting following
the topology direction. It does not make much difference with dyntopo/
remesher, but I think this should improve the quality of the brush when
working with Multires.
Reviewed By: JulienKaspar, jbakker
Differential Revision: https://developer.blender.org/D6587
Depsgraph RNA pointer would generate infinite loop in override comparisons.
Depsgraph pointer should never be considered here anyway, this is purely
runtime data.
Some fluid cache functions were not using Blender's more secure BLI_gzopen() function. On Windows there are some special cases which this function can handle compared to the plain gzopen().
Having that extra ID users handling at readfile level, besides generic
one ensured by libquery, has been something bothering me for a long time
(had to fix my share of bugs due to mismatches between those two areas).
Further more, work on undo speedup will require even more complex ID
refcount management if we want to keep it in readfile.c area.
So idea is instead to generalize what we did for linked data already
when undoing: recompute properly usercount numbers after liblink step,
for all IDs.
Note that extra time required here is neglectable in a whole .blend file
reading (few extra milliseconds when loading a full production scene
e.g.).
Notes:
* Some deprecated data (IPOs) are not refcounted at all anymore, this
should not be an issue in pratice since the are supposed to get deleted
after doversion anyway.
* Refcounting happens after `do_versions_after_linking`, i.e those
functions won't get valid ID usercounts currently. Again, this is not a
problem in current code, if needed we could recompute refcount before,
and then ensure `do_versions_after_linoiing()` actually handles properly
usercount, which it does not currently.
Differential Revision: https://developer.blender.org/D6881
Only use the hack in `key_block_get_data` for the bmesh shapekey.
The remaining shapekeys can use the original offset values.
Differential Revision: https://developer.blender.org/D6516
This commit actually fixes several issues in this modifier, but main one
from the report was caused by adding a `CD_NORMAL` layer to loops to
store temp real clnors. Unless we plan on modifying the topology itself,
this is useless, and would require some additional 'dirty normals'
tagging to work properly, so just switched to simpler, cleaner solution
of having a local array of computed clnors.
The node would render black in this case (but should use the
'active_render' layer choosen in the object data properties -- this is
now in line to how this is handled for e.g. UVs)
This introduces ATTR_STD_VERTEX_COLOR and uses this thoughout, if no
particular layer is specified in the node.
Maniphest Tasks: T73938
Differential Revision: https://developer.blender.org/D6887
The actual naming might also be a subject to change, especially the one
around `level`. Tricky part here is that at some point in the API there
will be change from Blender modifier's Quality to OpenSubdiv's Level,
but which API level is most suitable for this?
At least now meaning of settings is better documented ans should be
clear what's going on.
This applies the change from T73348 fix to related functions.
Instead of checking against an epsilon to avoid divide by zero,
perform the division and check the result is finite.
This is needed since small faces can have an area under 'FLT_EPSILON',
and dividing by values close to zero can result in 'inf'.
This patch provides an optimisation for Ease (Smoothstep) setting in the color ramp node.
This optimisation exists already for Constant and Linear modes.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6880
This patch adds a dedicated path to extract 16bit normals instead of packing them into 10bits/comp.
The tangents are also packed to 10bits/comp if not using the new High Quality Normal option.
Fix T61024 Degraded texture shading on dense meshes
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6614
Under some circumstances, MultiDrawIndirect was disabled to improve perf.
of average scene. But this conflicted with the normal instancing buffer
filling if only 1 or 2 instances were needed to fill the buffer. All
consecutive drawcalls could not be batched together and performance would
degrade rapidly.
This patch make my instance test scene go from 11fps back to 40fps where
it should have been.
Adds the invert vgroup option to the UVWarp modifier. Adds a flag and char padding to the DNA.
Differential Revision: https://developer.blender.org/D6841
Some users reported that the previous red color of the cursor was too
saturated and that can be distracting.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6828
This matches the name in the properties panel for the Blob and Snake
Hook brushes
Reviewed By: brecht
Maniphest Tasks: T73105
Differential Revision: https://developer.blender.org/D6805
cycles_util_avxf_avx2_test failed on the cross test, since
it wasn't immediately clear why, the test was disabled.
After looking into it, this test when build for AVX2 is
generating FMA instructions where the intermediate results
have "infinite" precision [1] leading to slightly different
results.
This diff re-enables the cross test and allows for a small error
in the results.
[1] https://www.felixcloutier.com/x86/vfmadd132ps:vfmadd213ps:vfmadd231ps
Differential Revision: https://developer.blender.org/D6873
Reviewers: brecht
This does not seem to be an actual issue in current master code (as
creating a new ID will assign 1 user to it by default), but is breaking
generic usercount handling in future changes.
The render passes didn't follow the DrawManager way of doing things. It added new geometry and shading groups during drawing. This would make it harder to migrate to Vulkan later on.
This change will re-implement this part by using uniform references.
Reviewed By: Clément Foucault
Differential Revision: https://developer.blender.org/D6875
Handling those through different ways /might/ be needed sometimes, but
in most case this is just a nest of issues, since you can easily forget
to take them into account.
Note that this should be a 'non-functional' change, as this new behavior
is not used anywhere yet.
Comparing the drive letter was case sensitive,
causing 'BLI_path_rel' to fail in common cases
(manually entering a lower case drive letter for example).
Surprisingly this issue dates back to 2005 and wasn't reported.
Revert change from 5f6c45498c
Excluding enum items meant Python scripts couldn't access them.
This is no longer needed now Area.ui_type is used for the menu.
When you switch between debug/release mode a lot, you could end up
in a situation where the generated RNA code was out of sync between
the two configurations.
There was one function optionally defined with an `#ifndef NDEBUG`
guard, this patch adds a stub implementation for the other configurations
to prevent build errors.
Differential Revision: https://developer.blender.org/D6855
Reviewers: brecht
Seems like sometimes files are being only partially ready, which makes it so there
are unsigned files, failing to deliver fully signed bundle.
Now expected archive file size is stored into stamp file and is checked against
size of the archive file on another side.
There are some bare prints used for debugging, would need to switch it to a proper
logger (or to be removed).
This extends FindOpenShadingLanguage.cmake to also look for the location of
stdosl.h and adds the path to the invocation of oslc to deal with the headers
being in different locations a little better.
Differential Revision: https://developer.blender.org/D6865
Reviewers: brecht
The redo panel does not consider the position of the mouse.
So it is not possible to know the direction to redo the operator.
The solution is to add a new `direction` parameter that can be
saved and used for redo.
Differential Revision: https://developer.blender.org/D6852
Each image that goes through preprocessing is already duplicated in `input_preprocess()`
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6790
Use gpu batch drawing for cache content in VSE.
Immediate drawing caused significant dorp in framerate.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6835
This node provides the ability to rotate a vector around a `center` point using either `Axis Angle` , `Single Axis` or `Euler` methods.
Reviewed By: #cycles, brecht
Differential Revision: https://developer.blender.org/D3789
Rendering with multiple CUDA devices but denoising with OptiX caused parts of the image to go
missing at the start while the resolution was scaled. This is because the copy operation in
`MultiDevice::map_neighbor_tiles` which slices the copy across all devices would slice based on the
full resolution and not the scaled one and therefore copy incorrect data between devices.
Since this is not the recommended way of using viewport denoising anyway, simply avoid those
incorrect copies for now by disabling denoising while the resolution is scaled. Doing both rendering
and denoising with OptiX is not affected by this, since it avoids those copies altogether anyway.
Once again those crappy weirdos IDs with their crappy weirdos 'loopback'
pointers...
This is a quick hack for now, think id_copy needs to be reworked a bit
to supported re-entrant sub-ID copying (also an issue with nodes I bet).
- Unlock property range.
- Use triangular noise to keep perceptual noise error more uniform.
Remap range to preserve perceptual intensity.
- Center noise distribution around 0 for GPU implementation because of
rounding.
- Do dithering after merging overlays.
Effect of using triangular noise is not really noticeable if you don't use
really low bitdepth. But doing a test in the shader were we artificially
reduce the bitdepth (`col = (col * 16) / 16;`) reveals the real difference.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6850
`invoke_props_dialog` and `invoke_popup` had a width and a height field. The height field was ignored as the height is determined based on the content. This change removes the field from the BPY + WM_api
Reviewed By: Campbell Barton, Jacques Lucke
Differential Revision: https://developer.blender.org/D6694
This change will not render the normals for faces that are hidden.
Before we had instance drawing the hidden faces were registered in the
index buffer. During the overlay refactoring the rendering was migrated
to instance rendering. Instance rendering does not use the index buffer
so the data was ignored.
This patch stored the normal visibility in the .w part of the normal or
for face normals it will set the normal to zero. The shader looks at
this and renders the normals fully transparent when detected.
Reviewed By: Clément Foucault
Differential Revision: https://developer.blender.org/D6798
This modifies the common CUDA implementation for adaptive kernel compilation slightly to support both CUBIN and PTX output (the latter which is then used in the OptiX device). It also fixes adaptive kernel compilation on Windows.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6851
* Direction mode X, Y and Z to align with axes rather than diagonal or
spherical as previously. X is the new default, existing files will
use diagonal or spherical for compatibility.
* Phase offset to offset the wave along its direction, for purposes like
animation and distortion.
https://developer.blender.org/D6382
This patch exposes the Window Manager `is_interface_locked` as a read
only property to python.
This property is needed to allow script writers to detect if a job is
running with locked interface like the Alembic exporter.
Reviewed By: Brecht
Differential Revision: https://developer.blender.org/D6749
There are still some issues left after the changes in master:
* Some optimizations need to be redone differently
* Color-space handling needs to be redone differently
* Upside-down drawing is broken
This separation between reader and writer code is part of Milesone 2 of
T73363.
In this commit the reader and writer classes are separated into their
own files, any `#include` and `using` statements are cleaned up, and
some separator comments have been removed.
No functional changes.
There are two issues solved in this commit:
- Our Windows buildbot has slightly different floating point errors than
the Linux one, which meant a larger delta was required for float
comparisons.
- The test performs an export to a temporary Alembic file and
subsequently imports it. Deleting the temporary file was impossible on
Windows because it was still in use. This is now resolved by first
loading the default blend file before deleting the Alembic file.
Since this tool doesn't perform selections when you click, it's important that the cursor helps communicate that something else will happen.
Also fix missing 'to' in Curve Edit Mode.
This avoids the ambiguity with the Cut operator in the Sequencer, which could be confused with Cut/Copy/Paste.
Use 'Split' for the operator and 'Blade' for the active tool.
Patch by Nathan Lovato, with edits
Differential Revision: https://developer.blender.org/D5542
There seems to have been no particular great reason why we were activating tools using a tweak event with MMB.
This change makes it so tools are activated with a press event instead.
The user theme is loaded after window creation so we should create the
window with a color that at least matches the default theme.
Differential Revision: https://developer.blender.org/D6858
Two main reasons for the lag:
- Allocation of memory with transfer to GPU.
- BLF_cache_clear();
The (partial) solution is to avoid memory allocating in some setups
through the `GPU_texture_clear`.
Differential Revision: https://developer.blender.org/D6837
This is needed so pose keymap doesn't override weight paint,
which currently prevents the weight paint context menu from showing.
Logically pose mode is secondary when used in combination with
weight paint, so evaluate it afterwards.
Also move vertex paint keymap since they have a lot in common,
it's simpler to keep them evaluated at the same place.
Resolves T73384
When a vertex between two edges is being collapsed,
it's important that edges between delimiting faces use the
angle between edges without scaling it down.
While faces with different materials wont ever be merged into a single
face, all the detail between the two faces may be removed.
Currently the only option is to warp based on the transform of other
objects, which is inconvenient if you want to e.g. control it through
a driver - you need to set up a dummy object and go through that,
which is clunky and should be unneccessary.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6690
This adds some extra functions recently added to the float Maths Node.
Not all functions have been ported over in this patch.
Also:
+ Tidy up menu
+ Change node color to match other vector nodes, this helps distinguish vector and float nodes in the tree
+ Move shared OSL functions to new header node_math.h
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6713
Steps to reproduce were:
* Add new (empty) scene
* Add some mesh object
* Change to texture paint mode
While it can be argued that we should already init the settings when
creating a new scene (so that the texture tab can show up),
BKE_paint_init() should always ensure paint settings are initialized
either way.
Related to T73611.
Sometimes the viewport buffer size is zero for a frame, which caused the denoising task to also try to
launch CUDA kernels with a launch size of zero, which in turn failed with a CUDA error. This patch
prevents launches from occuring in this case, similar to how it is handled in `copy_to_display_buffer`.
NOTE: This change shouldn't have any visible effect. It's just the
first (easiest) step towards decoupling gizmo redraws from viewport
redraws.
We currently redraw the entire region whenever a gizmo needs redrawing,
which would be nice to avoid in the future, see T73198. The first step
towards this would be having a separate tag for them, which is what
this patch implements.
The term "editor-overlays" was chosen because for the forseeable future,
we'll also have to redraw non-gizmo overlays in-between drawing 3D and
2D gizmos. Namely annotations.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6838
Fix ReDo and create a new operator to display only the required properties.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6849
The `convert_matrix()` function just converts between the Alembic and
Blender matrix data types, and doesn't do any coordinate system conversion.
To clarify this, the function has now been renamed to
`convert_matrix_datatype()`.
I also moved the implementations next to each other in the source file,
so that it's visible that there are actually two of them.
No functional changes.
The Alembic importer now works with local coordinates. Previously, the
importer converted transformations from Alembic to world coordinates
before processing them further; this processing often included
re-converting to local coordinates. This change made it possible to
remove some code that assumed that a child transform was only read after
its parent transform.
Blender's Alembic code follows the Maya convention, where in the zero
orientation the camera looks forward instead of down. This extra
rotation is now handled more consistently, and now also properly handles
children of cameras. This fixes T73269.
Unit tests were added to at least ensure that the importer and exporter
are compatible with each other, and that static and animated camera
transforms are handled in the same way.
This rename is to prepare for a future addition to the unit test file.
Currently it's named "import" and I will add an export test as well. The
rename is a separate commit to easily see the difference between the
rename and the addition of another test.
No functional changes.
Two main reasons for the lag:
- Allocation of memory with transfer to GPU.
- BLF_cache_clear();
The memory allocation seems to be unnecessary, so I removed it.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6837
Various fixes for smoke / fire flow objects:
- Apply inflow at every subframe (new: also emit during adaptive steps in between frames)
- Fix issue with fire not being emitted on first frame
- Higher value range for smoke flow density variable
MSVC does not have -march=native, so the kernel gets built without AVX2 and
BVH8 support. The code assumed it to be available and crashed
Differential Revision: https://developer.blender.org/D6082
This fixes the issue where sun shadowmaps needs a very big bias value to
make any difference.
The bias is now in world space and not dependant on shadow bounds.
Unfortunatelly this breaks compatibility with previous version and old
scene are likely to need user intervention to fix.
Also fixes the property range.
Fix T71661 EEVEE shadow from sun on incorrect face
Library data-blocks are supposed to be tagged as 'real user', i.e.
have one virtual user, since nothing else actually uses (refcounts)
them. That was done in liblink code, but not for newly added ID-LI IDs
when linking from a new libblend file e.g.
The current "Select Linked" operator works based on mouse position and
makes no sense to call from the menus and was removed in rBdd9dfadaac9b.
This patch adds an operator independent from mouse position that just
selects all keys to a corresponding point (and adds back menu entries,
adds keymap entry).
The original operator is renamed to 'select_linked_pick' internally
(this is now more in line to how "Select Linked" works for meshes,
curves etc)
Differential Revision: https://developer.blender.org/D6823
This panel is not for 2D paint, IMAGEPAINT_PROJECT_LAYER_CLONE is only
ever used in projection painting, not 2D painting.
Add a proper poll for this.
Maniphest Tasks: T73774
Differential Revision: https://developer.blender.org/D6836
Regression in aa919f3e82
Remove character margins, it complicated drawing & picking
to have one margin in pixels and a second margin in characters.
Replace this with an outer pixel-margin for drawing background colors.
MSVC has a conformance mode (/permissive-) where the C++ standard is more strictly
enforced. This mode is available on MSVC 15.5+ [1]
This patch enables this mode on compilers that support it and cleans up the few violations it threw up in the process.
- Mantaflow was using M_PI without requesting them using the _USE_MATH_DEFINES define to opt in to non default behaviour.
- Collada did not include the right header for std::cerr, this seemingly was fixed for other platforms already but put inside a platform guard.
- Ghost had some scoping issues regarding uninitialized variables and goto behaviour
Second landing of this patch, earlier commit was reverted due to some compiler configurations having slipped though testing
[1] https://docs.microsoft.com/en-us/cpp/build/reference/permissive-standards-conformance
Differential Revision: https://developer.blender.org/D6824
Reviewed By: brecht
This allows for more flexibility in Compositing compared to the
hardcoded alpha-over that is currently used.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6829
The introduction of python instancing for ID data in 2.80 completely
broke the 'smart & efficient' hack of using a same py object as key for
initial quick check, since rebuilding the RNAPointer in the
BPy_StructRNA would actually affect the py instance of the first ID used
to generate that py object...
TL;DR: No need for this complex and unclear optimization anymore, since
we do not actually rebuild a whole py object anymore every time we call
`pyrna_id_CreatePyObject()` from a same ID pointer.
Using a struct here allows to change given parameters to the callbacks
without having to edit all callbacks functions, which is always noisy
and time consuming.
15% to 20% improvement in cloth simulation performance with
self-collision.
The idea is to reduce the number of collisions computed by avoiding
overlapping tris with the same combination (eg. (1,0) and (0,1)).
Reviewed By: zeddb
Differential Revision: https://developer.blender.org/D6474
This partially reverts commit 5314161491.
Conflicts with Emulate "Middle Mouse Button".
This feature should be accessible under all supported configurations.
MSVC has a conformance mode (/permissive-) where the C++ standard is more strictly
enforced. This mode is available on MSVC 15.5+ [1]
This patch enables this mode on compilers that support it and cleans up the few violations it threw up in the process.
- Mantaflow was using M_PI without requesting them using the _USE_MATH_DEFINES define to opt in to non default behaviour.
- Collada did not include the right header for std::cerr, this seemingly was fixed for other platforms already but put inside a platform guard.
- Ghost had some scoping issues regarding uninitialized variables and goto behaviour
[1] https://docs.microsoft.com/en-us/cpp/build/reference/permissive-standards-conformance
Differential Revision: https://developer.blender.org/D6824
Reviewed By: brecht
The operator in its current state is based on mouse position and doesnt
make sense to be called from a menu.
(In fact it should be called 'select_linked_pick' internally and a
separate 'select_linked' should be implemented similar to how "Select
Linked" works for meshes, curves etc -- see D6823 for this)
Differential Revision: https://developer.blender.org/D6822
color
Selection and grooming only ever happens on hair keys (vertices), so
makes sense to use that theme color.
Sidenote: `CacheEditrPathsIterData` and `PTCacheEdit` have `sel_col` and
`nosel_col`, these can go, will do in a separate commit.
Maniphest Tasks: T68788
Differential Revision: https://developer.blender.org/D6725
Compared to 2.79, this also has the benefit of actually showing the curve
again if you set both bevel start and end to 0.0 for example.
Maniphest Tasks: T59271
Differential Revision: https://developer.blender.org/D6168
The option can still be modified in the bone settings in the properties
editor. In the future a correct menu entry (using an enum) can be
added back. For that, an operator like `wm.context_collection_boolean_set`
could be added for enums.
This reduces code duplication between the CUDA and OptiX device implementations: The CUDA device
class is now split into declaration and definition (similar to the OpenCL device) and the OptiX device
class implements that and only overrides the functions it actually has to change, while using the CUDA
implementation for everything else.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6814
Adds the invert vgroup mask option to the Vertex Weight modifiers.
These 3 modifiers share the same functions so they needed to be modified at the same time. They are all setup the same with the invert vgroup option being added. I had to add a flag to the Mix modifier but the others I use the existing flags.
Differential Revision: https://developer.blender.org/D6819
Provide different options for locking and unlocking vertex groups
using bone selection, accessible via a pie menu triggered via the
'K' hotkey. To implement a variety of operations, extend the old
operator with a new option to mask it by bone selection. If the
X Mirror option is enabled, selection is automatically mirrored.
This follows D6533 as the next step in improving accessibility of
vertex group locking during weight painting.
Differential Revision: https://developer.blender.org/D6618
There is a bug where the UVWarp modifier will stop working if an armature bone is set as the Bone To target and then later changed to a non armature object without clearing the Bone To string property. The modifier will still think it is needing a bone matrix and not properly use the object matrix.
A simple fix by adding a check for the object being an armature as well as the bonename string being set.
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D6809
There were to copies of stdosl.h one from stock OSL
and one in the cycles tree augmented with cycles
specific closures.
moved the cycles ones to stdcycles.h and copied
the stock stdosl.h and accompanying headers from
the OSL shader folder.
for further details see D6812.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6812
This enables an extra layer of control in the sculpt brushes.
For now it is enabled only in Scrape, but it should work in all brushes (like normal radius). In the future it may also be enabled in other brushes.
You can tweak in this property in the scrape brush to achieve a much better behavior when working on curve surfaces and control how much volume you want to trim. In most cases, it also fixes the bug where the brush keeps trimming in the same area without disabling accumulate.
It should be possible to fix some other artifacts in other brushes by tweaking this default property.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D5993
This patch allow to change the brush tip shape between a square and a
circle using a brush property.
After this change we are no longer testing the distance against a cube
(the Z axis is not used). I did not test this in depth, but if it does
not produce any artifacts I think we can keep it this way instead of
adding more complexity to the code.
In this new distance test the brush falloff is only applied on the
rounded parts of the square to avoid sharp artifacts in the
diagonals. Because of this, the round version is much softer than
the square one. The planned hardness property will fix this, but
this can also be avoided by setting the fallof to a custom curve.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6165
The IK chain was using an anchor point by default as it makes sense for
posing, but for creating curved shapes it is useful to be able to
disable it.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6584
This brush simulates deforming clay with your fingers, accumulating
material during the stroke. It has a plane that tilts during the stroke
in the front part of the brush to achieve this effect.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6238
The OptiX denoiser can be a great help when rendering in the viewport, since it is really fast
and needs few samples to produce convincing results. This patch therefore adds support for
using any Cycles denoiser in the viewport also (but only the OptiX one is selectable because
the NLM one is too slow to be usable currently). It also adds support for denoising on a
different device than rendering (so one can e.g. render with the CPU but denoise with OptiX).
Reviewed By: #cycles, brecht
Differential Revision: https://developer.blender.org/D6554
The code was using brush->alpha directly, so the alpha value from
UnifiedPaintSettings was not used when eneabled.
Reviewed By: brecht
Maniphest Tasks: T73656
Differential Revision: https://developer.blender.org/D6804
* Missing close dialog displayed to warn about unsaved changes.
* No reaction when the file was opened on a different desktop.
Differential Revision: https://developer.blender.org/D6765
ss->multires is set in sculpt_update_object, which is not called just
after disabling dyntopo, so it needs to be checked before running
reshapeFromCCG
Reviewed By: campbellbarton, brecht
Maniphest Tasks: T73706
Differential Revision: https://developer.blender.org/D6801
The channel color squares are no longer positioned under the visibility
icon, since contrast is poor. Instead they are thinner and drawn on the
left side, before the visibility icon. If the curve is hidden, the color
gets less opaque.
Differential Revision: https://developer.blender.org/D5882
This is an oversight of a fix for T69834, where I didn't realize the
ED_object_editmode_exit_ex() function is called for all objects as an
opposite of only the ones which were in edit mode.
Seems a simple fix: just move tag into a check that object was in the
edit mode prior to tag.
Differential Revision: https://developer.blender.org/D6808
Blender can crash while rendering, when scene data is changed from within
a `frame_change_pre` or `frame_change_post` callback function. This results
in bug reports like T60094, T67627, and T73530. Until this is properly
resolved, this limitation should be documented.
No functional changes.
While the file in this report had corrupted values,
this is avoidable without adding any extra overhead.
Use unsigned vertex group indices since we don't need negative values,
this is an alternative to checking they aren't negative in many places.
Vertex group values over INT_MAX is still considered invalid,
so any accidental unsigned wrapping won't be silently ignored.
This reverts commit 394b48029c.
Leaving this here does not affect performance in any significant way and it may be useful in the future.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6688
In some situations the symmetry point may be behind the camera, so the
projection is inverted and it looks wrong. This avoids drawing points in
screen space when they are behind the camera.
Reviewed By: jbakker
Maniphest Tasks: T72690
Differential Revision: https://developer.blender.org/D6487
This forces the full geometry when that overlay is enabled.
I would rather not support this and leave that bug as a know issues/
limitation because this is terrible for performance and people may not
be aware that by enabling the option they are disabling the
optimization.
Reviewed By: jbakker
Maniphest Tasks: T66349
Differential Revision: https://developer.blender.org/D6552
Even though we do not have much of those, this might change in the
future, and in any case having specific functions for this ID type in
generic `BKE_lib` area was really confusing.
The issue here is simple, IMA_SRC_TILED was missing from a check for whether the image should
be saved.
Note that tiled images won't be automatically saved if they have never been saved before.
For single images this is handled by packing them, but packing of tiled images isn't
supported yet.
However, in that case the file closing dialog will at least show a warning now instead of
completely ignoring tiled images.
Having functions defined in `BKE_lib_id.h` implemented into
`lib_remap.c` was confusing at best.
Besides trivial code splitting and header includes cleanup, had to add a
new `lib_intern.h` header for callbacks used by both remapping and
deletion code.
This commit adds the download, extract, patch, build, and install of the
Universal Scene Description (USD) library to the `install_deps.sh`
script.
Reviewed By: mont29, LazyDodo
Differential Revision: https://developer.blender.org/D6478
The changes come from the `functions` branch, where I'm using
these structures a lot.
This also includes a new `BLI::Optional<T>` type, which is similar
to `std::Optional<T>` which can be used when Blender starts using
C++17.
Make the Graph Editor Cursor panel fit in and use the split layout.
Also removed the snap buttons, since they were only a subset of the snapping possibilities, all of which are in Key > Snap anyway, and also in the context menu.
Update the theme to use correct hierarchy progression like in other editors.
Differential Revision: https://developer.blender.org/D6785
Reviewed by Brecht van Lommel
Don't use BKE_view_layer_context_active_PLACEHOLDER which is marked
as "never use this".
In fact, it isn't needed to lookup for camera in the tracking function
at all: camera object is always explicitly passed to it.
The user has preference to select the center point for alignment.
But in this case, it was not available for selection.
Also prioritizes selection of the middle point over the handles.
Reviewed By: #user_interface, brecht
Differential Revision: https://developer.blender.org/D6780
Was happening on macOS. The reason of this is because OpenImageIO
depends on boost, so it is to be passed to the linker after the
OpenImageIO libraries.
The layout was somewhat poor - hard to see what relates to the custom object, and also weirdly we show the Wireframe toggle above the Custom Object control, even though it can only be active if a bone has a custom object set.
Instead, I grouped everything in a Custom Shape sub-panel and used greying out.
Differential Revision: https://developer.blender.org/D6789
Reviewed by Brecht van Lommel
Patch adds an "Handle" option to the `SEQUENCER_OT_box_select` operator,
that allows to select the handles instead of whole strips.
Feature is mapped to Alt key modifier
A difference from the proposed design in T70730 is that covering the entire strip with the box actually selects both handles.
Reviewed By: iss
Differential Revision: https://developer.blender.org/D6372
Fluid guiding functionality was broken in the bake / read cache loop in fluid.c. Committing this to the release branch as otherwise fluid guiding would not have worked as expected (i.e. not at all).
Previously any of the named views could not have any roll,
this commit supports roll as long as it's axis-aligned (90,180,270 deg).
This is useful for snapping to views,
an improvement on cebd025e02.
Previously the MSVC environment was setup every time using vcvarsall.bat
when you ran you ran rebuild.cmd, Generally not an issue but after many
rebuilds on the same console, it grows the path environment variable
beyond what is supported and building breaks.
This patch adds a check to see if the environment is setup already
and skips the call to vcvarsall.bat
Also cleans up the double build in the msbuild's version of rebuild.cmd
install.vcxproj will build all that is needed, so no need to do a regular
build first.
oidn puts dllexport on all its functions causing the
blender binary to export these symbols.
this patch fixes this unwanted behaviour.
Differential Revision: https://developer.blender.org/D6647
Reviewers: brecht , sergey
This makes the code more consitent with the rest of blender
The `fractions_threshold` was changed to be active than enabled which is
also more consistent. This needs to be changed in other areas also.
libxml puts dllexport on all its functions causing the
blender binary to export these symbols.
this patch fixes this unwanted behaviour.
Differential Revision: https://developer.blender.org/D6646
Reviewers: brecht , sergey
Freeetype 2.9.1 tags dllexport on most of its functions so these
are now exported from the blender binary. (Same issue as D6563
which fixed it for USD)
Issue has already been fixed upstream so a simple version bump
fixes it.
This patch bumps freetype to 2.10.1
Differential Revision: https://developer.blender.org/D6645
Reviewers: brecht , sergey
OSL 1.10.9 fixes osl-bug 866 [1] which is long standing issue
on windows where paths get un-escaped and osl breaks when you
install it to for instance c:\blender-tests\new-boolean
This patch bumps osl to 1.10.9, and since osl is llvm's
only consumer, llvm/clang were bumped 9.0.1
Removed some of the patches that were no longer needed
Builds and passes all tests on windows and linux
[1] https://github.com/imageworks/OpenShadingLanguage/issues/866
Differential Revision: https://developer.blender.org/D6744
Reviewers: brecht
This works on selected_editable_objects, so no active object mandatory.
After deletion of the active object and using slection tools such as
e.g. box select, the lack of an active object would prevent this
operator from being called without good reason.
Also cannot think of any other existing poll that would really make
sense (Editor type, ...).
ref T68975
Maniphest Tasks: T68975
Differential Revision: https://developer.blender.org/D6773
This turns error condition checks into precondition checks, grouping the
non-error functionality together towards the bottom of the function and
error-handling functionality towards the top.
No functional changes.
This is not 100% a bug but a design change. The old method used the object origin as pivot point for Scale a nd Rotation, so when you moved the stroke in edit mode, the whole array ittems where offset because the pivot point distance changed.
Now, before applying scale and rotation, the stroke is moved to object origin to keep the offset when scale or rotate, so these transformations are done in stroke local space.
Glew.obj is one of the more expensive targets we have to build clocking in
at 34.907 seconds.
The root cause [1] is the msvc inliner is not super thrilled with the
glewIsSupported supported function, this patch changes the buildflags
of extern_glew to optimize for size rather than speed.
Given glew is only in play at initalization time there will be no
performance impact.
Time to build glew.obj
Before: 34.907s
After : 1.961s
[1] https://developercommunity.visualstudio.com/content/problem/732941/slow-compilation-of-glewc-for-visual-studio-2019-x.html
Differential Revision: https://developer.blender.org/D6656
Reviewers: brecht, sergey, fclem
Adding USD to a lite build fails to build due to boost errors, when you turn
boost on and rebuild still boost errors, boost was silently turned off since
it was not deemed needed. Once boost was forced on, it still fails due to TBB
being off.
This patch fixes:
- The Silent disabling of boost
- Add a check that USD is is not on before doing that
- move the TBB checks to a central location rather than the individual platform files
- Add USD to the TBB checks.
Differential Revision: https://developer.blender.org/D6479
Reviewers: brecht, sybren
Code was originally added to test a different approach to calculate the
irradiance buffer. The approach was just to slow so we never used it.
This change will remove it from the code base
Constraint stack similarly to modifier stack is fully operate on what
have been given to it, without requesting original or evaluated IDs.
Validness of datablocks passed to constraint stack are to be handled
on dependency graph/evaluation stream levels.
Mainly affects for() loops.
The reason why loop parameter was declared outside of the loop roots
back to the times when not all compilers supported C99.
This file was originally a placeholder for all the old functions that
have not yet been ported to the new draw system. Over time all the
functions that needed refactor were gone, and the functions here are
still needed.
While moving the functions around I removed dead code and made sure the
existent comments start with a capital letter and end with a full stop.
bf_intern_mantaflow lacked the `-DOPENVDB_STATICLIB` define
causing it to dynamically import openvdb, linked against
our static libs a happy time was not had by the linker.
- Move handling of private ID data (nodetree and master_collection)
under generic ID code. This shortens code a bit, but mostly avoids having
to modify all type-specific callback functions if/when we have to add
generic processing to IDs there.
- Seriously factorize `expand_xxx` area, in the same way we were already
doing in `direct_link_xxx` and `lib_link_xxx` areas.
Note that this actually also fixes some bugs (at least, potential ones),
like e.g. missing call to expand_id() for our beloved 'private ID'
(nodetrees & co), in current master code...
Differential Revision: https://developer.blender.org/D6764
Cache files are currently loaded via the Manta Python API. With very big caches this can slow down the viewport playback. Especially smoke simulations, which just load grids and no meshes, can suffer from this. This fix solves this problem by directly loading the cache files from disk (no Python). This fix has been in the works for some time. The developer of this patch is ready to handle any potential fall-out of this patch quickly.
Keyframes and channels were not updating immediately (you had to enter
channel and main regions with the mouse to force a redraw).
Differential Revision: https://developer.blender.org/D6762
Adds the invert vertex group option to the smooth modifier.
Setup same way as previous modifiers.
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D6745
Adds the invert vertex group option to the Curve modifier.
Adds a short flag and char pad to the Curve modifier DNA. Passes the flag into the curve_deform_verts function as the weight values are found there and not in the modifiers .c file.
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D6746
If the linked or append grease pencil object was added to the scene, the original file could have the datablock status in a different mode, so it's necessary to be sure is in object mode.
Differential Revision: https://developer.blender.org/D6759
Adds the invert vertex group option to the Lattice modifier.
Adds a short flag and modifies the existing char padding for the correct amount.
Adds a .invert_vgroup to the LatticeDeformUserdata.
Passes the flag into the lattice_deform_verts function where the weights around found and used.
For the other calls of lattice_deform_verts function they pass in NULL for the flag in the same way they pass NULL for the vgroup name.
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D6747
Not sure when this happened but apparently the lower bar is now windows 7 [1]
This patch bumps to API version to 0x0601 (Win7) and cleans up any uses that
worked around the globally set API version.
[1] https://www.blender.org/download/requirements/
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6758
To determine the albedo pass, Cycles currently follows the path until a predominantly
diffuse-ish material is hit and then takes the albedo there.
This works fine for normal mirrors, but as it completely ignores the color of the bounces
before that diffuse-ish material, it also means that any textures that are applied to the
specular-ish BSDFs won't affect the albedo pass at all.
Therefore, this patch changes that behaviour so that Cycles also keeps track of the
throughput of all specular-ish closures along the path so far and includes that in
the albedo pass.
This fixes part of the issue described in T73043. However, since it has an effect on the
albedo pass in most scenes, it could cause cause regressions, which is why I'm uploading
it as a patch instead of just committing as a fix.
Differential Revision: https://developer.blender.org/D6640
Apparently the compiled shader bump into some register limit and
the compiler instead of giving an error, does something incorrectly.
Differential Revision: https://developer.blender.org/D6759
When an edge loop is fully occluded the direction of movement
is not calculated resulting in unpredictable behavior.
So always calculate the direction of the groups of edge loops
but continue preventing occluded edges from affecting `mval_dir`
(global) value.
Differential Revision: https://developer.blender.org/D5247
The current edge slide is executed in this sequence:
- traverses all edges to find the ones selected,
- traverses each vertex of the selected edges,
- traverses each linked edge of the vertex to finally execute the code.
However the list of vertices that are part of selected edges are already
stored in `EdgeSlideData` through the `TransDataEdgeSlideVert *sv;` member.
Therefore, the code can be simplified as follows:
- traverses all `sv` in `EdgeSlideData`,
- get the `sv->v` vertex to finally execute the code.
Differential Revision: https://developer.blender.org/D5277
Liblink specific ID type function was so far running a loop over all IDs
of relevant type, unlike almost any other 'ID-callback-like' functions
in Blender, which usually let the looping controll to calling code.
The latter approach is more convinient when one want to add generic
(i.e. type-agnostic) code, since it typically only has to change code in
one place (caller function) instead of tens of places (all the callback
functions).
This commit also changes/sanitizes a few things that had nothing to do
in main liblink code, like mesh conversion from tessfaces to polys
(which can be done in after-linking versionning code), or scenes' cycles
detection/check regarding background 'set' scenes.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6727
In current `readfile.c` code we have that:
https://developer.blender.org/diffusion/B/browse/master/source/blender/blenloader/intern/readfile.c$3523
This is unconditionally clearing nodetree's recalc flags, and was added by rB81a762e79f83 ages ago. Thing is, in main ID read code we only clear that flag when **not** in undo context.
This proposed change intends to properly handle those cases, by moving `id.recalc` flags clearing from `read_libblock()` to `direct_link_id()`, which is also called for all 'local' IDs (ntrees and master collections currently).
I’d expect that change to be straightforward (and maybe even fixing some odd undocumented bugs), however there is no .blend file testcases associated with changes in rB81a762e79f83, so wouldn’t mind that to be double checked before it goes to master.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6711
Reduce the number of possible locations used to find libraries,
to simplify troubleshooting.
Only keep '*_ROOT_DIR' and the path used by 'install_deps.sh'.
The baking progress bar now uses the entire notification space in the UI. Before, old reports could still be visible when a bake job got started. This had the disadvantage that those message got frozen too with the bake UI freeze.
BKE_nurb_type_convert now takes r_err_msg and is more specific in the
error message...
ref T71672.
Maniphest Tasks: T71672
Differential Revision: https://developer.blender.org/D6275
Addons option
This removes the 'use_owner' option feature from rB61c8ed40f5df.
(this wasnt working well when addons are enabled and when switching
workspaces)
Now Addon filtering is just bypassed for Import/Export menus.
(by introducing/setting bl_owner_use_filter = False)
Maniphest Tasks: T58842
Differential Revision: https://developer.blender.org/D6740
WM_modalkeymap_operator_items_to_string() wasnt checking
WM_keymap_active(), so it was possible that e.g. when using the Industry
Compatible keymap, the shortcut from the Blender keymap was shown.
This also fixes the (now exposed) bug that the Industry Compatible keymap
would not have a ADD_CUT_CLOSED kmi defined for the Knife Tool [mandatory
for the status bar].
Maniphest Tasks: T67084
Differential Revision: https://developer.blender.org/D6748
The code removed was left there from an early prototype and it was not
doing anything.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6741
The conservative depth shader is ~4.5x slower than the normal one as it
uses geometry shader and fragment shader discard.
This patch also includes a hack to also fix the view parallel planar
geometry and the really small wire objects.
For some reason, the conservative raster fix does not work with normal
selection but does with box select.
This is a fix for T63356.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D6714
Lacunarity parameter determines scaling of subsequent octaves in fractal
noises. For example, Noise node have this scaling hardcoded to 2.0. Each
octave have twice bigger scale than previous one, resulting in finer details.
By design fractal noises that generate octaves with same seed should not
have Lacunarity set to 1.0, since then it just stacks up identical noises.
Differential Revision: https://developer.blender.org/D6742
This is a small optimization that makes sure the adaptive domain only considers active inflow objects. Ones with disabled fluid flow are skipped and thus the adaptive domain will not try to cover them.
An educated guess to put ensure order of static libraries initialization.
A bit weird, since OpenImageDenoise should be depending on TBB, but that
is likely being ensured by bf_compositor.
Linking succeeded on my Intel machine, and blenloader_test was passing
when doing manual test on buildbot.
The erase mode has an option to change the erase mode from soft/hard/stroke.
However, there are 4 brush types: hard/soft/point/stroke. Hard and Soft here are similar (they are both "soft" erase mode) only having different strengths.
The erase mode should be soft (Maybe this should be renamed to something more generic like "fade"), point, stroke.
Reviewed By: #user_interface, #grease_pencil, billreynish, antoniov
Differential Revision: https://developer.blender.org/D6733
From looking into builder's logs it seems that stamp file is picked
up prior to actual archive: sometimes worker reports missing archive
file, from a code path which is only possible if there is a stamp file.
Could be something with IO scheduling where bigger file is sent to
Samba server after smaller file.
Hopefully with this change this will not happen anymore.
When using duplicate linked meshes, objects that are not in edit-mode will be drawn as
it is in edit mode, when another object with the same mesh is in edit mode.
This will not be the case when one of the objects are influenced by modifiers. The change
reflects more how it was done in Blender 2.79.
The current change introduces a draw manager method that checks in detail who is responsible
for the drawing (render engine or overlay engine). If the edit mesh is not the original or
the object that is drawn doesn't draw the original mesh the object will be drawn by the render
engine.
Known Limitation of this patch is that the rendering outside edit mode doesn't reflect the
latest changes until the user switches between object and edit mode. When there are no
modifiers in use, the updating is done immediately.
IMO this would be sufficient for blender 2.82, it also fixes parts of T72733.
The updating of the surface batches requires more development and is
post-poned for now.
Reviewed By: fclem, brecht
Differential Revision: https://developer.blender.org/D6737
Added sanity check in the flow / obstacle object loops that check if the modifier data is valid. Ideally this should not be needed. However, in remove_particle_systems_from_object() the fluid modifier can get freed. It is not yet clear whether the modifier free call is really needed or not.
Font duplicator was not outputing dupli-objects using evaluated object
pointers, leading to crash because original object are not supposed to
be drawable.
This moves the backface culling to the fragment shader to avoid all the
limitations of the current system. This has a cost but it is unlikely that
bone drawing will be a bottleneck.
Usual issue from those dear Bone pointers in pose data...
Note that this is more like minimal-risk, quick fix, it's nothing like
'nice to have' code. Think proper solution would be to refactor handling
of those kind of 'caches' to ensure they are valid/up-to-date in a much
easier way, at the very least.
Ideal solution being to get fully rid of those horrors, of course, but
let's not dream here. ;)
Adds the invert vertex weights option to the Warp Modifier. Setup in the same way as the other modifiers.
Uses the existing flag char that is labeled unused.
Differential Revision: https://developer.blender.org/D6720
Clicking and dragging (aka tweaking) an unselected item should select
it, deselect others and translate. After rB395dfff103e1, it would not
deselect others anymore.
Adds the invert vertex weights option to the Displace modifier.
Adds a flag and char padding to the Displace modifier DNA for the invert group boolean.
Differential Revision: https://developer.blender.org/D6686
The function `gpu_texture_create_tile_array` checked for a valid
tile ibuf when determining the packing location. During the actual packaging it didn't.
As the tiles are already ignored when selecting the packing location, we
can also ignore it when copying it to the glTexture. Therefore this
patch removes the existing BLI_assert and replaces it with a NULL check.
Reviewed By: Brecht van Lommel
Differential Revision: https://developer.blender.org/D6738
Currently the UI for Symmetrize is confusing, for a few reasons:
- It exists as a sub-panel to the Dyntopo panel, even though it doesn't require Dyntopo to work
- It is inside a panel called Remesh, which is confusing because we now have a different Remesh panel
Changes:
- Put the Symmetrize controls in the Symmetry panel
- Rename Optimize to Rebuild BVH and move to the Sculpt menu
- Change the poll function - apparently it doesn't require Dyntopo
Differential Revision: https://developer.blender.org/D6735
Reviewed by Pablo Dobarro
Improvements to keyboard shortcuts shown in menus, mostly for Macs.
Differential Revision: https://developer.blender.org/D6730
Reviewed by Brecht Van Lommel
Using ellipsis character for line continuation since that glpyh is now narrower.
Differential Revision: https://developer.blender.org/D6728
Reviewed by Brecht Van Lommel
The function `modifiers_disable_subsurf_temporary` disables
temporarily only subsurf modifiers with the `On Cage` option enabled.
But a modifier can act on cage even with this option disabled.
Differential Revision: https://developer.blender.org/D6722
This meant a non-active selected object would be skipped based on the
mode last used, mostly an issue with 'Lock Object Modes' disabled.
Other transform operators (such as snap) don't do this.
Partially reverts b8aff06466
Font binaries updated with consistent keyboard symbols. Footer events now have per-platform icons.
Differential Revision: https://developer.blender.org/D6055
Reviewed by Brecht Van Lommel
Specifically, this happened when automatically hiding and showing of the
menu bar and dock was enabled in the system preferences.
Differential Revision: https://developer.blender.org/D6710
Some code was removed to avoid storing the combined pass when viewport
rendering other passes. But we can keep this by default, Blender overrides
the list of passes entirely.
- New bespoke icon for the universal brush toggles
- New icon for Redirected Folder (currently unused)
- New icons for handle types (Auto Clamped, Auto, Aligned, Vector, Free)
- New icons for Sequencer Text Strip anchor (Top, Bottom, Left, Right, Centre)
Thanks to Andrzej Ambroż for the new icons
Pixar recently released USD 20.02 [1]. I think it's important for people
to be able to figure out which version of the USD library is used in
Blender.
[1] https://github.com/PixarAnimationStudios/USD/releases/tag/v20.02
This commit exposes the USD library information via `bpy.app.usd`, and
includes that info in the `system-info.txt` saved via Help → Save System
Info.
Reviewed by: brecht
Differential Revision: https://developer.blender.org/D6724
Preset tracking settings is missing a comma here.
Without a comma this will result in an implicit string concatenation.
This was probably not wanted.
Reviewed By: sergey, Blendify
Differential Revision: https://developer.blender.org/D6717
Needed to protect against a case where clamp overlap limited
the offset to approximately zero, and the snap-to-pipe code
would therefore encounter an almost degenerate profile and
fail in matrix inversion.
It changes name to be blender-<version>-linux64.
Since CentOS is used as a base host for builds there is no real need
in specifying libc version. Is unlikely anything older could be used
anyway.
Also make bitness to be the same as windows. It is something what
users will read easier.
This patch fixes {T73051}. The cause of the issue was the absence of
relations in the depsgraph between IK solvers of overlapping IK chains.
Reviewed By: sergey, brecht
Differential Revision: https://developer.blender.org/D6700
Dont act on a base which cannot be found in the viewlayer.
Oversight in rB41858a73111d.
Maniphest Tasks: T73499
Differential Revision: https://developer.blender.org/D6721
- Remove the only_render arg from ED_editors_flush_edits
was only used in one place, the '_ex' version can be used instead.
- Split out the single object version of this function as currently
flushing is being done in-line, often only accounting for edit-mode,
ignoring sculpt mode for e.g.
Only Metaballs are left unsupported.
However, the implementation does not match 100% with cycles which converts
all objects to meshes.
Fixes T63424 EEVEE: Normal map node doesn't works with curve objects
This commit cleans up the flow emission code (i.e. the code that determines where flow is generated). It also addresses an issue with initial velocities.
Related issues (that might be fixed through this commit) are: T73422, T72949
Fluid domains with (auto-generated) particle systems were set to wireframe drawtype automatically. This was a convenience feature that made particles visible immediately. As the auto switching is sometimes a bit annoying though this has been removed. Instead, the Quick Liquid Operator switches the viewport shading type to wireframe.
- Initial velocities are no longer influenced by surface distance value.
- Added optimizations for different flow types (e.g. skip part of loop for liquid flow objects).
- Comments style cleanup and removed old todos.
This code is currently not in use and so removing it based on the YAGNI principle. If there really is need for a high-res emission loop it could be easily added again. However, I believe for the smoke noise it is sufficient to upscale the base emission map. A high-res emission map can easily be achieved by increasing the base resolution. Note also that in the new fluid system base loop and noise loop are decoupled making the need for a high-res emission loop even more unneccessary.
Code cleanup. Fixes a mismatch between definition and declaration of fsmenu_insert_entry().
Differential Revision: https://developer.blender.org/D6712
Reviewed by Julian Eisel
Changes to Info Editor making it easier to read. Only visual changes, no functional changes.
Differential Revision: https://developer.blender.org/D6491
Reviewed by Julian Eisel
This fixes the crash with pose_ik_segments = 0 in old files.
Some files were saved with the property set to 0, so the versioning code
won't work in that cases. I also changed that to fix the crash in those
files.
Reviewed By: brecht
Maniphest Tasks: T73054
Differential Revision: https://developer.blender.org/D6663
Tool system handlers had higher priority than the overlayed animation
"regions" (fake regions in fact). Give animation (and View2D) handlers
priority over toolsystem handlers.
Gizmo's still have higher priority, but I'll leave that unless there's a
reason to change.
More concretly, the crash would happen if a filebrowser is opened while
an error popup is visisble **in a different window**.
Code assumed the popup to be in the active window/screen, but it may
actually be displayed in a non-active window. Temporarily override
context to ensure this assumption is correct.
In blender 2.79 you could use a render result as a camera background
image. This is useful during layout/compositing. During Blender 2.80
development there were 2 issues introduced that removed this feature.
* to receive a render result the image required a lock. This lock wasn't passed and therefore no image was read from the result. Generating an GPUTexture from an Blender image also didn't do the locking.
* the iuser->scene field wasn't set what is required for render results.
This change adds an optional `ibuf` parameter to `GPU_texture_from_blender` that can be passed when available.
Reviewed By: fclem, brecht
Differential Revision: https://developer.blender.org/D6684
We have the problem that it's very hard to see which animation channels are selected. This is a very small tweak to make it more obvious, using the same text highlight color as the Outliner.
Really it should be overhauled to be more exactly like the Outliner, but this is just a quick fix to make it reasonably visible.
Differential Revision: https://developer.blender.org/D6708
Reviewed by Julian Eisel
Need to also offset markers in BKE_sequence_base_shuffle_time().
Also clarify/correct related comments.
Maniphest Tasks: T73014
Differential Revision: https://developer.blender.org/D6555
taking proportional size into account
Transform would always move all keyframes (e.g even when Proportional
size is 0.0).
'calculatePropRatio()' was setting td->factor correctly, but this was
not being considered in 'applyTimeTranslateValue()' if there was no
action [which greasepencil and masks do not have].
Maniphest Tasks: T55622
Differential Revision: https://developer.blender.org/D6697
What I removed here used to be needed in an earlier version of the
patch, to avoid use-after-frees. This is not the case anymore, so it
seems we can remove this.
Generally the context region code is very weak and inconsistent. If
things were done properly, the line I added shouldn't cause issues -
it's logically correct.
Also fixes T73475.
button in the Image slot in the Brush panel
Caused/exposed by rBaf9ca138ba7b
Also relevant was rB828905190e12
Above commits moved the WM_UI_HANDLER_BREAK around so it would not be
returned anymore.
We need WM_UI_HANDLER_BREAK, otherwise we wont get through to to
KM_RELEASE. There are two places that explicitly check for KM_RELEASE
(call to `ui_do_but_extra_operator_icon` in both
`ui_do_but_SEARCH_UNLINK` / `ui_do_button`), and without the above we
only ever get here with KM_PRESS.
This patch restores the behavior regarding WM_UI_HANDLER_BREAK to what
is was before the two culprit commits.
Checked that both this report (T72089) and T69755 are working.
Maniphest Tasks: T72089
Differential Revision: https://developer.blender.org/D6363
Since changing the active particle system uses NC_OBJECT | ND_DRAW
for 'RNA_def_property_update()' we need to redraw/refresh
BCONTEXT_PARTICLE in 'buttons_area_listener()' as well.
Maniphest Tasks: T67654
Differential Revision: https://developer.blender.org/D6702
Since we can now use separate icons for the items in the File Browser
lists, the following rounds out the "System" list for Linux.
Adds a few more folders that are available in most distributions
at installation, but only if they exist.
Try to never do operation twice and try to use MADD operations. Even if this
is very unlikely to make any difference, it can help compilers do some
optimization. I did not measure any difference as probes have much higher
impact on render time because of texture lookups.
Note that disk light is currently the most expensive light type so it
does not hurt to micro optimize.
Enable and Disable Edit Menu items based on whether those actions are currently applicable.
https://developer.blender.org/D4846
Reviewed by Brecht Van Lommel
This was due to the fact the drawing code was expecting the editpoints
to be equaly spaced. Reuse the code in particle.c to output the select
mask in red color channel of the particle (which is unused in new code).
This dependency was removed in fd0bc7e002,
as there already were dependencies ANIMATION -> PARAMETERS and
PARAMETERS -> CACHE, making ANIMATION -> CACHE unnecessary.
Upon subsequent inspection, the ANIMATION -> PARAMETERS dependency was
there due to the fallback behaviour in
`RNANodeQuery::construct_node_identifier()`. Now this is no longer
relied upon, and the required relation is made explicit again.
To issues caused this:
* The code to launch the filebrowser from a button didn't respect popup
menus stored in context
* There was actually no popup menu region stored in context for this
case. Doing so *may* introduce other issues, but the way I did things
already fixes all issues I've found.
Instead of changing the modifiers behavior, we make sure to always use
the data->totcol instead of the ob->totcol. Also we centralize getting
this number to avoid future issues.
Fix T72593 Blender crashes when separating mesh
Fix T72017 Crash on set visibility change
Not fully fixing it for now since I do not see an easy way to remove freed
pointer from libmap, and it does not seem to be an actual, practical
issue currently, but eeek...
In a minimal render engine blender crashes as it cannot create a preview
of the material yet. This patch adds a failsafe test for this specific
scenario.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6633
This was a missing notification because of wrong type. Also remove
the id tag as the changes are still picked up by workbench and does not
affect geometry or the particle system directly.
A big rework of the code now uses exact predicates for orientation
and incircle. Also switched the main algorithm to use a faster
divide and conquer algorithm, which is possible with the exact
predicates.
This changes the GL_TEXTURE_MIN_FILTER parameter from GL_LINEAR_MIPMAP_LINEAR
to GL_LINEAR_MIPMAP_NEAREST. It gives a sharper result, since with these
settings only down scaling is used.
Differential Revision: https://developer.blender.org/D6662
The NLA Editor sidebar is currently out of step with the rest of Blender, and in particular the Sequencer, which offers similar features and functionality. This patch makes a number of UI layout changes to make it fit in.
- Remove Active Track panel, since the track names can just be edited in the source list on the left
- Rejiggered the categories to be more understandable. 'Strip' includes settings for the current strip, and 'Edited Action' includes settings for the action that is currently being edited
- Use single column split layout
- Use name + mute toggle configuration like in the Sequencer
- Fix an issue with the theme, where the sidebar background was too dark, breaking the bright -> dark hierarchical logic
- Use sub-panels for animated strip properties controls
Differential Revision: https://developer.blender.org/D6687
Reviewed by Brecht van Lommel
This is done with two-finger double tap on trackpads, and is usually used to
zoom into an element under the cursor.
It can now be used for toggling local view in the 3D viewport.
Differential Revision: https://developer.blender.org/D6588
Now it shows "Macintosh HD" instead of "/", and a different icon for
removable and network drives.
Differential Revision: https://developer.blender.org/D6683
This is a step towards Wayland and headless rendering support, using EGL
instead of GLX. The EGL backend is not enabled by default, it can be tested
using WITH_GL_EGL=ON.
Differential Revision: https://developer.blender.org/D6585
This 'fixes' T68554: 'API mathutils.geometry.tessellate_polygon returns
bad results sometimes' by documenting the limitations of the current
implementation.
I've also added a unit test for the function, so that any change in this
behaviour will get noticed.
No functional changes.
Adds the Invert Vertex Group weight option to the Cast modifier.
Uses the same setup as similar modifiers invert weight.
Adds a boolean invert property next to the vertex group string in the modifier
and subtracts the current vertex weight from 1.0f if it is turned on.
Differential Revision: https://developer.blender.org/D6671
Minor modifications by @mont29.
While the behavior of this shortcut can be argued (it only works while
hovering the possibly invisible markers region), it was not my intention
to remove it entierly from the Graph and Sequencer Editor. Only wanted
to copy it.
Mistake in 6a49161c8c, the tool-header region was not created when
creating a new VSE editor (as opposed to an existing one in some
workspace).
There was also no way to get the tool-header to show in such cases.
Logic to open menus on hover changed since 2.7x for convenience
switching between popovers in the top-bar.
This also made hover open menus in situations where it isn't useful.
Restrict this to buttons placed side-by-side.
This implements a better heuristic for identifying if cloth or hair is
being dealt with (checking hairdata, instead of primitive_num).
The issue was caused by a change in primitive counting in rBd42a7bbd6ea5
I'm also adding some safeguards to avoid ever computing pressure for
hair. This shouldn't really be necessary, but it's good to be sure.
When building with precompiled libraries on Linux, CMake used boost libs
from the system outside the lib dir. This restricts CMake to use only the
libraries from the precompiled libraries.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D6659
Remove entries for the transform tools, since this is no longer necessary after recent changes. The default keymap entries are simply used instead, since the MMB is now added automatically.
Profile calculation now happens in a single pass rather than being spread
throughout the process. This means each profile will only be calculated a
single time.
Reviewed By: howardt
Differential Revision: https://developer.blender.org/D6658
Changes the default View3D mouse cursor to the OS-supplied arrow pointer. Subsequent cursor changes will now be set per-tool instead.
Differential Revision: https://developer.blender.org/D6485
Reviewed by Campbell Barton
The dependency graph has to know whether a driver must be re-evaluated
every frame due to a dependency on the current frame number. For python
drivers it was using a heuristic based on searching for certain sub-
strings in the expression, notably including '('.
When the expression is actually evaluated using Python, this can't be
easily improved; however if the Simple Expression evaluator is used,
this check can be done precisely by accessing the parsed data.
Differential Revision: https://developer.blender.org/D6624
The full name was "Topology Slide/Relax", but it didn't fit in the
toolbar UI. This was causing some problems:
- The mesh filter that does the same thing is called "Relax"
- We may want to add a "Topology Brush" tool in the future that is more
oriented to retopology task (like creating strips of quads), so by doing
this we avoid having two tools with the same name in the UI.
Reviewed By: billreynish
Differential Revision: https://developer.blender.org/D6590
This patch applies the original object scale to the coordinates of the
extracted mask mesh. Without this patch, the mesh was no taking the
original object scale into account and the shrinkwrap will fail.
The other solution would be copying the objecty scale to the new
extracted mask object, but I would like to avoid entering sculpt mode
with scaled objects as it may produce wrong behaviors in some tools.
Reviewed By: jbakker
Maniphest Tasks: T71373
Differential Revision: https://developer.blender.org/D6207
The 2D paint cursor in the 3D view does not support scene radius in
weight and vertex paint mode. This was also the case in 2.80 with the
old cursor. The option to change the units from view to scene is not
available in the UI in those modes, so I think it makes sense to remove
the support from the code.
The ideal solution could be supporting the new cursor in all paint
modes, but that is going to take more time and it is going to be
problematic in texture paint. After doing this, we can bring this code
back and add the option to change the units to the UI.
Reviewed By: jbakker
Maniphest Tasks: T73080
Differential Revision: https://developer.blender.org/D6586
It was printing "frame" time instead of FPS.
Other interesting thing to get solved is to solve "drop" in FPS
when there is an idle time in user input.
Current usecase is limited to observing FPS when there is a continuous
stream of events: for example, keep moving vertex in edit mode while
watching debug prints.
Boost could have picked up system-wide libbz2-dev installed and enable
this compression in iostreams. Nothing really wrong with this, but it
makes it so final Blender binary depends on bz2, which breaks default
linker flags.
This commit makes it so Boost is not using libraries which we don't
need, simplifying linking setup.
Differential Revision: https://developer.blender.org/D6668
The FPS here is measured based on a timestamp from when depsgraph
was previously evaluated.
Allows to ease investigating performance improvements/regressions
which are not related on animation system but on modifications on
a single frame (such as transforming vertex in edit mode).
Was caused by recent refactor of dependencies in 517870a4a1.
While there is no fully reliable solution to this issue other than
making TBB a dynamic library dependency (as documentation tells us
to do), there seems to be simple workaround which doesn't require
deeper changed in build process and packaging.
Tested on Brecht's computer who managed to reproduce the issue on
Linux (T72015#857423).
When there are ID properties on an object, and these are animated and
used by a driver, the depsgraph has proper connections between ACTION →
ID PROPERTY → DRIVER.
When these properties are defined on a mesh, however, the depsgraph
relations are incorrectly created between GEOMETRY → PROPERTIES_EXIT →
DRIVER (because it's assumed that 'source = ENTRY' implies 'geometry').
This patch solves this by first checking whether the targeted property
is an ID property and handling it accordingly. This also made it
possible to remove some special cases from pose bone relations.
Maniphest Tasks: T73001
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D6571
`RNA_property_pointer_set()` was just broken when assigning to an ID
pointer IDProp, on both debug/checks and actual assignment.
That was at least affecting RNA copying and liboverrides area...
even when target is not a collection
Looks like this was (accidentally) removed in {rBbe9e469ead22}
https://developer.blender.org/
rBbe9e469ead227aee8d4c29b98a125cf599c5c8bb#change-pFg0VUTAHY2q
This could also result in a missing update in
object_handle_update_proxy, see T72083 for an example.
Thx @Macroni investigating!
Maniphest Tasks: T72083
Differential Revision: https://developer.blender.org/D6651
This is a more correct fix to the issue Brecht was fixing in D6600.
While the fix in that patch worked fine for linking it broke ASAN
runtime under some circumstances.
For example, `make full debug developer` would compile, but trying
to start blender will cause assert failure in ASAN (related on check
that ASAN is not running already).
Top-level idea: leave it to CMake to keep track of dependency graph.
The root of the issue comes to the fact that target like "blender" is
configured to use a lot of static libraries coming from Blender sources
and to use external static libraries. There is nothing which ensures
order between blender's and external libraries. Only order of blender
libraries is guaranteed.
It was possible that due to a cycle or other circumstances some of
blender libraries would have been passed to linker after libraries
it uses, causing linker errors.
For example, this order will likely fail:
libbf_blenfont.a libfreetype6.a libbf_blenfont.a
This change makes it so blender libraries are explicitly provided
their dependencies to an external libraries, which allows CMake to
ensure they are always linked against them.
General rule here: if bf_foo depends on an external library it is
to be provided to LIBS for bf_foo.
For example, if bf_blenkernel depends on opensubdiv then LIBS in
blenkernel's CMakeLists.txt is to include OPENSUBDIB_LIBRARIES.
The change is made based on searching for used include folders
such as OPENSUBDIV_INCLUDE_DIRS and adding corresponding libraries
to LIBS ion that CMakeLists.txt. Transitive dependencies are not
simplified by this approach, but I am not aware of any downside of
this: CMake should be smart enough to simplify them on its side.
And even if not, this shouldn't affect linking time.
Benefit of not relying on transitive dependencies is that build
system is more robust towards future changes. For example, if
bf_intern_opensubiv is no longer depends on OPENSUBDIV_LIBRARIES
and all such code is moved to bf_blenkernel this will not break
linking.
The not-so-trivial part is change to blender_add_lib (and its
version in Cycles). The complexity is caused by libraries being
provided as a single list argument which doesn't allow to use
different release and debug libraries on Windows. The idea is:
- Have every library prefixed as "optimized" or "debug" if
separation is needed (non-prefixed libraries will be considered
"generic").
- Loop through libraries passed to function and do simple parsing
which will look for "optimized" and "debug" words and specify
following library to corresponding category.
This isn't something particularly great. Alternative would be to
use target_link_libraries() directly, which sounds like more code
but which is more explicit and allows to have more flexibility
and control comparing to wrapper approach.
Tested the following configurations on Linux, macOS and Windows:
- make full debug developer
- make full release developer
- make lite debug developer
- make lite release developer
NOTE: Linux libraries needs to be compiled with D6641 applied,
otherwise, depending on configuration, it's possible to run into
duplicated zlib symbols error.
Differential Revision: https://developer.blender.org/D6642
We compile zlib as own dependency, but are not informing BLOSC
to use it. This leads to zlib symbols defined twice when linking
Blender: one set comes from libz.a and another one from libblosc.a.
Tested on Linux Debian testing and CentOS 7.5.
It is possible that this change on its own will lead to linking
errors after libraries are re-compiled, This will be fixed as
a dedicated fix to Blender's build system.
Reviewed By: brecht, mont29, LazyDodo
Differential Revision: https://developer.blender.org/D6641
The problem was related to the smooth of the strength. As the factor was very low and the value was smoothed , the result was almost nothing when the radius was very small. Now the factor is higher and the smooth is done after clamping pressure.
This adds some kind of dashing to the tube and cone limits. Although the
dashing is in object space and is not a good as old dashing. But it is
the least time consuming and least complex solution.
- Register toggle is in the Text menu
- Run Script and Refresh OSL become an icon buttons next to ID block
Differential Revision: https://developer.blender.org/D5686
Reviewed by Campbell Barton
The operator always expect to be called form a button.
But this is not always the case.
This does not bring functional changes when not called from python.
This commit will prevent the crash. It does produce a result that may not
be desirable (only briding one face of the cube), but at least it's better
than crashing.
`posix_memalign` requires the `alignment` to be at least `sizeof(void *)`.
Previously, `MEM_mallocN_aligned` would simply return `NULL` if a too small
`alignment` was used. This was an OS specific issue.
The solution is to use a minimal alignment of `8` for all aligned allocations.
The unit tests have been extended to test more possible alignments (some
of which were broken before).
Reviewers: brecht
Differential Revision: https://developer.blender.org/D6660
Currently, the Info Editor default theme settings aren't consistent with other areas such as the Outliner. This change makes it fit in more by using the same default values for things like the background and selected elements.
Differential Revision: https://developer.blender.org/D6496
Reviewed by Pablo Vazquez
In our animation editors, we have a feature called View Frame. Problem is, it is not self evident what this feature does. Does it show frame numbers? Does it show all frames? Does it frame the view? Does it frame the view?
What this does, is it moves the view to where the playhead is. With clearer naming, we can communicate this much more clearly.
This is just a simple patch that changes the name from 'View Frame' -> 'Go to Current Frame'.
Differential Revision: https://developer.blender.org/D6437
In the brush properties, some masking-related controls are currently available as a sub-panel in the Options panel, while others are in the Mask panel. I want to generally lessen the amount of controls in generic 'options' panels, since it's non-descriptive and rather a rather random grab bag of various controls. Better to group things in terms of their functionality with descriptive names.
This change consolidates the masking-related controls under one panel.
Differential Revision: https://developer.blender.org/D6577
Reviewers: Pablo Dobarro, Julien Kaspar
The contents of the Rotation panel was already disabled when a bake
exists, but the checkbox in the header wasn't. Since rotations are part
of the baked data, it doesn't make sense to enable/disable rotations
after baking.
As pointed out by user @Peter Fog (tintwotin):
The current name for this property is wrong for two reasons:
- Because it makes incorrect use of the term 'window' (in Blender these are called 'editors')
- 'Lock Time' is incorrect, because it has nothing to do with locking the time at all. In Blender, the time is already locked and always the same across editors, no matter what. This property only syncs the visible range, which is something quite different.
For this reason, this is changed to 'Sync Visible Range' instead. I also added a tooltip, which was completely missing.
This patch also adds this property to the Sequencer, where it was missing.
Differential Revision: https://developer.blender.org/D6632
Reviewed by Julian Eisel
Changes the icon used for 'duplicate into new window' to a new one created for this purpose.
Differential Revision: https://developer.blender.org/D6463
Reviewed by Julian Eisel
Fixes the invisible scrollbar when opening preferences by 'Editor Type' by always creating scrollbars for V2D_COMMONVIEW_PANELS_UI objects.
This removes a fix (T47047) for dpi issues when dragging out the toolbox. The new toolbox in 2.8 does not seem to have the same DPI issues.
Exact steps for others to reproduce scrollbar invisible (T69413):
change the 3d viewport to preferences and go to addons. The scrollbar will not be visible
To verify there are no DPI issues:
Apply the patch and open blender. Hide the toolbar. Split the 3d view port into 4 windows, such that one is very small horizontally, another is very small vertically, and a third is small in both directions. Drag out the toolbar arrow on all four viewports. The Toolbar will be the same size in each. For further verification, adjust the scaling and repeat.
{F7805968}
Also this is my first time submitting a patch. Some instruction is welcome!
Reviewed By: #user_interface, Severin
Maniphest Tasks: T69413
Differential Revision: https://developer.blender.org/D6050
This commit belongs to T72894. It's related to (my) previous commits on pointer exchanges (today + yesterday). It cleans up the functions by describing their usage in the comments, adds additional nullptr checks, and fixes the reference count responsibilities of newly created PyObjects.
I also changed the Collada importer scuh that the diffuse Alpha is now set
for the generated PBSDF Shader AND for the base color.
Note: we can not export the base color AND the color
from the PBSDF Shaders at the same time. And when importing
we do not know if the color was originally coming from a base
material or from a shader. Setting them both in import seems
to give the most consistent results.
I'm still not entirely sure what was going on - I know that the execute
region didn't get initialized correctly, but doing that at a later point
didn't fix the issue. Apparently forcing the header region to
re-initialize does fix it, even though I was sure this was redundant.
Also fixes a memory leak in UI code after preferences were opened.
This function is a very special refresh function just for floating
regions. _initialize is more consistent with ED_area_initialize() so use
that too.
Also adds assert.
The OptiX SRT motion expects a motion defined by translation,
rotation, shear and scale, but the matrix decomposition code in
Cycles was not able to extract shear information and instead
produced a stretch matrix with the information baked in. This
caused conflicting transforms between traversal and shading
and lead to render artifacts.
This patch changes the matrix decomposition to produce factors
inline with what OptiX expects to fix that.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6605
Add toolbar to sequencer regions.
A bit of refactoring has to be done in RNA space.
Currently there is only cut tool implemented to serve as template for
anybody who would like to add more.
Importers/exporters should show the sidebar by default (if they have
any settings). For some reason this was missing for the Alembic and
Collada exporters.
Over-lines are used in other large files (keymaps for e.g),
using this add-hoc convention for sections make it easier to
configure editors to jump between them (as I have locally).
Waveform is freed in `sound_load_audio()` when sound length is `queried by BKE_sound_info_get()`.
Add argument free_waveform, so `BKE_sound_info_get()` can skip waveform freeing, as it is not expected to alter sound.
Reviewed By: sybren
Differential Revision: http://developer.blender.org/D6053
Add Adjust Last Operation panel to Sequencer.
`OPTYPE_REGISTER` was removed form some operators and few properties were hidden
So they don't show up on the panel
Author: a.monti
Reviewed By: ISS
Differential Revision: http://developer.blender.org/D6210
Refactor code to use `eSpaceSeq_Proxy_RenderSize` or corresponding `IMB_Proxy_Size`
enum items directly.
`SEQ_PROXY_RENDER_SIZE_100` has assigned value 99 to distinguish from `SEQ_PROXY_RENDER_SIZE_FULL`.
This caused error in image size calculation and because of that image had to be scaled.
Author: EitanSomething
Reviewed By: ISS
Differential Revision: http://developer.blender.org/D6368
Sometimes on flat open vertices the thickness would suddenly start
to be jumping in powers of 2, also when bending a plane there is a
noticeable jump in the geometry.
When offset is set to -1 or 1 weird glitches happen.
See D6559 for test cases that failed.
Some modifiers (like particle system, but also some other physics
simulations) require extra steps besides creating the modifier itself,
so had to switch to higher-level `ED_object_modifier_add()` to add
inserted modifiers in an overriding object data-block.
Due to how particle modifier handles copying, some extra nasty
hacks/work-around were also needed to get it working.
This data is only used to get current time/frame value, which is never
mandatory to add a modifier.
Needed by incoming fix to support particles modifiers in liboverrides.
Sometimes on flat open vertices the thickness would suddenly start
to be jumping in powers of 2, also when bending a plane there is a
noticeable jump in the geometry.
When offset is set to -1 or 1 weird glitches happen.
See D6559 for test cases that failed.
The `give_parvert()` function was only considering the mesh's original
vertex indices when the parent vertex index was valid for the evaluated
mesh. However, when using the Mask modifier the evaluated mesh can have
less vertices but still have the parent vertex.
Since the `if (nr < numVertex)` condition wasn't used to prevent any
out-of-bounds access, and seems just an incorrect optimisation, it could
be removed.
Even though we build USD as static, it still feels the need to mark its
symbols with declspec(dllexport) which means the blender binary now exports
these symbols.
this patch fixes that unwanted behaviour, however USD libs still need to
rebuild before this becomes visible in the blender binary
Differential Revision: https://developer.blender.org/D6563
Reviewed By: sybren
The USD landing broke building with clang on windows
due to a couple of reasons:
1) Some incompatibilities in their headers [1] only one
of them was important for us and is included in our patchset
now.
2) clangs lld wanted the full path to the libusd_b library
when using the whole archive link option, while msvc can
figure it out from just the library name.
Tested with clang/msvc and msbuild and ninja generators
[1] https://github.com/PixarAnimationStudios/USD/issues/1030
When using the color picker on a image editor it tries to read the color
from the original image. When there is no original image the code
crashes during the determination of the UDIM tile number.
There are 2 approaches to solve this.
1. Modify `BKE_image_get_tile_from_pos` to support NULL pointers.
2. Modify `ED_space_image_color_sample` with an early exit.
This patch modifies `ED_space_image_color_sample` with an early exit.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6629
Panel alignment was only updated when panel size changed. Now we can
also recognize changes in the category tabs offset and tag panels
for alignment updates.
Previously the window manager would receive the GHOST event and then query the
latest tablet data from the window to go along with it. If multiple events were
queued, it would then use too new tablet data for handling older events.
Fixes T62565: tablet pressure not working on macOS with some devices
Removing meaningless distinction between NULL pointer and EVT_TABLET_NONE,
and initialize pressure and tilt to 1.0 and 0.0 respectively when no tablet
is used.
There were two strange things going on:
- Start frame was clamped to 1, even when frame 0 is always a valid number.
This also ignored the 'Allow Negative Frames' user preference.
- Start frame was only clamped when setting the end frame, so first setting
the end frame and then the start frame would result in a different result
than doing it in the opposite order.
This commit fixes both issues by:
- Clamping the lower bound of the start frame only if negative frames are
not allowed, and
- apply that clamp both when setting the start and the end frame.
FreeStyle line styles were not part of the dependency graph, and
blacklisted from the Copy-on-Write system. As a result, animated
FreeStyle properties would not be updated by the animation system,
resulting in T72213. There was an explicit call to run the animation
system on the original datablocks, but that was (for good reasons)
removed in D5394.
This commit adds the FreeStyleLineStyle datablocks to the dependency
graph and allows them to be handled by the CoW system. As a result
- the UI now updates properly when properties are animated, and
- animated property values are actually used when rendering.
This commit includes @Sergey's patch P1222, which unifies two bits of
code that did the same thing: check whether datablock type is covered by
copy-on-write.
Reviewed By: sergey, brecht
Differential Revision: https://developer.blender.org/D6609
Related to T62882
When the SupportsMapping modifier flag was added to the dynamic paint
modifier it was added commented out. It may have not worked back then,
but it seems to work fine now.
MOD_weightvgedit.c was already updated similarly.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6072
Similar to the Microfacet Closures, the Principled BSDF Sheen closure is
added at a high weight but typically results in fairly low values.
Therefore, the default weight is a bad indicator of importance.
The fix here is the same as it was back then for Microfacets:
Compute an average weight using the normal as the half-vector
and use it to scale down the sample weight and the albedo channel.
In addition to drastically improving denoising of materials with
sheen when using the new Denoising node, this also can reduce noise
on such materials considerably.
This would happen if a tile is found on disk, painting would actually
request that tile (because corresponding uvs were in that range), but
that tile was not added in blenders list of tiles in that Image.
Need to also check tile in `image_quick_test` (regardless of iuser
having passed).
thx @lukasstockner97 for additional input!
Maniphest Tasks: T73110
Differential Revision: https://developer.blender.org/D6578
selection
False positive when a sequences end would be the same as active
sequences start.
Also thx @sybren for the heads up to make this more readable.
Maniphest Tasks: T72546
Differential Revision: https://developer.blender.org/D6451
I doubt we can do much better. Most of the aliasing comes from the edge
detection which does not use a lot of samples. We could use more samples
but then the detection becomes way more complex and expensive.
The second issue comes from the reconstruction (AA pass) that only bleed
adjacent pixels in if their line direction is perpendicular to the offset.
This makes corner gaps on certain diagonals.
Thanks @mano-wii for providing the patch!
The original code assumed that when `animation_player_preset` was not
`CUSTOM`, the `animation_player` path would be empty. This is a bad
assumption, as it can be quite useful to be able to toggle between a custom
and a built-in player. As such, `animation_player` should not be reset to
an empty string when a non-custom player is chosen.
'Straight Cut' is actually documented as creating N-Gons. However, the
code was disallowing this. This is probably a mix-up as the "allow N-Gons"
option is documented as a "force quad/tri" option.
There is already code for this, it just wasn't working. Now add the
slash after checking for an empty string.
Differential Revision: https://developer.blender.org/D6568
The gflags library was not fully configured to be built as a static
library, only one of two files was changed.
Differential Revision: https://developer.blender.org/D6611
This aligns with the VFX reference platform 2020 along with the decision
to stick to Python 3.7, see T68774.
Blosc was downgraded to 1.5 as recommended by the OpenVDB documentation.
IlmBase and OpenEXR are now built together with CMake rather separately
using autoconf.
Differential Revision: https://developer.blender.org/D6593
We can't use the fast path when the mesh is used by mulitple objects and so
slower sculpting is expected then. But fake users should not affect this. This
also fixes the same type of error in a few other areas.
Cache file loading for mesh and particle files now works through the direct update_structures functions. The final cache mode now also only bakes the most essential files and is therefore not resumable anymore.
When computing the roll value coming from the handle bone, the code
was using some strange unexplained math. It probably works fine when
the difference with the 'zero roll' orientation is pure twist, like
is the case when called from mat3_to_vec_roll. However, it appears
to break when significant swing is involved.
The issue is fixed by using the proper Swing+Twist decomposition
utility function that was added in a recent version for drivers.
This improves performance by reducing the amount of false positives.
A self overlap is made, so the distance from the vertices in the
overlap nodes is actually added.
This patch suppress the OSL warnings resulted from ambiguous calls
to the normalize function. This is done by casting to the vector
type before normalizing.
Reviewed By: Brecht
Differential Revision: https://developer.blender.org/D6621
Blender supports locking vertex groups to prevent changes to the
weights. However, as mentioned in comments for D3837, it is hard
to use this because there is no interface for locking in 3D View.
This adds a red shade to bones that are associated with a locked
weight group during weight paint mode, as the first step to adding
such interface. The next step is adding a pie menu for lock/unlock.
Differential Revision: https://developer.blender.org/D6533
Once again, am not exactly sure why that was working before, and not
anymore - but in any case, doing that kind of update here is not only
useless (since we have to do it at the end of the whole
collections/objects duplication and remapping anyway), it is also rather
dangerous, as collections are currently in rather invalid states at that
point of the code...
Note that in ideal world, `BKE_main_collection_sync()` & co would be
lazy (setting only a flag, then code actually needing this to be valid
again should call some sort of `BKE_main_collection_sync_ensure()`).
Then we would not have to worry about such things (and we'd get nice
performance improvements in some cases, also in main remapping code,
etc.).
Food for some refactoring, some day...
Simplify a bit, and remove some now redundant remappings.
As a side note, rBac723db57fd8ba5 actually also fixed some unreported
issues (missing remappings to new objects/collections in new copied
scene, that were not previously handled by the 'custom' code).
There are almost certainly still some missing remappings around,
though...
rBac723db57fd8ba5 makes proper remapping of all pointers in Scene...
including the object pointers of bases in ViewLayers.
Using `BKE_main_collection_sync_remap()` makes sense here anyway
(compared to `BKE_main_collection_sync()`, it additionally clears caches
in ViewLayers and Collections), this whole code makes a lot of
remapping.
Note that I do not really understand *why* that was working OK
before rBac723db57fd8ba5. I.E. *why* not remapping at all ViewLayers'
base object pointers seemed to be OK...
This whole collection/viewlayer caching is very hard to follow, and
really needs a full rework at some point (just as much as ID copying
code in general, and scene copying code in particular, in fact).
Both actual Collection datablocks and the horrible master collection
should share the same code (there were already some differences,
although probably not critical, but some callbacks from
scene->master_collection did not have the 'not self' flag...).
### Description of the problem
Until now, it is only possible to correctly add a lightprobe in python via an operator:
`bpy.ops.object.lightprobe_add()`
### Description of the proposed solution
The idea of this patch is to fix the lack of consistency lightprobe creation without operator.
It allow creation of different lightprobe type directly via `bpy.data.lightprobes.new(name, type)` (such as for curves).
In order to make it possible I had to:
1. Add a function `BKE_lightprobe_configure` in charge of lightprobe settings configuration (avoid code redundancy)
2. Allow an object to take lightprobe datablock as data during is initialization.
### A short example of this patch usage
```
lp = bpy.data.lightprobes.new('some_name','PLANAR')
bpy.data.objects.new('toto', lp)
```
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6396
There were some visual artifacts when the spin gizmo had a rotation
greater than 360 degrees.
Avoids this by drawing the arc over the span of one rotation only
and adjusting the background color based on the rotation count.
This way we remove the need for the srgb boolean uniform and a lot of code complexity. However, mesh update is going to be a bit slower.
I did not benchmark the performance impact.
This also fix a typo in draw_cache_impl_particles.c and fix hair not using vertex color in workbench.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6610
This will make this consistent with the behaviour of other start/end
ranges in blender. So start/end will instead be adjusted to always
satisfy the "start >= end" requirement instead of clamping the values.
EEVEE Soft shadows were not rendered correctly during viewport
rendering. The reason for this is that during viewport rendering the
shadow buffers were only update once and not per sample. This resulted
that all the samples calculated the same shadow.
This fix moves the call to `EEVEE_shadows_update` from cache finished to
draw scene. This needs to happen before `EEVEE_lightprobes_refresh`.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6538
There is a cornercase when the user edits an uvmap, that is not part of
the material (yet). When this is the case the uvmap was not added to the
uv buffer and the 'pos' alias was not created.
This change will always request the active uv map when uv editing.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6534
Even though we build USD as static, it still feels the need to mark its
symbols with declspec(dllexport) which means the blender binary now exports
these symbols.
this patch fixes that unwanted behaviour, however USD libs still need to
rebuild before this becomes visible in the blender binary
Differential Revision: https://developer.blender.org/D6563
Reviewed By: sybren
Embree's local intersection routine was not prepared
for local intersections without per-object BVH.
Now it should be able to handle any kind of local
intersection, such as AO, bevel and SSS.
Differential Revision: https://developer.blender.org/D6602
UI
Seems like we need to set the error with the evaluated ModifierData.
Pass this to 'surfacedeformBind' and report with that.
Differential Revision: https://developer.blender.org/D6601
This integrates hair collisions with the new cloth collision system,
greatly improving reliability, and reducing the amount of hair-specific
code paths in the cloth code.
The removes all the point constraint based collision stuff, instead
implementing segment impulse based collisions, using the same collision
response code as the normal cloth solver.
The hair system can now also collide with the emitter if it is a
collision object.
Reviewed By: mano-wii, Sebastian Parborg
Differential Revision: https://developer.blender.org/D6545
This avoids the problem where Blender doesn't start because
the PYTHONPATH points to an incompatible Python version,
see T72807.
Previously we chose to assume people who set the PYTHONPATH know what
they're doing, however users may have set this for non Blender projects.
So it's not obvious that this is the cause of Blender not to launch
on their system.
To use Python's environment vars, pass the argument:
--python-use-system-env
Note that this only impacts Python run-time environment variables
documented in `python --help`, Access from `os.environ` remains.
Old pre 2.5 files may have had non active spaces stored that doen't have
a header. The 2.5 versioning only added headers for active spaces, not
inactive (so invisible) ones.
Newer versioning code assumed there to always be a header though.
Inserted a version patch to make sure there's always a header now.
Fixes error reported to Debian,
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=949035.
The last handle wasn't corrected, also, there is no reason
to flip the handles while sorting (checking the same handles many times)
move this into it's own loop.
It's not certain this fixes the issue since I can't reproduce the crash, but
the code was wrong in any case.
Thanks to Ray Molenkamp and Anonymous for finding this.
`ARegion.alignment` unfortunately is a mixture of value and bitflag
enumerations. When checking for left/right/top/bottom region alignment,
the flags have to be masked out usually.
Most of the fixed cases here probably didn't cause issues in practice,
but could in fact break at any point when surrounding logic changes.
In fact the assert in #region_visible_rect_calc() failed in an older
file from https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=949035. This
fixes it.
Outliner tree building code was not handling properly empty libraries
(i.e. Lib datablocks in our bmain which have no used actual data
anymore).
Main issue here is unclean states of indirect hierarchies of linking
involving several libraries after undo operation.
This is not a critical issue though, just annoying and untidy.
Just some rewording of the documentation of `Particle.uv_on_emitter()`,
so that it no longer refers to 'derived mesh' but 'evaluated mesh', and
document that it expects a modifier from an evaluated object.
No functional changes.
The floating "Move" redo panel showed transform values with less
precision than in other places (e.g. sidebar and properties editor).
With Millimeters as unit it would even round to full integers, which
may be an issue since you typically work at higher precisions with this
unit.
Note that this only applies to the visual precision, internally we use
full floating point `float`s still.
Fixes T70367.
Previously the button would only show up in the status-bar located in
the same window that had the screen with the animation timer.
I don't see a reason not to show the button in all status-bars instead,
other animation feedback is also displayed in all windows.
Fixes T72194.
Based on @fclem's suggestion in D6421, this commit implements support for
storing all tiles of a UDIM texture in a single 2D array texture on the GPU.
Previously, Eevee was binding one OpenGL texture per tile, quickly running
into hardware limits with nontrivial UDIM texture sets.
Workbench meanwhile had no UDIM support at all, as reusing the per-tile
approach would require splitting the mesh by tile as well as texture.
With this commit, both Workbench as well as Eevee now support huge numbers
of tiles, with the eventual limits being GPU memory and ultimately
GL_MAX_ARRAY_TEXTURE_LAYERS, which tends to be in the 1000s on modern GPUs.
Initially my plan was to have one array texture per unique size, but managing
the different textures and keeping everything consistent ended up being way
too complex.
Therefore, we now use a simpler version that allocates a texture that
is large enough to fit the largest tile and then packs all tiles into as many
layers as necessary.
As a result, each UDIM texture only binds two textures (one for the actual
images, one for metadata) regardless of how many tiles are used.
Note that this rolls back per-tile GPUTextures, meaning that we again have
per-Image GPUTextures like we did before the original UDIM commit,
but now with four instead of two types.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6456
Not sure how the WorkSpaceLayout.screen pointer could end up being NULL,
but apparently that happens, or at least happened with older files.
Rather than just adding NULL-checks, prefer not keeping around those
invalid layouts at all.
It was too easy to end up with invalid region rectangles and we were
badly protected against them, so that they were hard to catch.
In fact we still create a main region for the top-bar, which ended up
getting a region height of -1. While this doesn't seem to have caused
issues in practice, we should prevent them entirely.
So idea was that at the end of region layout resolving,
`BLI_rcti_is_valid()` should return `true` for the region rectangle.
Further changes here ensure this is true: The `RGN_FLAG_TOO_SMALL` flag
is now set whenever there is not enough space for a region or if it
would get a size of zero or less.
Note: Should the assert fail, please do not just disable it and try to
actually address the root of the issue.
In Blender 2.80 Shift + H (`object_hide_view_set(unselected=True)`) used to
(temporarily) hide objects only if their collection was visible in the current
view layer.
This behavior was changed in rB0812949bbc3d7acfd1f20a47087ff973110aa177 (D5992)
by using `BASE_VISIBLE_DEPSGRAPH` for the decision which object's (temporary)
visibility should remain unchanged. Since the view layer visibility and depsgraph
visibility has been decoupled in said commit, the correct condition
to check is `BASE_VISIBLE_VIEWLAYER`.
This patch is a fix for T72302
Differential Revision: https://developer.blender.org/D6583
Seems that previous fix didn't work in all cases: Debian's build
environment didn't fully detect endianess, possibly due to typo,
possibly due to difference in various environments.
Using define magic from a more battle-tested project seems a safe
way to go.
There are more changes than just PPC since the upstream commit contains
full re-synchronization of all defines.
This commit updates numaapi to a latest library version from upstream.
rB4cde92303f3d made a MESH-only check (but that is no longer mandatory
since the Mesh is fetched by MOD_deform_mesh_eval_get instead of ctx-
>object->data now...). This would fail for projecting beveled curves
(resulting mesh missing), now also get the mesh for curves if we use
MOD_SHRINKWRAP_PROJECT (and only then).
Maniphest Tasks: T68489
Differential Revision: https://developer.blender.org/D6548
'Shift + S' is taken by toggeling Smooth Stroke in the keymap, should be
'Shift + Alt + S' (this also goes more in line with other editors, where
Alt is added to the shortcut used by 'Save' to do 'Save As')
Maniphest Tasks: T72803
Differential Revision: https://developer.blender.org/D6572
when using UDIMS
A more generic approach might be considered in the future (I assume
there are other operators around that need an update in shifting their
uvs) though.
Maniphest Tasks: T72648
Differential Revision: https://developer.blender.org/D6570
The Falloff property in the pose brush controls how much deformation
propagates through the ik chain, not by using the distance to the
vertices, so this option does not make sense (it does nothing in the
pose brush code and it disables the 3d paint cursor).
Reviewed By: jbakker
Maniphest Tasks: T73070
Differential Revision: https://developer.blender.org/D6574
When a node was partially/fully hidden, this was causing the mask flags
to update incorrectly because it was not checking all vertices, so they
were assigned the fully_masked state and not updating in the transform
tool and mesh filter.
Reviewed By: jbakker
Maniphest Tasks: T73094
Differential Revision: https://developer.blender.org/D6573
The PBVH usually is rebuild after a topology change, so it does not make
sense to keep the previous dynamic mesh preview vertex list. This may
cause a crash is the number of vertices of the new mesh (and preview) is
larger previous one. Now the list is deleted with the PBVH and a new one
will be generated using the new mesh when the cursor is updated.
Reviewed By: jbakker
Maniphest Tasks: T71712
Differential Revision: https://developer.blender.org/D6476
This add a basic sanity check that validates
the features we use from ffmpeg are actually
available
Differential Revision: https://developer.blender.org/D5999
Reviewed By: sybren
The USD landing broke building with clang on windows
due to a couple of reasons:
1) Some incompatibilities in their headers [1] only one
of them was important for us and is included in our patchset
now.
2) clangs lld wanted the full path to the libusd_b library
when using the whole archive link option, while msvc can
figure it out from just the library name.
Tested with clang/msvc and msbuild and ninja generators
[1] https://github.com/PixarAnimationStudios/USD/issues/1030
Simply loading factory settings and dragging an area separator
immediately after would cause an assert because of these invalid sizes.
Seems that since rB07499c04f612 we correctly initialize DPI related
UserPref values with 0, which caused DPI dependant initialization of
global areas to set ScrArea.global.size_min/max to 0 too.
Commit baeb11826b switched memory
allocation for the motion transform to use CUDA directly, instead of going
through abstractions. But no CUDA context was set active before those
were called, so the calls failed. This fixes that by binding a context beforehand.
The `do_versions_after_linking_280()` function was missing a placeholder
for the newly added versioning code. I copied the comments from the
`blo_do_versions_280()` function.
No functional changes.
During development of Blender 2.80 the `Object.hide` property was removed,
and later reintroduced in rB5e968a996a53 as `Object.hide_viewport`. Of
course there are some technical details missing in this summary, but this
is the view that's given in the 2.80 release notes.
FCurves on `Object.hide` weren't updated in versioning code, resulting in
the property no longer being animated. This commit corrects the RNA path
of such FCurves.
This might be useful in some places. Much of the code makes the implicit
assumption that the rectangle has valid coordinate order, good to make
it more explicit.
When splitting a Quad View by dragging a corner an quad area can become
negative size because of a one pixel offset to calculate the start of
the area.
This patch solves this to make sure that there are no negative areas. By
adapting the `ARegion.winrct`.
Reviewed By: Severin
Differential Revision: https://developer.blender.org/D6579
The original code had 0 as a magic number in the test whether the weight
belongs to a locked group, instead of comparing it to the actual group
number.
Thanks @mano-wii for providing the diff.
The Viewport Render Keyframes operator didn't include keyframes from
grease pencil animation or from mesh shape key animation (see T72861).
To find all related datablocks that could be considered as part of the
selected object, `BKE_library_foreach_ID_link(..., IDWALK_RECURSE)` is
used. To prevent relations like constraint targets from being visited,
the recursion is stopped when it finds a different object.
Reviewed By: mont29
Differential Revision: https://developer.blender.org/D6558
Never treat one of those horrorible 'IDs that are not real IDs' as
regular ID, and expect ID management code to do so. Unless there is a
very good reason, one should never explicitely pass those fake IDs to ID
management code directly.
In that specific case, user count is sort of 'disabled' in libquery
code, because master collections are not in bmain (`LIB_TAG_NO_MAIN`).
When sculpting and doing a image render the workbench used the sculpt
batch what is created by a different GPU context and would not show up.
In debug builds an assert for this case is checked. I rechecked all
`use_sculpt_pbvh` that it also checked if it was rendering. Only the
workbench deferred didn't do this for the default render mode.
With this change the user can render a workbench render directly from
sculpt mode.
Currently a curve's beveled geometry is built with duplicate geometry
along the seams between the "front," "back," etc, sections of the
curve. This builds them in one piece, resulting in smooth geometry.
Other than the duplicate geometry, the vertex positions are the same
as before.
Reviewed By: campbellbarton, mano-wii
Differential Revision: https://developer.blender.org/D6562
offset_meet creates offset lines that can't be directly intersected, so
the average of the points on each offset line is 'dropped' onto the
faces around the beveled vertex, which can depend on where
the loop starts.
This fix skips faces with the same normals as the "in plane" faces from
build_boundary.
Reviewed By: howardt
Differential Revision: https://developer.blender.org/D6521
The `poll_modal_item` was too restrictive.
`!validSnap(t)` already solves these cases, but for better readability
(and efficiency), I found it best to keep the `tsnap.mode` test.
This is because even if the glStencilMask is 0x00 the GL_DECR_WRAP and
GL_INCR_WRAP states still works and will modify the stencil.
Fix T73046 Overlapping parts of wireframes don't render at all in workench
with shadows turned on.
Patch from Habib Gahbiche (zazizizou) moves the "run operator and
compare mesh to a golden" paradigm used in bevel and boolean tests
into a general framework that separates the test specs from the
blend files. Then adds some other operator and modifier tests using
the new framework. Diff D5357.id20724.diff was applied.
New .blend files, modifiers.blend and operators.blend are needed
in the tests/modeling svn directory; those were separately committed.
When in sculptmode the wireframe coloring should always be disabled.
When an object was sculpted but had modifiers this was not detected and
the colors were still shown.
This fix will disable wireframe coloring for any object that is in
sculptmode.
Was caused by recent animation backup added to depsgraph as a part of
another bugfix.
This commit effectively disables the animation backup, restoring bug
related on handlers, but makes it possible to interact with objects
again.
Will re-iterate over handlers+animated properties.
It was possible to have object copy-on-write happening during
action's copy-on-write, which was causing access to a freed
memory from animation backup.
Solves crash reported in T73029.
Edge Index display were not showing up after performing
"Select Boundary Loops" when in face selection mode. The operation
tagged the mesh that the selection was changed, but in the end this
didn't do anything within the depsgraph.
The fix changes this to recalc the geometry of the object. What also
syncs the selectmode flags. A better solution would be to update the
select flags only. But that could be done as future work when we have a
performance problems.
Studiolights that were just created didn't had the
`STUDIOLIGHT_SPECULAR_HIGHLIGHT_PASS` flag. Without this flag the
studiolight the viewport didn't show the specular highlights and it
wasn't possible to switch the highlights on/off in the popover. After
reloading the studio lights this was possible.
This patch will always set the flag for any newly created, or being
edited studiolight, so the workbench render engine is fed with the right
data.
An optimisation in the collision system for cloth (static collisions),
broke the particle collisions, as they take motion into account. This
restores the moving BVH required for the particle collisions, while
keeping the optimisation for the cloth collisions.
Reviewed By: mano-wii
Maniphest Tasks: T71620
Differential Revision: https://developer.blender.org/D6560
Commit rB7e61e597253f3ca75f2fb86a57212ca750ffbbe8 broke viewport rendering when not displaying as halfs.
This fixes that by actually using the `scale` parameter that is passed into `film_map` and also de-duplicates code around it.
Differential Revision: https://developer.blender.org/D6557
The `optixAccelBuild` API throws an error when the property to get compacted size is passed in
without the `OPTIX_BUILD_FLAG_ALLOW_COMPACTION` flag set. This is not currently hit
because `background` is always true (set in `mem_alloc`), but would become an issue once that
is sorted out, so fixing it now to be safe.
OptiX always uses record-all behavior for transparent shadow rays, but did not check
whether the maximum number of hits exceeded the shadow hit stack. This fixes that.
- "Only Export Selected Objects" → "Selection Only", as this is the
standard used in other exporters.
- "Evaluation Mode" → "Use Settings for", as this is clearer for users
who aren't as familiar with the technical details of the depsgraph.
- "Use Instancing" → "Instancing", and moved the checkbox to the end of
the option list into its own box "Experimental". This looks a bit
cleaner than just plonking "EXPERIMENTAL" in the checkbox label.
This should only be used to change and recalculate region sizes in a DPI
independent way. To get the size with DPI applied, ARegion.winx/y should
be used instead.
Added note on this in comment.
When text drawing is disabled in the viewport the color of the sculpt
brh is set to the last used one. In th Light theme this is black what
makes it totally not visible.
This change will render the brush text using `TEXT_HI` as this is the
last one set when the text overlay is on.
The issue was caused by special handling of animation update after
manual edits in frame_post handler: to avid loss of manual edits
done on top of animated property. This was done as a separate pass
for non-animation update after frame_post did modifications.
This caused some other side-effect: non-modified animated property
was re-setting to the value which is used in the viewport.
Idea of this solution is simple: preserve values which came from
animation update through copy-on-write process. The actual process
of this is a bit involved: need to decode RNA path and do it twice
since f-curve might point to a sub-data which pointer will change.
Since this is only done for non-active depsgraph (aka depsgraph
used for render pipeline) this is probably fine since all this
extra overhead is just a fraction of overall render process.
Differential Revision: https://developer.blender.org/D6330
event_system.c has been collecting a lot of different functionality,
move generic query/access functions into their own file,
since these are used by operators and other parts of the windowing code
and aren't part of low level event handling.
Also move public last-property API to wm_operators.c.
Logic to convert double-click events into press events wasn't running
in the case an operator had a modal keymap, causing bevel for e.g.
to ignore keys pressed quickly.
Change event handling logic so modal handlers never
receive double click events, so checks for press/release are reliable.
While this is an old issue for mouse events in practice it wasn't
a problem since the first event typically executed/canceled.
Support for keyboard double-click exposed the problem
for all modal operators that take numeric input.
The exporter constructs an export hierarchy, and then traverses that
hierarchy from the root to the children. When an object is to be exported,
but its parent is not, it meant that this traversal from (great)parent to
children would skip these objects. This commit fixes that by inserting the
missing parents as 'transform only' into the hierarchy.
The way the USD exporter currently works, it is not possible to export
invisible objects. As such, the 'Visible Objects Only' option was
confusing.
Exporting invisible objects means obtaining invisible evaluated objects
from the depsgraph, which is not something that's currently implemented.
Once that's done, we can reintroduce this option.
This is in response to @brecht's remark in rBec62413f803e, where he
states that the approach was problematically interpreting the holdout
setting in a different way than what it was designed to do.
If we later want to add back a different "never include this in exports"
criterion, it can be easily done in
`AbstractHierarchyIterator::mark_as_weak_export()`. If such a criterion
should be file-format-specific, it can be done by overriding that
function in the file-format-specific subclass.
The original geometry referenced in `vtable` was deleted by the
`extrude_face_region` operator.
It is read soon after, so don't delete the original geometry
(param `use_keep_orig`).
This may have a small impact on performance.
transform_direction() can't handle parameters in constant address space.
Creating a local copy of the parameter satisfies the OpenCL compiler.
CUDA and CPU compilers should be able to optimize this away I hope.
As per T71295, the "duplicate+move" macro fails to store TRANSFORM_OT_translate properties once it's been used with rotation. I believe this is due to it being re-initialized with incorrect properties, reading bogus values from stored TRANSFORM_OT_rotate properties.
Force storing of actual operator id name instead of one defined in the macro, which in turn forces a name mismatch on initialization.
Reviewed By: #modeling, campbellbarton
Maniphest Tasks: T71295
Differential Revision: https://developer.blender.org/D6413
There are deeper issues than just updating the regression test .blend file
and the solution is dragging for far too long.
Considering this a known broken feature, which will either be fixed next week
or completely removed from the interface for the coming release.
Move redraw tagging to the gesture modal operator
to make sure this only runs when it's needed.
Caused by d591c8a350, which tagged the region to redraw when the
gizmos were tagged to refresh, however they wont redraw when hidden.
Thanks to @jbakker for finding the root cause.
In transform_motion_decompose, successive quaternion pairs are checked to be
aligned such that their interpolation is rotation through the shortest angle
between them. If not, the first in the pair was flipped. This can cause
problems for sequences of more than 2 quarternions, since flipping the first
in a pair might misalign the previously pair, if unlucky.
Instead, this change flips the second in the pair, which is safe when
iterating forwards.
Differential Revision: https://developer.blender.org/D6537
This patch adds support for the OptiX denoiser as an alternative to the existing NLM denoiser in Cycles. It's re-using the same denoising architecture based on tiles and therefore implicitly also works with multiple GPUs.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D6395
E.g. "Cube" would be placed after "Cube.001", which is not what you'd
expect. 2.80 handled this correctly.
Loosely based on D6525 by @radcapricorn, but found a bug in that and
prefered to do some further adjustments.
Also activates test for this case.
Adds tests covering a good amount of common cases and corner cases for
`BLI_strcasecmp_natural()`.
Could of course always add more tests for more cases, but don't want to
spend too much time on this.
This commit restores old metaball workaround which was forcing their
update from a single thread.
The root of the issue comes to the fact that metaball evaluation needs
to access metaballs from duplilists, so they are properly polygonized
with corresponding motherball which is outside of duplilist.
In a more ideal world this will be implemented in a way that will not
require iterating over all duplilists, but only through the ones which
actually contain metaballs for the given motherball. In practice this
ends up in a huge refactor in both relations builder (which meeds to
see whether there are metaballs in duplilists without actually
creating duplilist as it can not be done prior scene is evaluated)
and in metaballs area which need to use new relations information.
Additionally, metaball evaluation must become thread-safe, which is
currently not a case with dupli-object matrices. There might be issues
deeper in polygonization code which I am not aware of.
Having this forced single-thread evaluation is same as Blender 2.79
was doing.
Think it's better to have slower but simpler solution than to invest
time in refactoring area which requires deeper design changes.
Reviewed By: dfelinto
Differential Revision: https://developer.blender.org/D6539
Caused by own rBe02ecd599bdc.
Can happen with e.g. cloth.
Also fixes T59583
Maniphest Tasks: T72235, T59583
Differential Revision: https://developer.blender.org/D6547
The `in int flag;` in `gpu_shader_2D_edituvs_faces_vert.glsl`
don't have the values `FACE_UV_ACTIVE` and `FACE_UV_SELECT`.
Add face flags then.
Original patch is from @EitanSomething
Differential revision: https://developer.blender.org/D6520
This file had become disorganized, it wasn't clear which structs/flags
were deprecated.
- Add comments explaining what each struct is for.
- Use doxy sections.
- Remove outdated notes, unused flags.
- Group custom-data.
- Group deprecated structs in their own section.
T
This brush should be added to the set of brushes where we know which
vertices are going to be affected by the brush when starting the stroke.
This way we can limit the automasking only to those vertices instead of
flood filling the whole mesh from the active vertex.
All brushes that are not in this set will automask by flood filling the
mesh when starting the stroke. To improve this and make it work as most
users expect, we need a fast way to calculate topological distances on
high poly meshes.
Reviewed By: jbakker
Maniphest Tasks: T72251
Differential Revision: https://developer.blender.org/D6376
Before this it was possible to use the operator with Dyntopo sample mode
with a PBVH type GRIDS or FACES, causing a crash. Now we check first if
the PBVH type is correct before calling the sampling function.
We also check if the PBVH exists, which may also cause a crash.
Reviewed By: jbakker
Maniphest Tasks: T72647
Differential Revision: https://developer.blender.org/D6475
This was causing a crash when the mesh does not have the mask data
initialized. I also added the same check to mask extract as it works the
same way.
Reviewed By: jbakker
Maniphest Tasks: T72830
Differential Revision: https://developer.blender.org/D6513
This commits introduces the pose_ik_segments brush property in the Pose Brush. When increasing the IK segments count, the brush generates more segments and weights associations following the topology of the mesh. When moving the brush, these segments are transformed using an IK solver and they are used to deform the mesh.
When pressing Ctrl, the brush controls the segments' roll rotation instead of using the IK solver. The brush falloff controls how much rotation is propagated from the first to the last segment in the chain.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6389
working properly
'outliner_extract_children_from_subtree()' (introduced in
rB40a1c671655c) was extracting the children of non-matching parents
regardless of their own matching state.
Now properly filter the subtree prior to extracting.
Maniphest Tasks: T69246
Differential Revision: https://developer.blender.org/D6517
If pre-deselecting takes place, then flushing was not happening
('changed' never became true because no new faces were being selected).
This rectifies the logic. (also removed redundant double initialization
to false)
Maniphest Tasks: T72499
Differential Revision: https://developer.blender.org/D6459
Remove this pointer since it's linking Mesh data back to the object,
where a single edit-mesh may have multiple object users,
causing incorrect assumptions in the code.
Resolves dangling pointer part of the T72667 crash,
although there are other issues which still need to be fixed.
In EDBM_op_finish and EDBM_update_generic,
full Main lookups have been added which should be replaced with mesh
argument or the update tagging moved elsewhere.
Note that this is a bit clumsy having both edit-mesh and mesh,
BKE_editmesh_ensure_autosmooth & BKE_editmesh_lnorspace_update
are often called together, these could be made into a single functions.
The pointer is causing issues since two objects can share an edit-mesh,
removing in stages, see T72848.
Also fixes the material index being clamped by every object.
Now this also works for the UV Editor
Note: Only the transform tools currently support this. I would like to make this work for all the tools with gizmos, but need to find the nicest way to do it.
This was caused by a missing call to BKE_mesh_flush_hidden_from_verts()
when a SCULP_UNDO_HIDDEN undo step is processed.
Reviewed By: jbakker
Maniphest Tasks: T72700
Differential Revision: https://developer.blender.org/D6488
Users reported that Whiskey lake has the same issue as other intel
platforms where an extra glFlush is needed. This change will
add Whiskey Lake to that exception.
Patch provided by Philip Luk
There were two reasons the USD Exporter was listed as experimental:
- Originally there was no deduplication of mesh normals & UV coordinates
(resolved in rBf5e00f735106b5ec635806a4c795a2bc46ae8369), and
- the way materials were exported was incompatible with instancing with
USD 19.07. This seems to be resolved with the current version of USD
(19.11).
Blender (more specifically, `makesdna`) doesn't seem to like empty DNA
structs, so I couldn't remove all properties from
`UserDef_Experimental`. Instead I have just kept `char _pad0[8]`.
Reviewed by: campbellbarton
Differential Revision: https://developer.blender.org/D6519
For some reason socket flag and actual presence of link got out of sync.
Check for link to be present prior to access it.
Fixes crash opening file from T48684.
Was only using first ID instead of all of them.
Might have been causing issues when updating motion paths of
multiple objects.
Spotted by Jack C, thanks!
Nice reversed-logic mistake in rB693721cc7e7d.
How this could remain unnoticed for almost one year is fairly
mysterious, this should have basically broke all node tree copying,
would expect such bug to get reported within days, weeks at most...
Probably because that function is not that much used in current code.
Nice reminder also that those bloody nodetrees still need a lot of
cleanup/refactor/simplification when it comes to ID management code.
Reported/fixed as part of D6484, but this really needs its own commit.
In the Auto Merge & Split feature, multithreading was only used to
find duplicates between vertex and another vertex.
But with this patch, multithreading is now used to find intersections
etween edge and edge and between edge and vertex.
In my tests I noticed a performance improvement of around 180%
(0.017151 secs to 0.009373 secs)
Differential Revision: https://developer.blender.org/D6528
In 2.79 when moving a marker with `G` while holding `ctrl` the marker
would snap to one-second movements. This was (probably accidentally)
removed when the Timeline Editor was removed in rB5374865523faf253.
Reviewers: mano-wii
Differential Revision: https://developer.blender.org/D6527
Many things weren't working correctly in GP Edit Mode, such as select/deselect/invert all, basic tools, extrude, snap etc.
Now this mode works consistently with other modes.
`BLI_task_parallel_range` counts the number of tasks depending on the
number of items.
In the case of `BLI_bvhtree_overlap` the number of items is always
between 2 and 16, which makes it always run in single thread.
So, set the maximum number of items per thread to 1.
In my tests the cloth collision system (which calls that function)
went from 0.80fps to 0.88fps.
Differential Revision: https://developer.blender.org/D6523
As you can see, here is testing the "children" of `node1`.
So bvhtree is `tree1`.
This problem has never been observed because usually `tree_type`
of `tree1` equals `tree_type` of `tree2`.
Although not using multithreading, the `thread` parameter in the
`BVHTree_OverlapCallback` callback always returned a value between the
"number of threads".
This parameter should always be 0 in such cases.
Also a `BLI_Stack` was created for each "thread" and used.
This small overhead is no longer seen.
Differential Revision: https://developer.blender.org/D6510
T71495 describes two problems with animation of textures not showing up in
the dope sheet:
1. textures connected to force fields
2. textures of brushes
This patch resolves the first case.
An alternative would be to switch to iteration of dependencies using
`BKE_library_foreach_ID_link()`. This is a good idea to do at some point,
but adding these few lines was considerably easier & safer to do.
Enable by default since it allows selection
without gizmos getting in the way which is a common annoyance.
Although this has the drawback that subtle adjustments with
gizmos won't be detected until the drag threshold is reached.
This defaults to selection when not using a gizmo.
The previous behavior to drag anywhere can be set in the tool settings
or by selecting the fallback tool (Alt-W).
See: T66304
The entire particle system was skipped when the particle system itself did
not have any animation data. This caused the animation data on the texture
to be skipped.
Was caused by 6183688c35 (thanks ronsn for nailing it down!).
The issue is that order of copy-on-write operations is not defined, so
can not use flags set by that operation to make decision.
Since we are already respecting the User Preference for 'auto_smoothing'
in 'BKE_nlastrip_validate_fcurves', we should also do this for default
interpolation and handles.
Maniphest Tasks: T70986
Differential Revision: https://developer.blender.org/D6490
Previously the default values were left non-zero to avoid having to
update scripts. However, this meant it wasn't possible to setup
non-modal key bindings for smooth & randomize.
Now these operators follow logic of many other operators where setting
the value executes immediately, leaving unset runs modal.
Existing keymaps & scripts will need to be updated.
Addresses issue raised in f4a4ec8425.
Previously, compared to `Auto Merge` without `Split Edges & Faces`,
`Auto Merge` with this option ignored duplicates between selected
vertices. It only considered duplicates between selected vertices and
unselected vertices.
This is a regress and not a progress.
This commit implements this behavior, so this option matches the other
`Auto Merge`.
Currently the islands are created depending only on the visible UVs.
This can be confusing because compared to Edit Meshes, islands are created based on the selected elements.
T68284 shows a case where this confusion is observed.
Differential Revision: https://developer.blender.org/D6502
Previously, non-default alpha or float settings had be set manually.
With this change, the Add Tile and Fill Tile operators initialize
width, height, alpha and float from the currently selected tile
if it has a vaild ImBuf, otherwise from the first tile.
Previously the alpha was hardcoded to 0.7. Now it is possible to control
the cursor alpha by changing the alpha color of the cursor color
property. New alpha default is 0.9. This, with the new saturated colors,
should make the cursor more visible on highdpi screens.
I also removed the cache location preview as it is too visible right now
with the new alpha and color values.
Reviewed By: billreynish
Differential Revision: https://developer.blender.org/D6433
Partial updates are not currently supported in the Workbench engine, so
we can skip the rect calculation during the stroke.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6444
This was causing an ArithmeticError in OpenVDB with a voxel size that
small.
Another solution could be scaling both the mesh and the OpenVDB grid
with opposite values to support a larger range of voxel sizes, but I
would like to have a better solution to the voxel remesher compatibility
with small objects.
Reviewed By: JacquesLucke
Maniphest Tasks: T72747
Differential Revision: https://developer.blender.org/D6498
In a recent refactor we splitted the lines extractor in `extract_lines` and
`extract_lines_loose`. When an object is in edit mode the extracted
lines loose also had to include a dummy bmesh edge iterator. This change
adds this missing dummy method.
Reviewed By: antoniov
Differential Revision: https://developer.blender.org/D6499
Blender crashes when selecting a marker in the 3d viewport that is from the non active scene camera. This patch will solve this crash, but introduced a new scenario that isn't thought out. In the new scenario it is still hard to select a marker via the 3d viewport.
I would expect that when selecting a marker in this case would select the camera where the marker belongs to and select the marker that is under the mouse button.
Reviewed By: Sergey Sharybin
Differential Revision: https://developer.blender.org/D6360
In Blender 2.81 Look Dev was renamed to Material Preview
also the old look dev HDRIs could also be used in both
cycles and eevee. A more generic name was need.
As shown in the T68805, non-sized bones (such as the resulting extruded
bone) have no direction or orientation.
This can be bad for operators like `extrude_move` since the user might
want the resulting bone to be aligned with the bone that originated it.
The solution here is to get the parent bone orientation in the
transform operator if the bone has no size.
Differential Revision: https://developer.blender.org/D6486
Rewrite the checks for determining if the solution is actually within
the triangle to fix stability issues when the correct solution is on
an edge, and step is very small, i.e. the solution is already very
close. Also, comment more clearly what is happening geometrically.
This should fix problems when vertices that should project exactly
onto an edge actually miss, resulting in weird spikes. This made
Target Normal Project unusable for the voxel remesher.
Currently the action channels are applied after the existing
transformation, as if the action controlled a child of the
bone. This is not very natural, but more importantly, the
transform tools are not designed to work conveniently with an
additional 'pseudo-child' transformation, resulting in effects
like an unexpected pivot location.
Implementing a Before mode that integrates the action channels
as if applied to a parent allows using the special transform
tool code intended for dealing with such constraints.
Note that in either mode, Action constraints should be added
in reverse order, putting a new constraint before the existing
ones that the Action was keyframed to work together.
In order to implement the option, extract a utility from
the Copy Transform constraint code for combining transforms
with special anti-shear scale handling that matches the
Aligned Inherit Scale mode.
The Before mode also requires switching the constraint to
the Local owner space, while the After mode can still use the
World space for efficiency as before. Since the constraint
doesn't have an Owner space option in the UI, this has to be
handled in an RNA setter.
For full backward compatibility, the original simple matrix
multiplication mode is preserved as the third option, but it
is not recommended due to creating shear.
Differential Revision: https://developer.blender.org/D6297
For no apparent reason, when building Alembic the script would always
re-download and re-extract the Alembic source code. This is no longer the
case, and it now only happens if the source directory is missing. Since the
source directory name contains the Alembic version, it will automatically
trigger a download+extract when the version changes.
Previously, when an unknown parameter was passed to `install_deps.sh`,
the script would just show "Wrong parameter!" without any context. This
can make it hard to figure out what's exactly going wrong. Now it prints
which parameter it thinks is wrong.
- Object/pose checks were performed in same loop,
so selected pose bones were moved instead of the object.
- Snap selected to cursor/active incorrectly used unique object-data.
The //Distance from Object// and //Distance from Camera// modifiers still used the old 2.79 matrix multiplication syntax.
Differential Revision: https://developer.blender.org/D6468
This was disabled as part of b66ae8259e which disabled
animation for display mode and other cases where it doesn't make sense.
However it's useful to be able to overwrite frame ranges,
adding this back.
This adds UDIM support to e.g. the Displacement modifier.
The implementation is straightforward: If the image is tiled, lookup the
tile based on UVs and shift the UVs into the tile's coordinates.
The problem is described in a comment in the change.
Short version: If a pass was used as a divide_type but also requested
explicitly (e.g. diffuse color), it was added to the passes list
twice because the names of the two requests didn't match.
Then, when searching for the pass to divide by, the wrong one (not
the one that the kernel was writing to) was picked.
A bug in llvm < 9.0.1 causes the compiler to crash when
openmp is enabled. Since mantaflow uses tbb we can safely
disable this flag temporarily for this module.
Reviewed By: sebbas
Differential Revision: https://developer.blender.org/D6446
Actually, to purge orphans datablock you need go to Outliner, enable Orphan mode and press Purge button (that sometimes is out of the view because the window is too narrow).
To have this option hidden make very difficult to users use and understand what means orphan data, so this patch just adds a new Clean Up menu to File menu with this option. This menu could be used in the future for more clean up options. To have a general Clean Up menu is common used in other softwares.
Reviewed By: billreynish, mont29
Differential Revision: https://developer.blender.org/D6445
This commit affects `id_sort_by_name()` and `check_for_dupid()` helper:
* Add a new parameter, `ID *id_sorting_hint`, to `id_sort_by_name()`,
and when non-NULL, check if we can insert `id` immediately before or
after it. This can dramatically reduce time spent in that function.
* Use loop over whole list in `check_for_dupid()` to also define the
likely ID pointer that will be neighbor with our new one.
This gives another decent speedup to all massive addition cases:
| Number and type of names of IDs | old code | new code | speed improvement |
| -------------------------------- | -------- | -------- | ----------------- |
| 40K, mixed (14k rand, 26k const) | 39s | 33s | 18% |
| 40K, fully random | 51s | 42s | 21% |
| 40K, fully constant | 40s | 34s | 18% |
Combined with the previous commits, this makes massive addition of IDs more
than twice as fast as previously.
This commit affects `check_for_dupid()` helper:
* Add a special, quicker code path dedicated to sequential addition of a
large number of IDs using the same base name.
This gives a significant speedup to adding 'randomly'-named IDs:
| Number and type of names of IDs | old code | new code | speed improvement |
| -------------------------------- | -------- | -------- | ----------------- |
| 40K, mixed (14k rand, 26k const) | 49s | 39s | 26% |
| 40K, fully random | 51s | 51s | 0% |
| 40K, fully constant | 71s | 40s | 78% |
Note that 'random' names give no improvement as expected, since this new code
path will never be used in such cases.
This commit affects `check_for_dupid()` helper:
* Further simplify the general logic of the code (we now typically only do
one loop over the list of data-blocks, instead of two).
This gives a significant speedup to adding 'randomly'-named IDs:
| Number and type of names of IDs | old code | new code | speed improvement |
| -------------------------------- | -------- | -------- | ----------------- |
| 40K, mixed (14k rand, 26k const) | 62s | 49s | 27% |
| 40K, fully random | 76s | 51s | 49% |
| 40K, fully constant | 77s | 71s | 8% |
Note that 'constant' names give little improvement, as in that case the first
loop over the list of IDs (checking whether base given name was already in use)
was aborting very quickly.
This commit affects `check_for_dupid()` helper:
* Fix (serious though rare) bug where several IDs could end up with
exact same name (happened with over 10k IDs with same very long name).
* Fix (relatively harmless) func reporting it did not change the given
name when it actually had truncated it.
* Sanitize handling of supported min/max number suffixes (it now handles
between 1 and 1 billion, which should be way more than enough).
* Sanitize general logic to (hopefully!) make it easier to follow.
* General cleanup (naming, comments, variables scope, remove dead code, etc.).
Note that general performances here remain the same, there is no
measurable gain or loss. Algorithm remain also the same globally.
Attempt to use a GHash to speed up checks of used names proved to be
much worse, just building the GHash would already take as much time as
the whole process with current code...
This alone can make e.g. adding 40k IDs with default name (i.e. 'fully
constant' case in table below) about 15-20% faster:
| Number and type of names of IDs | old code | new code | speed improvement |
| -------------------------------- | -------- | -------- | ----------------- |
| 40K, mixed (14k rand, 26k const) | 75s | 62s | 21% |
| 40K, fully random | 90s | 76s | 18% |
| 40K, fully constant | 94s | 77s | 22% |
Idea is to use a first pass, where we just check one item every nth ones,
until we have found the chunk in which we want to insert the new item,
then do a basic linear comparison with all items in that chunk as before.
Also, list is now walked backward, as in most common 'massive insertion'
cases new ID's names have higher number, hence ends up towards the end of
the list.
This new sorting function can be between a few percents and 50% quicker than
original code, depending on various factors (like length of common parts of
ID names, whether new IDs should be added towards the end or not, how high
in numbering we are, etc.).
Note: another, full bisecting approach was also tried, which gives a massive
reduction of comparisons (from n items to log2(n)), but it only gave minor
improvements of speed from original fully linear code, while adding a fair
amount of complexity to the logic. Only case it was shining was the unlikely
'very long, almost similar ID names' case (since `strcasecmp()` is rather
costly then).
Those tests are here mostsly to ensure ID name management is working as
expected (the code ensuring we never have two ilocal data-blocks of the
same type with the same name in a .blend file).
Note: Currently fails in some cases, fixes are incoming.
Note: Ideally this would be in C, but we already have too many tests
linking the whole Blender and its libraries, this is becoming a real
pain to link debug + ASAN + tests build these days... So until we find a
better way to handle those dependencies, sticking to simple python
scripts.
Since rB6bc6d016c5e7, outliner was not opening back up from the found
active element (but only its ID instead -- all occurances of this ID in
any collection).
Now expand from the active element as well (but only do this for the
first occurance of the corresponding ID)
Maniphest Tasks: T71844
Differential Revision: https://developer.blender.org/D6329
updates in particle editmode
Particles themselves were cleared correctly but this was not tagging
batch cache dirty.
Might move this to a utility function later [since it is used in more
places], but that is for after going over some more reports...
Reviewers: sergey
Differential Revision: https://developer.blender.org/D5925
are visible
Not freeing PTCacheEdit and tagging batch cache dirty on undo will have
a couple of consequences. This patch fixes:
- crash deleting a particle, then undo
- basically any edit (combing, ...), then undo will leave child hairs
untouched
- adding hairs (through mirror, add tool, ...), then undo will leave
'orphaned' child hairs
See also D5755 for a related discussion
Fixes the crasher mentioned in T69000
Might move this to a utility function later [since it is used in more
places], but that is for after going over some more reports...
Reviewers: sergey
Differential Revision: https://developer.blender.org/D5912
This feature makes it possible to do a viewport render (a.k.a.
playblast) by only rendering those frames on which the selected objects
have a keyframe.
The frames to render are stored in a `BLI_bitmap`, which has a bit for
each frame set to 0 (skip) or 1 (render). An alternative approach would
be to construct a set of all keyframes to render, but that would make
both constructing the list and looking up frames in the list more
complex.
The only thing this feature does is skip OpenGL rendering of a frame. As
a result, 'skipped' frames are still included in the output, but just
use the render result of the last-rendered frame. This is exactly what's
described in T72229.
Differential Revision: https://developer.blender.org/D6443
Reviewed By: zeddb
Design task: T72229
option (even though it is being used for the ADD brush)
Add the option to the UI in that single case.
Maniphest Tasks: T72473
Differential Revision: https://developer.blender.org/D6418
Make indices accommodate into the measures of edgelength and edgeangle
so this results in a nice stack of up to three rows.
Maniphest Tasks: T72128
Differential Revision: https://developer.blender.org/D6357
Since BM_uv_vert_map_create would return NULL for an empty mesh, code
would then return from uv_select_linked_multi [where it should just skip
and continue instead...]
Maniphest Tasks: T63407
Differential Revision: https://developer.blender.org/D6441
Note that this removes image_paint.detect_data from
UnifiedPaintPanel.get_brush_mode, I can't see why it's needed,
it causes issues where the texture paint UI isn't used when it can be.
The ratio for area stretching was packed into an unsigned int, but could
contain negative numbers. This flipped the negative numbers to high
positive numbers and rendered the wrong color in the stretching overlay.
I can remember during {T63755} I had to flip the sign to get the
correct result, but couldn't find out why that was needed. Now I know.
Reviewed By: fclem, mano-wii
Differential Revision: https://developer.blender.org/D6440
All panels were calling poll to draw in their section causing a lot of
repeated boiler plate poll functions.
Also rename 'PreferencePanel' to 'CenterAlignMixIn'
since this is it's purpose.
The tooltip was static, so it was the same for viewport-rendering the
current frame and for the entire animation. It is now different for
those two.
The structure of `screen_opengl_render_description()` is such that it
allows for adding a new description for a soon-to-come feature (T72229).
This resolves a logical problem using tweak as a fallback tool.
See: T66304#828742
The select action would immediately show the gizmo underneath it,
then the tweak would be handled by the gizmo instead of moving the item
under the cursor.
Currently this works by hiding the gizmo until the tweak event ends.
While it's simpler to check if the gizmo received a mouse-down event,
it causes flickering before each drag event which feels like a glitch.
This is optional for each gizmo type because there are cases where this
can be useful to activate the gizmo immediately (mesh rip for example).
Vertex colors behaved differently as the paint overlay mixed the colors
in display mode and the results was multiplied on top of the original
shading.
This patch will align the implementation to texture painting where the
colors are drawn by the workbench engine so the correct shading is
applied.
This also means that we don't show the vertex colors overlay when not
in solid mode.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6436
Introduced by {T67981}. We changed the python API for curve evaluation.
Freestyle still used the old call that failed. This patch updates
FreeStyle to use the new API. I checked other areas in freestyle but it
seemed to be the only `evaluate` for curves that is called directly.
Reviewed By: zeddb
Differential Revision: https://developer.blender.org/D6430
Due to recent refactoring of the overlay unification the camera limits
were also visible when the overlays were turned off. This was because
the `draw_extra` had an exception for when looking through the camera.
This change also takes the global hide overlays into account. So now the
camera limits will not be drawn when overlays are turned off. This also
fixed other camera related overlay drawing.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6394
Due to the refactoring of the overlay engine the draw caches were
changed. The sphere batch used to have positions and normals. After the
refactoring it didn't had the normals anymore. The normals are needed
for shading. As they were not there the look dev spheres were rendered
black.
This change add the `nor` attribute to `DRW_cache_sphere_get` batch.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6393
The previous cursor colors were chosen to match the toolbar icon color,
but the contrast on top of the default matcaps was not enough. This
commit uses more saturated colors, which are more visible on top of the
default matcaps.
Reviewed By: jbakker, billreynish
Differential Revision: https://developer.blender.org/D6194
With the brush reorganization included in D5928, the brush panels had covered the topbar and make impossible to use the drawing tools.
Now, the options are grouped in Stroke popover, similar to Sculpt.
Also fixed some missing or wrong texts.
There are options for the exporter that are relevant for each export, for
example the choice between Viewport or Render settings, or whether to
export the current frame or an animation. It's better to have the options
panel opened by default.
When going from EDIT to OBJECT mode, Blender updates the object data from
the edit-mode data. This took care of renaming FCurves that animate Curve
control points when control points are added/removed, but this didn't keep
the FCurve groups intact. Since the FCurve groups are tightly connected to
the Action channels, it's hard to keep the group pointers intact during
this process. Instead of making the code even more complex in an attempt to
do that, I implemented a function (`BKE_action_groups_reconstruct()`) that
rebuilds the group channel pointers.
The call to `action_groups_add_channel()` had to be removed because it
updates the the next/prev pointers of the FCurve while we're looping over
them, causing infinite loops.
- Use abbreviations min/max
this is used throughout the existing API.
- Rename use_adaptive_stepping to use_adaptive_timesteps
since this is used with timesteps_min/max it's clearer
to use matching terms.
These attributes checked for any baked / baking
since this is a common test that was performed in layout code.
Also follow our naming convention - using an 'is_/has_' prefix
in this case since "cache_baked_data" reads as if it's used to access
the baked data.
Fixed sized strings are always initialized & this is not done
elsewhere before calling BLI_path_join.
Remove since it's not needed and makes it read as if the function
might not initialize the output argument.
gpu_draw.c had generic sounding utility functions which were specific
to smoke drawing.
Split into it's own file so the functionality is clearly separated.
Allows enabling debug prints in this file easily,
differentiates it from code which has been disabled for other reasons.
Also remove unused DEBUG_TIME.
The particle system needs some tweaks so that it can be used for particles created in Mantaflow (i.e. to read both FLIP and secondary particles from Mantaflow).
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3857
Cycles needs some smaller updates so that the up-res smoke wavelet noise and the liquid mesh speed vector export work correctly.
Reviewed By: mont29
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3854
A collection of smaller changes that are required in the /blender/source files. A lot of them are also due to variable renaming.
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3855
Probably the most significant changes are in smoke.c.
New functionality includes:
- support for adative time steps (substeps)
- write flow objects to grid structure so that Mantaflow can generate levelsets
- no more distinction between FLUID_3D and WTURBULENCE objects. Everthing that communicates with Mantaflow now lives in a FLUID object.
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3861
All new simulation parameters need to be registered in the RNA. Especially for liquids a lot of new ones need to be added.
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3858
Similarly to physics_fluid.c (in same directory) which handled the baking process for Elbeem, there is now physics_manta.c which handles it for Mantaflow.
There are two types of jobs: one for baking and another for freeing. The generic jobs will be used to bake / free specific parts of the simulation (e.g. bake mesh, free particles, etc.).
The jobs are only being used in the "modular" cache mode where the simulation has to be baked in parts.
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3856
Smaller changes in the build files to reflect the new Mantaflow macro.
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3853
All the new fluid simulation parameters need to be added the according structs (i.e. domain settings, flow settings).
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3860
With Mantaflow the current smoke modifier UI will accommodate both smoke and liquids.
In addition, there is now an option for Mantaflow liquids in the quick effects section ("Quick Liquid").
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3852
Files from /intern/mantaflow handle the communication between core Blender code and Mantaflow itself. It's the bridge to communicate with Mantas Python functions.
Code from /intern/mantaflow/intern/strings/ is pure Manta code and would likely need less attention in the review.
Reviewed By: sergey
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3851
Includes preprocessed Mantaflow source files for both OpenMP and TBB (if OpenMP is not present, TBB files will be used instead).
These files come directly from the Mantaflow repository. Future updates to the core fluid solver will take place by updating the files.
Reviewed By: sergey, mont29
Maniphest Tasks: T59995
Differential Revision: https://developer.blender.org/D3850
Previously this limit was rather high, but with UDIMs it's fairly easy
to reach this many images. Even though this exceeds the texture limit
on most hardware as far as I can tell, it should at least not crash.
The old code uses a fixed array which overflows eventually, this fix
replaces the array with a GSet.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6416
The ImageTextureNode incorrectly used the tile slot encoding for box-
mapped textures as well. Since box-mapping always generates UVs that
lie in the 1001 tile, there's no need to support tiles here.
This was caused by a call to paint_2d_bucket_fill that I missed in the
UDIM changes.
Now, this feature does not only not crash, but it even works with tiled
images!
- Use the clip mask context menu for the image editor.
- Remove redundant CLIP_MT_mask_handle_type_menu.
- Remove "Add" items (was only in image mask context menu)
as the convention is not to include these in the context menu.
The tools passed to _tools_flatten, _tools_flatten_with_tool_index
already have dynamic tool items expanded, so there is no need to check
for callable tool items.
The code checked for the presence of more than one tile before
substituting the tile number into the filename, so if a one-tile
UDIM was used (or all but one tile were culled), the substitution
was skipped and as a result the file was not found.
With this change, the code explicitly tracks whether substitution
is required, avoiding this problem.
This also fixes another problem: The Environment texture never
does substitution since it doesn't support UDIMs, but before the
syncing code still inserted the placeholder into the filename if the
user selected a tiled background image.
This makes a number of changes to the tool settings brush UI:
- All brush-related controls are now grouped together, so you can see which items are brush settings are which are not. Previously it was all jumbled together.
- The brush picker is in a separate panel, so that you can switch brushes without worrying about the settings, or vice versa.
- Custom Icon settings moved from the Display settings(now known as Cursor) to the Brushes panel.
- UnifiedPaintSettings panels are removed and the contained options are now next to their relevant setting with a globe icon toggle. This is not displayed in the header.
- 2D Falloff and Absolute Jitter toggles were changed into enums, to make it clearer what happens when they are on or off.
- Adjust Strength for Spacing option was in the Options panel in some modes, but in the Stroke panel in others. It is now always under Stroke.
- Display (now Cursor) panel was reorganized, settings renamed.
- 2-option enums are annoying as a drop-down menu, so they are now drawn with expand=True.
- Smooth Stroke and Stabilizer options in grease pencil and other paint modes are now both called "Stabilize Stroke", for consistency and clarity.
- De-duplicated some drawing code between various painting modes' brush options. I tried to keep de-duplication reasonable and easy to follow.
- A few more tweaks - see D5928 for the extensive list.
Most of the patch is written by Demeter Dzadik, with some additions by myself
Differential Revision: https://developer.blender.org/D5928
Reviewers: Pablo Dobarro, Bastien Montagne, Matias Mendiola
Shift-dragging with the Box Select tool would also select the nearest point.
Using a Click instead of Press event for selection (just like the default keymap) fixes this.
This was causing errors to show in the terminal.
Reviewed By: campbellbarton
Maniphest Tasks: T72409
Differential Revision: https://developer.blender.org/D6407
- The default value of USD attributes is now only set on the first sample.
Previously this was done for every sample (so the final default was the
last value, rather than the first value).
- More use of the sparse value writer, now also for UV coordinates, mesh
normals, and curve points.
This commit introduces the first version of an exporter to Pixar's
Universal Scene Description (USD) format.
Reviewed By: sergey, LazyDodo
Differential Revision: https://developer.blender.org/D6287
- The USD libraries are built by `make deps`, but not yet built by
install_deps.sh.
- Only experimental support for instancing; by default all duplicated
objects are made real in the USD file. This is fine for exporting a
linked-in posed character, not so much for thousands of pebbles etc.
- The way materials and UV coordinates and Normals are exported is going
to change soon.
- This patch contains LazyDodo's fixes for building on Windows in D5359.
== Meshes ==
USD seems to support neither per-material nor per-face-group
double-sidedness, so we just use the flag from the first non-empty
material slot. If there is no material we default to double-sidedness.
Each UV map is stored on the mesh in a separate primvar. Materials can
refer to these UV maps, but this is not yet exported by Blender. The
primvar name is the same as the UV Map name. This is to allow the
standard name "st" for texture coordinates by naming the UV Map as such,
without having to guess which UV Map is the "standard" one.
Face-varying mesh normals are written to USD. When the mesh has custom
loop normals those are written. Otherwise the poly flag `ME_SMOOTH` is
inspected to determine the normals.
The UV maps and mesh normals take up a significant amount of space, so
exporting them is optional. They're still enabled by default, though.
For comparison: a shot of Spring (03_035_A) is 1.2 GiB when exported
with UVs and normals, and 262 MiB without. We probably have room for
optimisation of written UVs and normals.
The mesh subdivision scheme isn't using the default value 'Catmull
Clark', but uses 'None', indicating we're exporting a polygonal mesh.
This is necessary for USD to understand our normals; otherwise the mesh
is always rendered smooth. In the future we may want to expose this
choice of subdivision scheme to the user, or auto-detect it when we
actually support exporting pre-subdivision meshes.
A possible optimisation could be to inspect whether all polygons are
smooth or flat, and mark the USD mesh as such. This can be added when
needed.
== Animation ==
Mesh and transform animation are now written when passing
`animation=True` to the export operator. There is no inspection of
whether an object is actually animated or not; USD can handle
deduplication of static values for us.
The administration of which timecode to use for the export is left to
the file-format-specific concrete subclasses of
`AbstractHierarchyIterator`; the abstract iterator itself doesn't know
anything about the passage of time. This will allow subclasses for the
frame-based USD format and time-based Alembic format.
== Support for simple preview materials ==
Very simple versions of the materials are now exported, using only the
viewport diffuse RGB, metallic, and roughness.
When there are multiple materials, the mesh faces are stored as geometry
subset and each material is assigned to the appropriate subset. If there
is only one material this is skipped.
The first material if any) is always applied to the mesh itself
(regardless of the existence of geometry subsets), because the Hydra
viewport doesn't support materials on subsets. See
https://github.com/PixarAnimationStudios/USD/issues/542 for more info.
Note that the geometry subsets are not yet time-sampled, so it may break
when an animated mesh changes topology.
Materials are exported as a flat list under a top-level '/_materials'
namespace. This inhibits instancing of the objects using those
materials, so this is subject to change.
== Hair ==
Only the parent strands are exported, and only with a constant colour.
No UV coordinates, no information about the normals.
== Camera ==
Only perspective cameras are supported for now.
== Particles ==
Particles are only written when they are alive, which means that they
are always visible (there is currently no code that deals with marking
them as invisible outside their lifespan).
Particle-system-instanced objects are exported by suffixing the object
name with the particle's persistent ID, giving each particle XForm a
unique name.
== Instancing/referencing ==
This exporter has experimental support for instancing/referencing.
Dupli-object meshes are now written to USD as references to the original
mesh. This is still very limited in correctness, as there are issues
referencing to materials from a referenced mesh.
I am still committing this, as it gives us a place to start when
continuing the quest for proper instancing in USD.
== Lights ==
USD does not directly support spot lights, so those aren't exported yet.
It's possible to add this in the future via the UsdLuxShapingAPI. The
units used for the light intensity are also still a bit of a mystery.
== Fluid vertex velocities ==
Currently only fluid simulations (not meshes in general) have explicit
vertex velocities. This is the most important case for exporting
velocities, though, as the baked mesh changes topology all the time, and
thus computing the velocities at import time in a post-processing step
is hard.
== The Building Process ==
- USD is built as monolithic library, instead of 25 smaller libraries.
We were linking all of them as 'whole archive' anyway, so this doesn't
affect the final file size. It does, however, make life easier with
respect to linking order, and handling upstream changes.
- The JSON files required by USD are installed into datafiles/usd; they
are required on every platform. Set the `PXR_PATH_DEBUG` to any value
to have the USD library print the paths it uses to find those files.
- USD is patched so that it finds the aforementioned JSON files in a path
that we pass to it from Blender.
- USD is patched to have a `PXR_BUILD_USD_TOOLS` CMake option to disable
building the tools in its `bin` directory. This is sent as a pull
request at https://github.com/PixarAnimationStudios/USD/pull/1048
This patch contains the work that I did during my week at the Code Quest - adding support for tiled images to Blender.
With this patch, images now contain a list of tiles. By default, this just contains one tile, but if the source type is set to Tiled, the user can add additional tiles. When acquiring an ImBuf, the tile to be loaded is specified in the ImageUser.
Therefore, code that is not yet aware of tiles will just access the default tile as usual.
The filenames of the additional tiles are derived from the original filename according to the UDIM naming scheme - the filename contains an index that is calculated as (1001 + 10*<y coordinate of the tile> + <x coordinate of the tile>), where the x coordinate never goes above 9.
Internally, the various tiles are stored in a cache just like sequences. When acquired for the first time, the code will try to load the corresponding file from disk. Alternatively, a new operator can be used to initialize the tile similar to the New Image operator.
The following features are supported so far:
- Automatic detection and loading of all tiles when opening the first tile (1001)
- Saving all tiles
- Adding and removing tiles
- Filling tiles with generated images
- Drawing all tiles in the Image Editor
- Viewing a tiled grid even if no image is selected
- Rendering tiled images in Eevee
- Rendering tiled images in Cycles (in SVM mode)
- Automatically skipping loading of unused tiles in Cycles
- 2D texture painting (also across tiles)
- 3D texture painting (also across tiles, only limitation: individual faces can not cross tile borders)
- Assigning custom labels to individual tiles (drawn in the Image Editor instead of the ID)
- Different resolutions between tiles
There still are some missing features that will be added later (see T72390):
- Workbench engine support
- Packing/Unpacking support
- Baking support
- Cycles OSL support
- many other Blender features that rely on images
Thanks to Brecht for the review and to all who tested the intermediate versions!
Differential Revision: https://developer.blender.org/D3509
Also adds a generic popover that can be used whenever an active tool has too
many settings than can fit in the horizontal area. The popover calls the active
tool's draw_settings with "extra" set to True.
Users should be able to know explicitly what they are testing.
By having them all enabled we run into a scenario where a new
experimental feature may have been introduced, and the user
is now using it without being even aware of that.
Differential Revision: https://developer.blender.org/D6404
The instructions for this part are not exactly clear, but better to document
what took us a while to figure out to get a new system set up for signing.
It seems it does work along multiple axes in some cases, but not at all. So
restored the UI for now, waiting for a better fix.
This reverts commit 9569425b02.
With upcoming light group passes, for them to sum up correctly to the combined
pass the clamping must be more fine grained.
This also has the advantage that if one light is particularly noisy, it does
not diminish the contribution from other lights which do not need as much
clamping.
Clamp values on existing scenes will need to be tweaked to get similar results,
there is no automatic conversion possible which would give the same results as
before.
Implemented by Lukas, with tweaks by Brecht.
Part of D4837
Part of T70240
This is the initial implementation of Weld Modifier.
New features will be added later.
ToDo:
- Seams: restrict welding to vertices along boundary edges.
- Edge Collapse: collapse edges below the length threshold.
- New icon.
- Some customdata are not being correctly interpolated.
Differential Revision: https://developer.blender.org/D6383
Viewport: Disable Clipping For EEVEE and External Renderers
Currently it is possible that, when using viewport clipping, the display and tools communicate
different information to the user then the renderer does. The reason is
that the renderer does not support viewport clipping. Both EEVEE and
Cycles do not support it.
This patch will disable the clipping in all the tools and drawing code
when the viewport drawing mode is `Material Preview` or `Rendered`.
This patch introduces a `RV3D_CLIPPING_ENABLED` util that checks if
clipping is enabled for the given `rv3d` and `v3d`. Also in places where
it was needed we added the `ViewContext` as a carrier for the `View3D`
and `RegionView3D`.
There are a few areas in the tooling (select, projection painting) that
still needs to be tackled after this patch.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6047
Scale Cage was using the same shortcut as Insert Scale Keyframe (Shift-R)
This isn't neccessary anyway since you can use the same key to cycle the tools.
Fix for T72309. The crash was caused by a missing check of the draw state, which resulted in dereferencing of a null pointer.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6380
The `DEG_depsgraph_query.h` file uses the `ITER_BEGIN` and `ITER_END`
macros defined in `BLI_iterator.h` without including that header.
No functional changes.
In the current OpenCL implementation we have a work-around for platforms
that didn't support NULL pointers. We used to replace all NULLs and
empty arrays with a pointer to a single byte on the OpenCL Device.
During investigation of {T65924} it was asked to remove this work-around
for testing. This change improves the render times.
SCENE | BEFORE | AFTER
--------------------+--------+-------
bmw27 | 108 | 89
barbershop_interior | 867 | 673
classroom | 270 | 173
fishy_cat | 244 | 196
koro | 249 | 207
pavillon_barcelona | 582 | 414
Note that this change does not fix T65924 it just improves the
rendering performance for OpenCL. We haven't tested this patch on all
platforms so we should keep an eye out on the tracker.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D6391
Updates blender subversion in order to properly handle recent userdef theme changes.
Differential Revision: https://developer.blender.org/D6388
Reviewed by Campbell Barton
This seems to only affect some drivers. Rendering to multiple targets
without all output variables defined seems to be undefined behavior.
Thanks to gaiaclary for reporting on blendercoders.
Custom render passes are added in the Shader AOVs panel in the view layer
settings, with a name and data type. In shader nodes, an AOV Output node
is then used to output either a value or color to the pass.
Arbitrary names can be used for these passes, as long as they don't conflict
with built-in passes that are enabled. The AOV Output node can be used in both
material and world shader nodes.
Implemented by Lukas, with tweaks by Brecht.
Differential Revision: https://developer.blender.org/D4837
Steps to reproduce were:
* Change File Browser display mode to fullscreen in Preferences
* File > Save As (make sure existing .blend is selected)
* File > Open
The file name would still be shown in red and the + and - icons would be
visible, which should only be the case for saving files, not opening.
Note that this change makes all `FileSelectParams.flag` values be reset
on re-opening a File Browser, which *may* in fact cause other issues.
It's easy to fix those though, and I'd prefer properly resetting the
flags and only keeping specific ones in that case.
Issue likely caused by 8b31f6fb21.
With this, initializing the toolsystem (e.g. for a new workspace-scene
combination) would skip the entire create->initialize routine for
image/texture painting settings. Reason being that these are not
allocated, unlike other paint settings. So while correctly skipping the
create part, it also skipped the initialization, which was still needed.
This does further changes in related code to avoid NULL pointer
accesses.
The issue is that `something.path_resolve('"custom_property"')` raises a
`ValueError` when the custom property is set to `None`. Since `None`
cannot be stored in a keyframe anyway, the property is now silently
skipped. Not having an explicit value is the closest we can get to
`None`. This of course breaks when the value should be `None` in between
not-`None` values, but I would consider that as a problem with the rig,
and not something Blender can fix.
To follow the standard conventions, the simplify options for grease pencil are inverted now, so to disable the file, uncheck the fill option and not as before that must be checked..
Before it was confusing because the simplify was activated checking the option and this was opposite to the mesh simplify options.
The result of normalising the bone vector wasn't checked, so it resulted
in a zero-length vector being multiplied with the desired length. Choosing
an arbitrary vector prevents the bone being 'stuck' at zero length.
Thanks @mano-wii for the patch.
The edit overlays were drawn dispite of the global hide overlays toggle.
This patch checks the global hide overlay toggle to see if it needs to draw
these overlays.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6371
This can be used to make closed surfaces behave more like a soft body.
Reviewed By: Jacques Lucke
Differential Revision: http://developer.blender.org/D5788
When size pressure is enabled, the radius may not be enough to sample
any vertices to calculate the area normal. This lead to a vector 0 plane
and a division by 0 when calculating the signed distance to the plane,
which produces a NaN that was being written to the vertex coordinates.
Reviewed By: brecht
Maniphest Tasks: T72006
Differential Revision: https://developer.blender.org/D6326
This commit fixes 3 bugs:
- Fix the crash reported in T72054. The BMesh elem table and the vd.no was null, so we now ensure that the table exists before running any sculpt tool in dyntopo. The relax function also uses vd.fno in case that vd.no is not available.
- Fix missing updates of the bounding boxes when running the mesh filter. This can be optimized by running the updates only when the filter finishes. Without this, it is impossible to sculpt the user modifies the mesh too much with the filter.
- Fix incorrect solution of relax vertex when using EEVEE. Relaxing the mesh requires the updated normals after each iteration. This was done by the PBVH rendering code, but when running EEVEE it was using incorrect normals. Now normals are updated after each iteration.
Reviewed By: brecht
Maniphest Tasks: T72054
Differential Revision: https://developer.blender.org/D6333
Some other areas in the brush code outside sculpt mode assume that
pressure is multiplied directly on top of the initial size. This patch
calculates the pixel radius correctly using the brush size from sculpt
mode to get the dyntopo detail size.
When the new brush input system is in place, all these values will come
directly from the brush input code with all the custom curves applied
per brush, so all paint modes will have a correct brush behavior and all
this sculpt mode specific code won't be necessary.
Reviewed By: jbakker
Maniphest Tasks: T72092
Differential Revision: https://developer.blender.org/D6339
The old mask brush implementation was adding the brush value to the
previous vertex mask value and clamping the result. This leads to
visible artifacts in the mask gradient as the value approaches 0 or 1,
so it was not possible to paint a smooth mask with this brush.
Now we are also multiplying by the previous mask value before clamping,
fixing all those gradient artifacts.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6341
This adds compaction support for OptiX acceleration structures, which reduces the device memory footprint in a post step after building. Depending on the scene this can reduce the amount of used device memory quite a bit and even improve performance (smaller acceleration structure improves cache usage). It's only enabled for background renders to make acceleration structure builds fast in viewport.
Also fixes a bug in the memory management for OptiX acceleration structures: These were held in a dynamic vector of 'device_memory' instances and used the mem_alloc/mem_free functions. However, those keep track of memory instances in the 'cuda_mem_map' via pointers to 'device_memory' (which works fine everywhere else since those are never copied/moved). But in the case of the vector, it may decide to reallocate at some point, which invalidates those pointers and would result in some nasty accesses to invalid memory. So it is not actually safe to move a 'device_memory' object and therefore this removes the move operator overloads again.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6369
This makes is so repeated key presses will cycle the tools in the same slot in the toolbar.
Users can tap Q repeatedly to switch selection tools, or press Ctrl-E repeatedly to enable the various Extrude tools, etc.
This is only supposed to happen when copying nodes that are part of the user
editable database, not temporary copies for the dependency graph.
The LIB_ID_COPY_LOCALIZE test was wrong because it is a combination of multiple
bitflags as pointed out by Bastien, and was actually redundant anyway since
LIB_ID_CREATE_NO_MAIN is part of it.
Previously Noise and Wave texture nodes would use noise functions within a [0,1]
range for distortion effects. We either add or subtract noise from coordinates,
never do both at same time. This led to the texture drastically shifting on the
diagonal axis of a plane / cube. This behavior makes the Distortion input hard
to control or animate. Capabilities of driving it with other texture are also
limited, diagonal shifting is very apparent.
This was fixed by offsetting the noise function to a signed range and making it
zero-centered. This way noise is uniformly added and subtracted from coordinates.
Texture pattern sticks to main coordinates which makes it way easier to control.
This change is not strictly backwards compatible, there is versioning to ensure
the scale of the distortion remains similar, but the particular pattern can be
a little different.
Differential Revision: https://developer.blender.org/D6177
Adds theme settings to allow change of front and back faces of the Face Orientation overlay
Differential Revision: https://developer.blender.org/D6262
Reviewed by Jeroen Bakker
Modes: Linear interpolation (default), stepped linear, smoothstep and smootherstep.
This also includes an additional option for the **Clamp node** to switch between **Min Max** (default) and **Range** mode.
This was needed to allow clamping when **To Max** is less than **To Min**.
Reviewed By: JacquesLucke, brecht
Differential Revision: https://developer.blender.org/D5827
When creating shaders and using maths functions it is expected that Blender should match functions in other DCC applications, game engines and shading languages such as GLSL and OSL.
This patch adds missing functions to the Blender maths node.
Ideally, it would be nice to have these functions available to vectors too but that is not part of this patch.
This patch adds the following functions trunc, snap, wrap, compare, pingpong, sign, radians, degrees, cosh, sinh, tanh, exp, smoothmin and inversesqrt.
Sign function is based on GLSL and OSL functions and returns zero when x == 0.
Differential Revision: https://developer.blender.org/D5957
Allows each File Browser list item in Volumes and System to use individual icons.
Differential Revision: https://developer.blender.org/D5802
Reviewed by Julian Eisel
Changes file names of two icon 'dat' files to reflect previous identifier rename.
Differential Revision: https://developer.blender.org/D6166
Reviewed by Julian Eisel
Implement T66304 as an experimental option,
available under the preferences "Experimental" section.
- When enabled most tools in the 3D view have a gizmo.
- Dragging outside the gizmo uses the 'fallback' tool.
- The fallback tool can be changed or disabled in the tool options
or from a pie menu (Alt-W).
Tests were missing a manifest, and were importing the
wrong version of Microsoft.Windows.Common-Controls
causing blenloader_test, bmesh_core_test and alembic_test
to fail due a loader error.
This reverts commit 3d9cc4d3f1. The commit
message and intent behind the commit is unclear. We generally want the
discrete GPU to be used for best performance, not the integrated GPU.
In older versions the ms crt was only a few dlls, in recent versions
this jumped to over 40 leading to quite a bit of clutter in our
bin folder.
This change moves the CRT into its own folder.
For developers that generally already have the runtime globaly
available on their machine, there is a new cmake option
(WITH_WINDOWS_BUNDLE_CRT, default ON) that you can use to toggle
installing the runtime to the blender bin folder, and save some
time during the initial build, this option is off by default for
only the developer profile.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6132
Multisample buffers were used for smooth line drawing. As we now have
an algorithm that doesn't need the multisample buffers we can remove
them.
The user preference for viewport multi_sampling is replaced by single
toggle overlay `use_overlay_smooth_wire`. By default this setting is
enabled as the new drawing is really quick (<1ms) and uses zero hacks.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6367
Enables Windows 10 feature that automatically adjusts non-client area (title bar) on high-DPI displays.
Differential Revision: https://developer.blender.org/D6370
Reviewed by Brecht Van Lommel
We had a manifest file, but it was seemingly not used, some
settings were done using linker pragmas, some of them visual
studio would set by default for us, others where not set at all.
This patch changes:
- Single manifest file where we can maintain all settings in
a single location, removal of any linker pragmas related to
the manifest.
- Compatibly settings for win vista - win10, without this any
call to any of the GetVersion and related functions (GetVersionEx,
VerifyVersionInfo, IsWindowsXxxx) will by default say we are
on vista and OS specific optimizations in external libraries may
be missed.
-Rather than having it in the .RC file in an #ifdef which may
or may not trigger depending on the build tool used, we tell
cmake to treat it as a source file and it will do the right
thing for both the ninja and visual studio generators.
Differential Revision: https://developer.blender.org/D6136
Reviewers: brecht
When adding a modifier to an object, the drop-down select box has the
same mouse-over tooltip for each modifier type: "Add a procedural
operation/effect to the active object". This isn't helpful when you
don't know what the modifier does from just reading the name.
This patch adds descriptions for most modifier types, so that it's
clearer what the modifiers do. The text was provided by @werwack after
discussion on DevTalk[1] and taken from the Blender Manual, and
subsequently updated by @billreynish and @HooglyBoogly.
Not all modifiers have a short one-line description in the manual, hence
those are still missing in this patch.
[1] https://devtalk.blender.org/t/ui-contextual-modifier-description-in-the-tooltips-of-the-add-modifier-dialog-window/10382
Reviewed By: billreynish
Differential Revision: https://developer.blender.org/D6352
Switch to Gregory basis patches which are tangent continuous across their
boundaries.
Originally we've used BSpline basis patches to be more compatible with the
old subdivision code, but a lot of things changed anyway.
The Camera Background Images uses a alpha under blending. For alpha
under blending to work correctly the framebuffer containing the result
of the render engine needs to be active
(`DefaultFramebufferList.default_fb`) and the source blend color needs
to be premultiplied with alpha.
Due to recent refactoring this wasn't the case and it seemed that the
background image was drawn in front of the scene. This patch sets the
correct state so it seems to be drawn behind the scene.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6365
Introduced by 9c337fcfe2 mistaken that `MeshExtract.use_threading` set
to false means that no threading is used at all. This is not the case it
will still perform threading for large objects, it will only run the
different subtasks in serial.
Fixed by scheduling the `lines_loose` in the task_pool after the rest
have been executed. This is cleaner than the previous implementation as
it sticks more to the actual design.
This use the overlay AA pass to antialias the selection outlines.
This also do all search and expand in one pass and reduce the computation
time and memory used (2 x 32bit/pixel buffer less).
Note that the aliasing is a bit worse than the old FXAA that we used to have.
Previously the decimation would take the whole curve into account when
decimating and not just the selected part.
This also contains various smaller bug fixes for the fcurve decimation.
Reviewed By: Sybren
Differential Revision: http://developer.blender.org/D6286
Bone relationship lines needs to be hidden in:
* object mode
* or when relationship lines are turned off
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6356
The lines index buffer can contain all edges (edit mode) or only loose
edges (object mode). When switching between these modes the wrong
content of the index buffer can be used.
This patch will clear the lines index buffer when a `loose_edges` is requested. Making sure it is always up to date.
Note that this is supporting an exising hack where the IBO is truncated
during the creation. We should find a different way how to solve these
kind of issues.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6349
When using Render Viewport Animation the meta data
was only stamped when a float buffer was used. In recent changes it was
possible to use a char buffer for the result.
This commit will also support stamping of metadata using a char buffer.
Add new icons and panels Grease Pencil Dopesheet to manage layers without having the properties panel visible.
Also, the icons are in the same order in Dopesheet, Layers and Material list to keep consistency.
As the number of columns for icons is limited to 3 and we also need use a factor, I have impleted the change using slider area. Also, the slider option is enabled by default for 2D Template.
See T72026 for more info.
Reviewed By: mendio, pepeland, billreynish
Differential Revision: https://developer.blender.org/D6328
Introduced a way to specify cloth pressure force influence with a vertex
group. This will allow users to only have pressure affect certain parts
of the mesh.
In addition to this, the "shrink factor" is now also unlocked to allow
negative values and thus allowing the cloth mesh to grow as well.
Reviewed By: Jaques Lucke
Differential Revision: http://developer.blender.org/D6347
Rigify uses a property group to contain options of its rigs, so
currently it is impossible to use Alt-Click or Copy To Selected
to change a setting for multiple rigs at the same time.
The main problem here is that there is no efficient way to find
which bone the property group belongs to. To maintain performance,
implement this by checking the active bone if it is known. Copy
Data Path and related features still don't work, as data path
calculation can't use context.
Differential Revision: https://developer.blender.org/D6264
The function armature_vert_task had a special case for Vertex Paints in Grease Pencil data. This patch
removes this special case. Not sure why it was needed in the first place.
Reviewed By: antoniov
Maniphest Tasks: T70378
Differential Revision: https://developer.blender.org/D6348
Now when change the setting the strokes are recalculated.
To do this, it was necessary to move the UV recalc to BKE module in order to share with Draw Engine. If the recalc it was done in draw engine, the factor was only calculated for evaluated version and there was a problem when draw a new sytroke.
Now, the RNA parameter recalc the original datablock instead of tag for be calculated in Draw Engine.
Fix for T72136: Bounding boxes were still drawn even with disabled
overlays. Also the outline displayed the original mesh
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D6350
- Margins used duplicate define between files.
- Cursor selection ignored margins.
- Cursor wasn't scaling with DPI.
Add a 'draw_rect' member which is the region rect with margins applied
to make these checks clearer.
This resolves issue pointed out in D6300,
which complicated further refactoring.
To do this we do a half pixel offset in the normal direction.
However, we use the shading vertex normal instead of the geometric normal
(for speed) and this leads to imprecisions.
So we try to mitigate the common case and leave the corner cases as is.
Instead of relying on undefined behavior to discard the line, we put the
vertex at the camera origin so that interpolation does not produce any
fragment. We don't use FLT_MAX as it also gives unpredicatble results on
NVidia.
Stupid mistake, 'original' filepath is a blender-flavored one, with
potentially weird things like the '//' relative 'header'... This can
work on linux (also it could have broken in other places too), but on
windows that is fully invalid path and python `os.path` library just
generates empty result here.
Simply using proper valid path instead fixes it...
This Simplify and should also speedup the drawing a tiny bit.
We now discard the edges in the vertex shader by setting one of the verts
at vec4(0,0,0,0) to not produce any fragment
This is the unification of all overlays into one overlay engine as described in T65347.
I went over all the code making it more future proof with less hacks and removing old / not relevent parts.
Goals / Acheivements:
- Remove internal shader usage (only drw shaders)
- Remove viewportSize and viewportSizeInv and put them in gloabl ubo
- Fixed some drawing issues: Missing probe option and Missing Alt+B clipping of some shader
- Remove old (legacy) shaders dependancy (not using view UBO).
- Less shader variation (less compilation time at first load and less patching needed for vulkan)
- removed some geom shaders when I could
- Remove static e_data (except shaders storage where it is OK)
- Clear the way to fix some anoying limitations (dithered transparency, background image compositing etc...)
- Wireframe drawing now uses the same batching capabilities as workbench & eevee (indirect drawing).
- Reduced complexity, removed ~3000 Lines of code in draw (also removed a lot of unused shader in GPU).
- Post AA to avoid complexity and cost of MSAA.
Remaining issues:
- ~~Armature edits, overlay toggles, (... others?) are not refreshing viewport after AA is complete~~
- FXAA is not the best for wires, maybe investigate SMAA
- Maybe do something more temporally stable for AA.
- ~~Paint overlays are not working with AA.~~
- ~~infront objects are difficult to select.~~
- ~~the infront wires sometimes goes through they solid counterpart (missing clear maybe?) (toggle overlays on-off when using infront+wireframe overlay in solid shading)~~
Note: I made some decision to change slightly the appearance of some objects to simplify their drawing. Namely the empty arrows end (which is now hollow/wire) and distance points of the cameras/spots being done by lines.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6296
Some points on the OpenXR-SDK dependency:
* The repository is located at https://github.com/KhronosGroup/OpenXR-SDK (Apache 2).
* We use the OpenXR loader lib from it, the headers, and some CMake utilities.
* It contains a bunch of generated files, for which the sources are in a separate repository.
* To use the injected OpenXR API-layers from the SDK (e.g. API validation layers), the SDK needs to be compiled from this other repository.
* We could use that other repo by default, but I'd rather go with the simpler solution and allow people to opt in if they want advanced dev features.
* For Windows a patch is needed to link the CRT in a compatible way.
All this is entirely untested on macOS.
Maniphest Tasks: T71365
Differential Revision: https://developer.blender.org/D6188
Changes for the higher level, more Blender specific side of the implementation.
Main additions:
* WM-XR: Layer tying Ghost-XR to the Blender specific APIs/data
* wmSurface API: drawable, non-window container (manages Ghost-OpenGL and GPU context)
* DNA/RNA for initial management of VR session settings
* Utility batch & config file for using the Oculus runtime (Windows only)
Differential Revision: https://developer.blender.org/D6193
## Design Overview
* For code using this API, the most important object is a GHOST_XrContext handle. Through it, all API functions and internal state can be accessed/modified.
* Main responsibilities of the Ghost XR-context are to manage lifetimes of the OpenXR runtime connection (represented by XrInstance), the session and to delegate operations/data to the session.
* The OpenXR related graphics code, which is OS dependent, is managed through a `GHOST_IXrGraphicsBinding` interface, that can be implemented for the different graphics libraries supported (currently OpenGL and DirectX).
* Much of this code here has to follow the OpenXR specification and is based on the OpenXR [[https://github.com/KhronosGroup/OpenXR-SDK-Source/tree/master/src/tests/hello_xr | `hello_xr`]] implentation.
* In future we may want to take some code out of the context, e.g. extension and API layer management.
* There are runtime debugging and benchmarking options (exposed through --debug-xr and --debug-xr-time, but not as part of this patch).
* Error handling is described in a section below.
## Why have this in Ghost?
Early on, I decided to do the OpenXR level access through GHOST. Main reasons:
* OpenXR requires access to low level, OS dependent graphics lib data (e.g. see [[https://www.khronos.org/registry/OpenXR/specs/0.90/man/html/openxr.html#XrGraphicsBindingOpenGLXlibKHR| XrGraphicsBindingOpenGLXlibKHR]])
* Some C++ features appeared handy (`std::vector`, RAII + exception handling, cleaner code through object methods, etc.)
* General low level nature of the OpenXR API
After all I think much of the functionality is too high level to live in GHOST however. I would like to address this by having a separate `VAMR` (virtual + augmented + mixed reality) module, placed in `intern/`. The main issue is getting this to work well with Ghost data, especially how to get the mentioned low level data out of Ghost.
This is something I'd like to look into again before too long, but for now I think having this in Ghost is reasonable.
## Error Handling Strategy
The error handling strategy I chose uses C++ exceptions, a controversial feature. Let me explain why I think this is reasonable here.
The strategy requirements were:
* If an error occurs, cleanly exit the VR session (or destroy the entire context), causing no resource leaks or side effects to the rest of Blender.
* Show a *useful* error message to the user.
* Don't impair readability of code too much with error handling.
Here's why I chose an exception based strategy:
* Most alternatives require early exiting functions. This early exiting has to be 'bubbled up' the call stack to the point that performs error handling. For safe code, early exit checks have to be performed everywhere and code gets really impaired by error checking. Tried this first and wasn't happy at all. Even if error handling is wrapped into macros.
* All `GHOST_Xr` resources are managed via RAII. So stack unwinding will cause them to be released cleanly whenever an exception is thrown.
* `GHOST_Xr` has a clear boundary (the Ghost C-API) with only a handful of public functions. That is the only place we need to have try-catch blocks at. (Generally, try-catch blocks at kinda random places are a bad code smell IMHO. Module boundaries are a valid place to put them.)
* Exceptions allow us to pass multiple bits of error information through mulitple layers of the call stack. This information can also be made specific with a useful error message. As of now, they conain a user error message, the OpenXR error code (if any), as well as the exact source code location the error was caught at.
So the strategy I went with works as follows:
* If a VR related error occurs within `GHOST_Xr`, throw an exception (`GHOST_XrException` currently).
* OpenXR calls are wrapped into a macro throwing an exception if the return value indicates an error.
* Useful debugging information and user messages are stored in the exceptions.
* All resources must be managed through RAII, so throwing an exception will release 'dangling' ones cleanly.
* In the GHOST C-API wrappers, the exceptions are caught and contained error information is forwarded to a custom error handling callback.
* The error handling callback is set in `wm_xr.c`, prior to creating the XR-Context, and implements clean destruction of the context.
Differential Revision: https://developer.blender.org/D6192
Needed for DirectX-only OpenXR runtimes (e.g. Windows Mixed Reality).
Adds a minimal DirectX 11 Ghost context, plus some shared DirectX-OpenGL resource interface using the NV_DX_interop2 WGL extension.
I know that the current implementation fails on some systems, which is something I'll have to fix at some point. Don't know if this is a showstopper though, I can fix that in master too. For now this isn't going to be used by many people anyway. Recently I also learned that OSVR uses the same extension, see https://github.com/sensics/OSVR-RenderManager/blob/master/osvr/RenderKit/RenderManagerD3DOpenGL.cpp. Their implementation may be useful to fix the issue, according to a OSVR dev, it works quite reliably for them.
Note: Didn't actually test just this patch on Windows yet.
Differential Revision: https://developer.blender.org/D6190
Some points on the OpenXR-SDK dependency:
* The repository is located at https://github.com/KhronosGroup/OpenXR-SDK (Apache 2).
* We use the OpenXR loader lib from it, the headers, and some CMake utilities.
* It contains a bunch of generated files, for which the sources are in a separate repository.
* To use the injected OpenXR API-layers from the SDK (e.g. API validation layers), the SDK needs to be compiled from this other repository.
* We could use that other repo by default, but I'd rather go with the simpler solution and allow people to opt in if they want advanced dev features.
* I copied `presentation.cmake` and `xr_platform_defines.cmake` from the SDK. They contain logic that is not needed for us and prints at CMake generation. We could change that but figured it would also be helpful to keep the files equal to the SDK ones.
* For Windows a patch is needed to link the CRT in a compatible way.
* @LazyDodo already pushed the precompiled binaries for Windows.
All this is entirely untested on macOS.
Differential Revision: https://developer.blender.org/D6188
Just like the Windows Mixed Reality runtime, we have to apply a sRGB
OETF to get visually correct-ish looking colors.
We plan to further investigate if it's really WMR and Monado being wrong
here, or if it's Oculus (which doesn't need this additional transform),
but there are some details to check.
Reduces code duplication for the DirectX specific upside down drawing.
Didn't actually test the upside down drawing, since I need to do that on
Windows.
* Adds the needed bits to support VR session settings and access them in
the VR view drawing code.
* Added settings for: shading mode, grid floor, annotations, clipping.
More can easily be added.
* The Add-on adds a "VR" tab in the side bar, containing buttons for the
added settings
This fixes linking errors with the Monado runtime. The specification
says that extension functions have to be gotten through
`xrGetInstanceProcAddr()` which we did for some extensions, but not for
the graphics extensions. Worked fine earlier, but broke meanwhile.
Code now assumes that the view-draw callback left the OpenGL state for
ongoing use by Ghost-XR. So the state doesn't have to be set by the
swapchain image submission and thus it we don't need to pass the Ghost
OpenGL context to it in a hacky way.
Also, rename drawViewEnd (to submitToSwapchain) and remove
drawViewBegin.
These were previously used to blit from a offscreen (non-OpenGL)
context, into an onscreen one. We do this differently, and only within
GHOST_XrGraphicsBinding now, so this can be removed.
Rather than max. one shared resource per DirectX context, code can now
request a handle for a shared resource and ask the DirectX context to do
operations on that.
There were a few typos here and there, and the openxr_sdk does not respect the cflags we give it hence a patch was added to work around this undesirable behaviour.
Adds the SDK so that `make deps` works, on all platforms. On Windows,
the SDK headers and libraries are now assumed to be in the usual lib\
directory.
Also fixes a mistake in install_deps.sh
Note that none of this is tested as it requires an older Visual Studio
version than I have.
Removes the OPENXR_USE_BUNDLED_SRC option which allowed using the
OpenXR-SDK sources bundled in extern/. It would be too much hassle to
keept these updated.
Now these have to be provided on the system Blender is compiled on. I've
already added necessary bits to install_deps.sh and our Windows
maintainer is ready to provide builds with OpenXR-SDK linked.
Once this gets into master, platform maintainers will have to be
notified about the added dependency.
This also removes the JsonCpp dependency, it was only needed for the
OpenXR-SDK sources.
Discussed this in length with @sobotka, and it seems WMR has an utterly
broken pixel color pipeline. So we apply a SRGB OETF for this specific
runtime to compensate.
Only tested on Windows.
Updates loader sources from the OpenXR SDK to latest 1.0 SDK and updates
requiremed version to 1.0. The compile time generation of files is a
thing of the past now (although you can still force it).
1.0 got released yesterday. Only one line needed fixing in our OpenXR
code to get it to compile. Rendering is black though.
For now I tried to keep edits to CMakeList.txt files minimal. So now
there are OpenXR CMake options exposed (with bad names), CMake prints,
etc.
This way we avoid the big overhead of context switches. Makes frames
render about twice as fast here. For heavy Spring scenes I'm getting
around 20 FPS here, classroom scene is at 50 FPS.
This is great given that drawing itself still isn't optimized for dual
eye rendering.
Outputs frame render time in milliseconds and FPS this time would add up
to. We could average times so FPS is a bit more stable, but the
precision of un-averaged results may be helpful too.
I'm not really sure why the leak happened - draw manager kept allocating
certain buffers - but I figured I could avoid it by not recreating
GPU_offscreen/GPU_viewport on every redraw.
This should also improve performance a bit.
Adds generic unique_oxr_ptr to wrap xrCreate and xrDestroy functions of
OpenXR handles into a unique_ptr like RAII interface.
While for most cases, OpenXR resources can be freed by their owning
object, sometimes errors may occor before final ownership is established.
E.g. swapchain ownership is only transfered to the session object once
its swapchain-images are created - which may fail. With this RAII
wrapper, the swapchain would be freed on error (as this triggers stack
unwinding through an exception), no matter who holds ownership to it
currently.
There are other solutions to this problem, e.g. by establishing final
ownership right after/upon creation, or by explicit freeing in case an
error is spotted; it's too easy to make mistakes here though. Plus, we
may want to experiment with using this API for all OpenXR resources, to
entirely avoid the possibility of them leaking.
Fixes:
* Destruct surface when destroying session on error. Fixes null pointer
dereference when trying to draw the surface on next redraw.
* Fix trying to enable same extensions/API-layers multiple times due to
static array usage not being cleared after error.
* Null pointer dereference with OpenGL drawing
E.g. with an active OpenXR runtime installed, but no HMD plugged in,
Blender will now show: "Failed to get device information. Is a device
plugged in?".
In case of such errors, the VR-session will be cleanly exited, with
no side effects to the rest of Blender (at least if there are no bugs).
The GHOST_Xr API now allows setting a custom error handling callback
which may cleanly destroy all GHOST_Xr data.
This wraps all functions that could fail into some proper (although
unfinished) error handling.
These were the requirements for the error handling strategy:
* If an error occurs, cleanly exit the VR session (or the context if the
error happend during its set up), causing no resource leaks or side
effects to the rest of Blender.
* Show a *useful* error message to the user.
* Don't impair readability of code too much with error handling.
After some back and forth I decided Ghost internal exceptions are the
best way to go about this. I get exceptions are a controversial feature,
IMHO that's because most people use them 'wrong' however. Here's the
rationale:
* Most alternatives require early exiting functions. This early exiting
has to be 'bubbled up' the call stack to the point that performs error
handling. So the code gets really impaired by error checking. Tried
this first and wasn't happy with it at all. Even if error handling is
wrapped into macros.
* All GHOST_Xr resources are managed via RAII. So stack unwinding will
cause them to be released cleanly whenever an exception is thrown.
* GHOST_Xr has a clear boundary (the Ghost C-API) with only a handful of
public functions. That is the only place we need to have try-catch
blocks at.
(Generally, try-catch blocks at kinda random places are a bad code
smell IMHO. Module boundaries are a valid place to put them.)
* Exceptions allow us to pass multiple bits of error information through
mulitple layers of the call stack. This information can also be made
specific with a useful error message.
As of now, they conain a user error message, the OpenXR error code (if
any), as well as the exact source code location the error was caught
at.
So if an error is caught inside GHOST_Xr code, an exception is thrown
with specific and hopefully useful information in it. In the Ghost C-API,
these exceptions are caught and passed on to a custom error handling
callback. This callback will be defined by the Blender window-manager
and output the error information via a usual user report popup (not done
yet). It can also ensure the entire session is shut down.
Note that the majority of errors OpenXR can return are for invalid API
usage. Assuming the API usage is valid though, most error messages
should never reach users and can be left a bit more vague. Maybe we can
add something like "This is probably a bug and should be reported" to
those.
Unfortunately, enabling XR_EXT_debug_utils crashes instance creation
with the Windows Mixed Reality runtime. Maybe I'm doing something wrong,
for now, just disable it.
* Initialize all class member variables
* Add version to runtime name printing
* Separate functionality code from debug prints
* Improve code structure using Doxygen groups
* Make accessors const functions
* Add (Doxygen) comments
* Naming
* Reorder functions
Makes GHOST_Xr much more consistent with the rest of GHOST. Basically I
added a GHOST_IXrContext interface which can be called by the GHOST
C-API. The internal GHOST_XrContext class implements this.
Outside of GHOST only the opaque GHOST_XrContextHandle is accessible.
Kept GHOST_Xr_intern.h and GHOST_Xr.cpp for now. I'll probably remove
them soon. There's not much reason for both of them to be there.
Enables (or tries to) the XR_EXT_debug_utils extension which allows
setting a custom callback for additional debug prints. Note that I
haven't been able to test this really, as the Monado runtime appears to
not have this fully implemented (had to force sending a custom message
to find that out...). Will test with the Windows MR runtime in a bit. We
can also improve message quality then (it can print the exact function
name the message was sent from, print additional custom labels which
could indicate session state, etc.).
For this to work two environment variables have to be set:
XR_API_LAYER_PATH and LD_LIBRARY_PATH, both have to point to the API
layers built by the OpenXR SDK. It also requires you compile Blender
linked to your own SDK build (OPENXR_USE_BUNDLED_SRC disabled).
Further changes will make this set up unnecessary, so validation layers
can be enabled via some flag.
I was trying to set the camera pose as OpenXR reference pose. But then I
got really strange poses for drawing back from OpenXR. Couldn't figure
out a way to solve this.
Now we just apply OpenXR's draw pose to the Blender camera pose. If we
later want to support moving around in the scene (e.g. via teleporting),
that has to be changed.
Also uses flipped drawing for DirectX surface to correct DirectX upside
down drawing (compared to OpenGL).
Session code and data structures are now localized. But also, this
enables sessions to use deterministic destruction to end itself cleanly.
So whenever an error occurs, we can use stack unwinding which will cause
graceful ending of the session.
* Move GHOST_Xr types to GHOST_Types.h
* Add GHOST_XrPose. We'll have to pass around pose data at multiple
places.
* Don't require extra call to prepare session rendering, handle
everything through GHOST_XrSessionStart
* Naming
Finally: This makes it possible to render a viewport to an HMD via
OpenXR. Pure OpenGL rendering will need some more tweaks to work.
To my great delight, performance is quite good for reasonably sized
scenes.
Had to do some hacks and marked some TODOs. Nothing too bad though.
Here are a couple of notes:
* Current initial pose is pretty useless, think it just looks downwards
from world origin. Will change that soon.
* The rendered viewport has some issues: Too dark (bad lighting?), grid
doesn't show up even though I told it to, lighting seems to change with
view position/rotation, etc. Needs some polish.
* Ideally we'd just use the D3D11 Texture given to us via the OpenXR
swapchain and blit the OpenGL framebuffer into that. However the
NV_DX_interop extension fails doing this. Seems like this is a NVidia
Optimus only issue, but I'm missing the hardware to confirm.
So instead, we blit into the D3D11 back buffer first and then into the
Texture.
* The draw-manager uses its own offscreen context so we have to get the
render result from the draw-manager context to the VR session's
context first. Luckily I've already added code to support blitting from
one OpenGL context into another. But it requires blitting twice.
Blitting should be very cheap, but still...
Draw-manager could get a context to use passed instead.
Not visible yet, but it should draw in the offscreen. The way this is
now, we don't depend on the Window->Workspace->bScreen->... chain. We
simply draw an offscreen viewport in the draw callback of the XR session
surface.
The drawing also uses view and projection matrices from OpenXR (or
calculated from OpenXR data).
Uses the new wmSurface type (non-window drawable container) to manage
the OpenGL, DirectX and GPU module contexts. The draw callback of the XR
surface calls the GHOST_Xr session drawing routines.
What you should see when starting a VR session now (using the WMR
runtime): The Windows Mixed Reality Portal pops up, and a blue
background is drawn on the HMD. This is from the blue color clear call
we do in the drawing preparations of the GHOST_Xr session drawing.
Adds a wmSurface type which acts as a container for non-Window (offscreen)
draw surfaces. Ideally wmWindow would of course also just do C-style
inheritance from wmSurface, but I guess they can co-exist too.
For the VR session a surface is created on Linux and passed to the
graphics binding to use.
Note this is not used on Windows yet, it still opens a window there.
* Allow passing custom data to session draw function, passed to the
callback
* Actually call the callback
* Create and bind a WM level callback. Will later be used to draw the
viewport.
Also, check if session is actually visible before drawing.
Just a clear call for now, so all you see is a blue world. This blue
"world" is however drawn by Blender!
Also fixes use after destruction of compositor layer data.
Using a dummy identity pose at {0, 0, 0} to start with. This should
later use the current viewport position & orientation I guess.
Also adds function to set a draw callback. There's quite some OpenXR
related stuff to be done before and after drawing anyting, as well as
before and after drawing each view (eye). Quite some info would have to
be exposed to WM to let it manage drawing. So I think using a callback
called from GHOST_Xr to draw each eye instead is a good way to go.
VR session currently crashes after opening. Seems to be related to
blocking frame wait call. I'm not too worried about that though, might
disappear once OpenXR frame sync functions get proper timing info
passed.
Following the OpenXR SDK's example code very closely here. We have to do
some nasty converting of graphics binding specific image vectors to
generalized base ones. The SDK's approach seems like a good way to go
about this.
With this, the Windows Mixed Reality Portal finally pops up when
starting the session. That is how it's supposed to work. After it's
initialization phase all you see is black. That's expected too as we
don't send anything to the device yet.
Nothing special to say. Just calls needed OpenXR functions to create
swapchains. Swapchain images are not created yet.
Interestingly, the Windows Mixed Realtiy portal now pops up when closing
Blender after having created a VR session. So we're getting closer ;)
Monado now opens a window here when asking it to start a session. That
seems to be the case because it doesn't detect the HMD as direct mode
capable yet. But that shouldn't be an issue from our side.
Refactors function into a class, and make this class a friend of
GHOST_ContextGLX. That seems like a better way to access low level
graphics data for this specific case, rather than giving anyone access
via a getter.
Creating the context causes the OpenXR loader to try connect to a
runtime. That would involve reading the OS'es active_runtime.json and
dynamic linking based on that. So better avoid doing this on startup.
Also: don't pay for what you don't use!
DirectX is Y coordinates top to bottom, while OpenGL is the opposite.
For the final window manager on-screen drawing, draw to a texture first
and draw that upside down if needed.
Phew! That a fight. But this is also a pretty important feature as it
allows interfacing with Windows Mixed Reality OpenXR runtime and the
HMDs supported by it.
Important remaining issue: The rendered viewport is upside down :) That
is of course because DirectX has the opposite vertical direction than
OpenGL.
When creating a DirectX window, we also create an OpenGL offscreen
context to use for all drawing. Just before swapping framebuffers, the
DirectX window blits the framebuffer of the offscreen context into its
swapchain. This requires the WGL_NV_DX_interop and WGL_NV_DX_interop2
extensions.
For testing/dev purposes, also adds:
* Version of the offscreen to onscreen blitting that's OpenGL only. So
it blits the default framebuffer of the offscreen context into the one
of the onscreen context.
This is disabled by a #define.
* Code to draw a colored triangle using DirectX, also for testing.
Requires the D3DCompiler.lib to be available at compile time. This is
also disabled using a #define.
With this the VR window should open fine and get cleared in a red-ish
orange using Direct3D 11 calls (well, on Windows that is).
The window still draws a 3D view to an offscreen buffer. Where we
usually just swap the buffers, we now allow calling a GHOST function to
blit the offscreen OpenGL buffer to whatever type of graphics buffer the
window uses (DirectX here).
The nice thing about this approach is that all DirectX code stays in
GHOST_ContextD3D.cpp. And the entire compatibiliy code can go into a
single function higher level modules don't need to care about.
This also fixes a number of issues introduced in earlier commits.
The window doesn't show anything of course. However we draw (at least I
assume it does) as regular, just into a window offscreen context.
A valid 3D view is created in the window. It's not visible but you see
cursor changes as you move over the window. So handling works.
The window immediately crashes, hence keeping it disabled for now.
Not sure how much of this I'll leave in, for now this is mainly for
testing DirectX compatibility.
* Retrieve graphics context to bind through callbacks so the XR code can
manage their lifetime.
* Use static union to store system & chosen graphics lib specific
graphics binding data.
Makes some things a bit cleaner, too.
Adds support for creating a DirectX 11 Ghost context. It's not used yet
and more stuff besides creation is needed. Also, other versions than 11
should probably be supported.
We need a DirectX context to support the Windows Mixed Reality OpenXR
Runtime, a rather important one to support. Idea is to use an extension
for OpenGL-DirectX interoperability for drawing the OpenGL offscreen
viewport render using DirectX.
Similar to gizmo/ and message_bus/, there's now a xr/ directory
containing header files and a intern/ directory for source and internal
header files.
Guess this is reasonable to do. And better to do early on to avoid
loosing much git history.
OpenXR needs to interface with some graphics library (OpenGL, Vulkan,
DirectX, etc.). This is done through graphics binding extensions. The
OpenXR specification requires these to be properly set up before a
session is created.
Adds the following:
* Support priority list of multiple graphics binding extensions (e.g.
check OpenGL extension availability first, DirectX on Windows second,
etc.)
* Barebones for passing graphics library data to OpenXR session
creation. This is highly system dependent, e.g. it requires GLX data
for OpenGL on X11 systems (XrGraphicsBindingOpenGLXlibKHR). More work,
including additions to GHOST, will be needed once I get to the more
graphics related stuff.
* Create an own graphics context for the VR session. It's not doing
anything useful yet. This is just to fool the Monado OpenXR runtime
so that it actually attempts to create the OpenXR session.
* Had to add two CMake modules for platform dependent #define's required
by the OpenXR specification.
To correctly start a session, a graphics extension specific object needs
to be passed to the OpenXR runtime. E.g. for the Windows Mixed Reality
runtime, XrGraphicsBindingD3D11KHR needs to be passed with a valid
DirectX device. Since we don't have any DirectX compatibility working,
I can't test this on Windows yet.
So to test this I finally need to get Monado to work on Linux and
correctly setup the OpenGL extension there.
Adds operator to toggle a VR session, exposed in the Window top-bar
menu. It triggers the needed calls for session creation and destruction.
Setting up the XR-system (a configuration of related devices) is also
done now.
Calling WMR runtime functions fails currently. Not sure why. So while
this executes required routines, it doesn't really work.
We don't actually enable any extension or layer yet. We just put their
names into arrays and print them. The printing can be disabled via a
compiler define, later we should put them behind a proper
logging/debugging mechanism (CLOG).
Also fixes a Visual Studio compile error.
Adds wm_xr.c for an XR management API and creates/destroys the OpenXR
instance through this. This is as planned in my proposal, to lock the
OpenXR calls behind an abstraction.
XR data will be stored in a (non-public) wmXRContext struct within the
window-manager.
For now, creates the OpenXR instance on startup. I think it's better to
lazy setup this, as in, only creating the instance once the user starts
the first XR session. Just to avoid costs for something that may not be
used (the OpenXR loader we use will try loading and parsing the system's
OpenXR active_runtime.json on instance creation). That's for later when
I introduce session management though.
Also added a context getter for the xr-context, which is unused but may
be handy later.
Adds OPENXR_USE_BUNDLED_SRC so that if disabled, CMake tries to find
the SDK headers and libraries in system paths or in specified root
directory.
I guess this is the way we'd want to do this in master. However for
people testing the branch the bundled sources are much more convenient
(should work out of the box, no need to compile the SDK manually).
Calling OpenXR functions should now work on both Windows and Linux.
Also includes:
* Silence Linux only warnings
* Remove common_cmake_config.h and let CMake generate it when
configuring (contents vary on system it's compiled on)
* Remove JsonCpp CMake install target
* Remove unnecessary CMakeLists for JsonCpp includes
* Remove unnecessary CMake commands
* Style cleanup
Adds needed headers for OpenXR, the loader from the OpenXR SDK and
JsonCpp into extern. Took a while to get this to compile/link, but on
Win10 it works fine now without patching #includes. Linux probably
needs more work.
Added a compile option WITH_OPENXR to toggle XR feature compiling.
Also does a dummy xrCreateInstance() call to test linking.
2019-05-28 01:21:25 +02:00
3424 changed files with 543703 additions and 444144 deletions
# This may have issues with C++ initialization order, needs to be tested
# on all platforms to be sure this is safe to enable.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
