Affected by rB8621fdb10dc4
Crash if single-user data is created when we apply transform
on multi-user image data. Crash occurs because creation of new copy
was not handled in `single_obdata_users` for empty objects (image for example)
Reviewed By: dfelinto, mont29
Maniphest Tasks: T100040
Differential Revision: https://developer.blender.org/D15587
The Python based importer had a special case handling of "no faces in
the whole file at all", where it ended up treating the whole file
as essentially a point-cloud-like object (just loose vertices, no
faces or edges). The new importer code was missing this special case.
Fixes T100017. Added gtest coverage that was failing without the fix.
# Conflicts:
# source/blender/io/wavefront_obj/importer/obj_import_file_reader.cc
The importer code was written under incorrect assumption that vertex
data (v, vn, vt commands etc.) are grouped by object, i.e. follow the
o command, and that each object has its own vertex data commands. This
is not the case -- all the vertex data in the whole OBJ file is
"global", with no relation to any objects/groups; it's just that the
faces belong to the object, and then they pull in any vertices they
like.
This patch fixes this incorrect assumption in the importer:
- Vertex data is now properly global; no need to track some sort of
"offsets" per object like it was doing before.
- For each object, face definitions track the minimum & maximum vertex
indices referenced by the object, and then all that vertex range is
created in the final Blender object. Note: it might be (unusual, but
possible) that an object does not reference a sequential range of
vertices, e.g. just a single face with vertex indices 1, 10, 100 --
the resulting Blender mesh will have all the 100 vertices (some
"loose" without belonging to a face). It should be possible to track
the used vertices exactly (e.g. with a vector set), but I haven't
done that for performance reasons.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D15410
# Conflicts:
# source/blender/io/wavefront_obj/importer/obj_import_file_reader.cc
# source/blender/io/wavefront_obj/importer/obj_import_mesh.cc
# source/blender/io/wavefront_obj/importer/obj_import_objects.hh
# source/blender/io/wavefront_obj/tests/obj_importer_tests.cc
Address the issue by re-working line continuation handling: stop
trying to parse sequences like "backslash, newline" (which is the
bug: it should also handle "backslash, possible whitespace, newline")
during parsing. Instead, fixup line continuations after reading chunks
of input file data - turn backslash and the following newline into
spaces. The rest of parsing code does not have to be aware of them
at all then.
Makes the file attached to T99536 load correctly now.
# Conflicts:
# source/blender/io/common/intern/string_utils_test.cc
# source/blender/io/wavefront_obj/importer/obj_import_string_utils.cc
# source/blender/io/wavefront_obj/importer/obj_import_string_utils.hh
The stroke points were changed but the bounding box calculation was not done and this produced a problem in any bounding box check done by different tools.
Any script that raised a SystemExit called by --python, --python-expr
command line args or by executing the text block would exit without
printing a message. This caused the error from T99966 to be hidden.
Add explicit handling for SystemExit to ensure the message is always
shown before exiting.
More details noted in code-comments.
Was accidental regression in rBed9b21098dd27bf9364397357f89b4c2648f40c2
Remove the input slider's PROP_FACTOR subtype in favor of the default to
align with other IOR sliders. This provides much better control when
dragging the value with the mouse.
Differential Revision: https://developer.blender.org/D15477
It was smart enough to check if the buffer had the right
size but neglected to cast to a 64 bit value so it
overflowed.
Differential Revision: https://developer.blender.org/D15457
Reviewed By: brecht
Specifically BKE_texpaint_slots_refresh_object was being called, which
causes cycles to reset at strange times (like moving the mouse cursor
in pose, boundary and various other tools).
This (along with some code that checks if the pbvh pixels need
to be rebuilt) is only run if is_paint_mode (which used to be
needs_colors) is true.
Note there is a bug in BKE_object_is_visible_in_viewport, it
returns false when the object is in local mode.
The transform operator poll should do a similar test. That
would allow us to move the test from sculpt_brush_strok_invoke
to SCULPT_mode_poll (at the moment we cannot do this due to
the brush operator falling through to the translate keymap
item in global view3d keymap).
The falloff was applied to scale by error. Now, the falloff is only applied to the rotation.
Differential Revision: https://developer.blender.org/D15364
.
Regression in [0] which didn't account for the bounds of empty objects.
Add support support calculating bounds from empty draw-type to use in
pose-bone culling.
[0]: 3267c91b4d
Caused by [0] which made accessing the drag-start require a function
instead of being the value written into the event cursor coordinates.
[0]: b8960267dd
Regression in [0] caused operations such as file-load or file-new
from any window besides the first to write into the freed:
`wmWindow.eventstate`.
Resolve by copying the event instead of restoring the region relative
cursor position after modifying it.
[0]: 789b1617f7
This changes the boolean modifier material index handling to be
consistent with the mesh boolean geometry nodes, which was
last changed in 1a6d0ec71c. The issues was that the
material maps were retrieved at the object level, which doesn't
really make sense because the boolean is a geometry-level
operation. It was also confusing and prone to incorrect behavior
because it's more complex to retrieve information from two places.
Differential Revision: https://developer.blender.org/D15365
Baking assumed that color attributes could only have two configurations:
float color data type on vertices, or byte color type on face corners.
In reality the options can be combined to make four total options.
This commit handles the four cases explicitly with a somewhat
more scaleable approach (though this should really be C++ code).
Differential Revision: https://developer.blender.org/D15244
Optionally use `sphinx_changelog_gen.py` to dump current version of the
API in a JSON file, and use closest previous one listed in given index
file to create a changelog RST page for Sphinx.
Part of {T97663}.
Main change is to make it use JSON format for its dump files, instead of
some Python code.
It also introduces an index for those API dump files, mapping a blender
version to the relevant file path.
This is then used to automatically the most recent (version-number wise)
previous API dump to compare against current one, when generating the
change log RST file.
Part of {T97663}.
* TBB MEX version is now 2021, since this versin introduces 'oneTBB'
which brings a lot of incompatibilities with previous versions.
* Fix several typos and mistakes in OSD, Embree and OIDN build code that
prevented proper usage of a local TBB build.
There were two problems:
1) The checking of the collision was not working with one point only.
2) For one point, the masking was checked always and if the masking was not activated, the stroke was skipped.
The resource binding were missing from the shading group
(`shgroup->uniform_attrs`), leading to no custom property UBO creation
(`drw_uniform_attrs_pool_update`) when issuing the drawcall,
resulting in a missing UBO bind.
The fix make sure to no duplicate the bindings by creating a simple
shader bind instead of a `GPUMaterial` bind.
Candidate for 3.2.1 corrective release.
Using Light output is supported in Cycles. This patch adds support for it
and remove the crash in `ntree_shader_weight_tree_invert()` by treating
it as any other outputs.
Candidate for 3.2.1 corrective release.
This happened because of the false assumption that `std::array<char, 32>`
would be treated as a container and not relocate their content if the
`Vector` would grow. Replacing with actual object allocation fixes the
issue.
Candidate for 3.2.1 corrective release.
In perspective mode the snap point direction needs to be taken into
account to define which side of the face is being looked at.
If there is no face under the mouse cursor, there is no direction
adjustment and the element normal will be used.
Addendum to previous fix, which was for point selection, this fixes the
face selection mode. The issue is caused by wrong flags used for paint
mode (the edit mode flag was always used). Also add back flag which was
accidentally removed in 16f5d51109.
The normals flags were not setup properly which made normals for all
elements (vertices, faces) to be drawn when using the normals overlay.
Also remove usage of uints for the flag in the APIs.
Before [0] mouse-motion events left the 'event.value' un-changed,
so a mouse-move would be set to PRESS/RELEASE based on previous events.
Support accessing the previous event value directly
to address feedback from T99102.
Note that the previous cursor location is already exposed.
[0]: 52af3b20d4
Some OBJ files out there (see T98782) have face definitions that
contain vertex normal indices, but the files themselves don't
contain any vertex normals. The code was doing a "hey, that's an
invalid index" and skipping these faces. But the old python importer
was silently ignoring these normal indices, so do the same here.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D15177
All of the operators in vertex paint mode didn't work properly with
the new color attribute system. They only worked on byte color type
attributes on the face corner domain.
Since there are four possible combinations of domains and types now,
it mostly ended up being simpler to convert the code to C++ and use
the geometry component API for retrieving attributes, interpolating
between domains, etc. The code changes ended up being fairly large,
but the result should be simpler now.
Differential Revision: https://developer.blender.org/D15261
It was looking up the last modifier in the stack, ignoring visibility, instead
of mesh->runtime.subsurf_runtime_data set by the modifier evaluation and used by
the drawing code.
One case of copying image formats was not properly using BKE_image_format_copy.
To fix this for existing .blend file we need to do versioning, ensuring the curve
mapping is properly copied.
Commit 277fa2f441 added channels region to unintended editors if sequencer was
used in area. This caused issues with some editors having 2 tool regions and
non functioning side panel.
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D15253
When input file is changed, `orig_height` and `orig_width` fields are
reset, which causes thumbnail dimensions to be incorrectly calculated.
Only draw thumbnails if both mentioned fields are non 0.
Fix by always testing unhandled double-click events as press events,
irrespective of the previous event type.
**Details**
Handling double-click events only ran when the previously pressed-key
matched the current pressed-key.
Originally when double-click support was added the previous type was
compared (ignoring it's press/release value) and while not necessary
it was harmless as it matched the check for double-click events being
generated.
As of [0] the logic for click/drag detection changed to ignore release
events as release this could interrupt dragging.
This made it possible to generate double-click events that failed the
`event->prev_press_type == event->type` comparison.
In these cases it was possible to generate double-click
events that would not fall-back to a 'press' value when unhandled.
[0]: 102644cb8c
When GPU subdivision is used, and the modifier is not set to be applied
on the cage, UV selection is not synced with the face selection in the
viewport.
This happens because the extraction, despite being in edit mode, is set
to `MESH` instead of `BMESH` (or `MAPPED` in some cases) like for CPU
subdivision, and since the mesh is not always synchrnised with the BMesh
the edit mode flags are not always updated.
With GPU subdivision, when creating the `MeshRenderData`, the condition
`has_mdata && do_final && editmesh_eval_final != editmesh_eval_cage` is
true which forces the `MESH` extraction. Following comment in D14485,
this replace the `has_mdata` in the condition with `use_mapped` which
solves the issue.
Differential Revision: https://developer.blender.org/D15248
The old Python OBJ importer had a (somewhat confusingly named) "Keep
Vertex Order -> Poly Groups" option, that imported OBJ groups as
"vertex groups" on the resulting mesh. All vertices of any face were
assigned the vertex group, with a 1.0 weight.
The new C++ importer did not have this option. It was trying to do
something with vertex groups, but failing to actually achieve
anything :) -- the vertex groups were created on the wrong object
(later on overwritten by "nomain mesh to main mesh" operation);
vertex weights were set to 1.0/vertex_count, and each vertex was only
set to be in one group, even when it belongs to multiple faces from
different groups. End result was that to the user, vertex groups were
not visible/present at all (see T98874).
This patch adds the import option (named "Vertex Groups"), which is
off by default, and fixes the import code logic to actually do the
right thing. Tested on file from T98874; vertex groups are imported
just like with the Python importer.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D15200
Eevee rendered an empty image for aov nodes that weren't linked to
any other nodes. When connected the result was OK. The root cause was
that the AOV nodes were not marked as output node and pruned when not
connected to any other nodes. The pruning process is there to reduce
the complexity of the GLSL and improve compilation time and
execution time.
Regression introduced by {rBca37654b6327}. This commit reversed the
order of loading uniforms. The bloom renderpass used the previous
loading order to overwrite an existing uniform (bloomBaseAdd).
Due to the new ordering this doesn't work anymore where the render
pass outputted an image similar to the final image. This was fixed
by loading the correct value for bloomAddBase and remove the rewrite.
To match the existing Python .obj importer, and to make it easier for
the user to determine which object is which, use the filename for the
default object name instead of "New object".
Differential Revision: https://developer.blender.org/D15133
Since the occlusion input is going to be removed in EEVEE-Next, I just
added a temporary workaround. The occlusion is passed as SSS radius
as the Specular BSDF does not use it.
The final result matches 3.1 release
Issue is caused by an off by one error which would map some edge loops to
the loops of some the next polygon in the list of polygon, which may not
be a topological neighbor.
On MacOS Eevee cyptomatte shaders fails as it doesn't ignore the `attrib_load`
parameter. I validated that removind the parameter works on Linux/AMD and MacOS
Intel. It could be that there are other platforms that require the dummy parameter.
If this should use a forward declaration and implement an emoty function in the
cryptomatte vertex shader.
The call to `get_component_for_write` would sometimes copy the mesh
even when the mesh is replaced with itself. The `replace_mesh` method
handles that case already, so just use that instead.
Intel iGPU (HD4000) supports OpenGL 4.4 but doesn't support conservative
depth. (GL_ARB_conservative_depth). This change will only check for the
availability of the extension.
There were two problems here:
1) Console warnings due to brush was None.
2) It was impossible to recreate a brush.
This patch fixes both issues and it is now possible to recreate any brush.
Differential Revision: https://developer.blender.org/D15213
Reviewed by: @dflelinto
NURBS curves can be invalid when the order is less than the number
of points, or in a few other situations. Currently the evaluated data of
an invalid NURBS curve is empty. This is inconvenient because it
requires checking for empty curves when it otherwise wouldn't be
necessary. This patch replaces that fallback with copying the original
data to the evaluated points. This makes conceptual sense too, as if
the curve couldn't be evaluated-- which wouldn't necessarily delete it.
Usually the UI protects against this happening, but it's currently
possible to create an invalid curve with some operations like the
delete geometry node.
Differential Revision: https://developer.blender.org/D14837
Whats happening is that the modifier keeps adding new frames to the evaluated object resulting in an exponential increase. This is because when preparing the data for the modifiers we only copy visible strokes to the eval object. But the modifiers do not consider visibility and will generate the mirrored strokes even for layers that are hidden. Because those layers have not been copied (only their structure) we run into this issue.
The solution is always copy the active frame of all layers (even if the layer is hidden).
The new OBJ importer is producing "sharp" edges on some meshes that
should be completely smooth. Only observed on UV-Sphere type meshes
so far (see T97820).
I'm not 100% sure what is the root cause, but my theory was that
maybe due to limited number of float digits that are printed for
vertex normals in the file, the normals that are read in are not
always exactly 1.0 length. And then the Blender's "set custom loop
normals" function (which expects normalized inputs) wrongly marks
some edges as sharp.
Adding explicit normalization for the normals that are read from the
file fixes the wrongly-sharp edges in test cases from T97820. I
have not observed measurable performance impact in importing large
models (e.g. 6-level subdivided Monkey) that contain vertex normals.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D15202
The importer was not doing a notification that the scene has changed, so
the bottom status bar scene stats info was not updated right after the
new OBJ import.
Reviewed By: Julian Eisel
Differential Revision: https://developer.blender.org/D15015
Re-organize de-selection so that there is less conflict between tracking
and masking operators.
Still not fully ideal: the LMB selection does not de-select everything
now since the `mask.select` with `deselect_all` is only added to the
keymap when the RMB is the select mouse. While this is sub-optimal, this
seems to be how mask selection behaved in the Image Editor in 3.1.
Not sure it worth looking into a more complete fix, as it will likely be
too big to be safe for a corrective release.
Differential Revision: https://developer.blender.org/D15183
The comparison between dot products of each point of the poly were
not taking into consideration negative values. FLT_MIN was used rather
than -FLT_MAX due to a misunderstanding of the FLT_MIN definition.
Maniphest Tasks: T98718
Differential Revision: https://developer.blender.org/D15161
When searching for the active NLA strip, avoid overwriting the found strip
pointer with NULL if it was already found in a previous iteration.
The active strip is searched for while looping over the NLA tracks. If the
active strip was found on a previous track, and not on the current track,
this would effectively set `actstrip = NULL`. This is now avoided.
Another benefit is that the search for the active strip is stopped as soon
as it's found, which should increase performance a tiny bit.
The `update_active_strip_from_listbase()` function took meta-strips in
the "source" list into account, but didn't recurse into the
corresponding meta-strip of the "destination" list. This is now fixed.
`update_active_strip_from_listbase()` needed a few changes to resolve
the issue:
- It was renamed to `find_active_strip_from_listbase()` to limit its
reponsibility to just finding the active strip. It now leaves the
assignment to the caller. This reduces the number of parameters by 1
and makes recursion simpler.
- The destination strips are now, like the source strips, passed as
`ListBase`, so that both source & dest can be recursed simultaneously.
This issue was only exposed by ba49345705. The ID pointer of the
material's parent tree-element wasn't actually pointing to an ID, but to
the list-base containing the IDs. It was just unlikely to cause issues
in practice, although an assert was thrown.
Just don't do anything if the object or object-data to unlink the
material from could not be found. The following commit will introduce a
error message about this.
Instancing with geometry nodes uses just the evaluated Mesh, and ignores the
Object that it came from. That meant that it would try to look up the subsurf
modifier on the instancer object which does not have the subsurf modifier.
Instead of storing a session UUID and looking up the modifier data, store a
point to the subsurf modifier runtime data. Unlike the modifier data, this
runtime data is preserved across depsgraph CoW. It must be for the subdiv
descriptor contained in it to stay valid along with the draw cache.
As a bonus, this moves various Mesh_Runtime variables into the subsurf runtime
data, reducing memory usage for meshes not using subdivision surfaces.
Also fixes T98693, issues with subdivision level >= 8 due to integer overflow.
Differential Revision: https://developer.blender.org/D15184
The value of disabled buttons shouldn't be changed through dropping onto
it. Check for the disabled state in the drop operator poll, so the
dragging code will change the cursor to show that dropping isn't
possible at the given cursor location.
This could spam the console with errors (potentionally slowing down in
cases).
Was added in rBeae36be372a6, but not used.
Maniphest Tasks: T98565
Differential Revision: https://developer.blender.org/D15113
The initial point count check was only being done for Bezier curves.
This revision fixes T98624 by adding the check for NURBS curves as well.
Reviewed By: HooglyBoogly
Maniphest Tasks: T98624
Differential Revision: https://developer.blender.org/D15140
The operator set the color but the factor of the mix value was not updated and as the default value was 0, the color was not vivible and only worked when the stroke was previously painted.
Caused by {rB6a3c3c77b3eb}.
Displacement and wave were tagging the original mesh normals dirty,
instead the result's normals need tagging. Seems like a typo in above
commit (similar to rBfe43c170831f).
Maniphest Tasks: T98727
Differential Revision: https://developer.blender.org/D15165
This bug was unreported. This was triggering a linking error caused by
the vertex shader not having a local version of `attr_load_temperature_post`
and `attr_load_color_post`.
This was caused by the `copy_m4_m4` trying to copy the `object_to_texture`
from `drw_grid` which was `nullptr`.
Fixing this also exposed that rendering such volumes (without any valid
grid attributes) is not supported and we should follow what Cycles does.
Differential Revision: https://developer.blender.org/D15147
This avoid leaving a `GPUMaterial` in a `GPU_MAT_QUEUED` state which would
block rendering.
Fix T98603: Hang when saving project in material preview mode
Maniphest Tasks: T98603
Differential Revision: https://developer.blender.org/D15135
The original assert did not take into account the offset size in the loop being -1. The tests were then run in non-debug mode, so while the mesh regressions still passed, the false positive asserts that happened were not caught.
Differential Revision: https://developer.blender.org/D15136
Rendering directly to a resource using OpenGL interop and Hgi
doesn't work in Houdini, since it never uses the resulting resource
(it does not call `HdRenderBuffer::GetResource`). But since doing
that simultaneously disables mapping (`HdRenderBuffer::Map` is
not implemented then), nothing was displayed. To fix this, keep
track of whether a Hydra viewport does support displaying a Hgi
resource directly, by checking whether
`HdRenderBuffer::GetResource` is ever called and only enable use
of OpenGL interop if that is the case.
Differential Revision: https://developer.blender.org/D15090
This was leading to some crashes and warnings such as:
"Code marked as unreachable has been executed. Please report this as a bug."
Differential Revision: https://developer.blender.org/D15116
This did not refresh the Image editor, but more importantly this now
appeared cropped (a regression from the partial image updater).
Solved in the RNA function by:
- calling BKE_image_partial_update_mark_full_update
- sending appropriate notifier
Maniphest Tasks: T98573
Differential Revision: https://developer.blender.org/D15110
Regression in [0]. When zoomed in, we can be within the face of an
island but too far from an edge, in this case
uv_find_nearest_face_multi_ex is used instead of
uv_find_nearest_edge_multi with the consequence that hit.l cannot be
used in uvedit_uv_select_test (it is NULL).
Instead, use uvedit_face_select_test instead in this case.
[0]: d356edf420
Reviewed By: campbellbarton
Ref D15100
The original index layer was not initialized properly.
Supporting original indices properly for this node is
doable, but for now it is better to simply initialize them
to the "none" value to fix the crash.
Differential Revision: https://developer.blender.org/D15105
The problem was that copying a `CDDerivedMesh` (`CDDM_copy`) doesn't
copy the `vert_normals` reference that it takes from a mesh. Since this
entire area is almost completely broken anyway (mainly in terms of
ownership handling), for now we can just avoid copying the `DerivedMesh`
in the zero levels case.
Longer term, this area should be refactored to remove `DerivedMesh`
and use the newer subdivision evaluation system.
Differential Revision: https://developer.blender.org/D15099
Crash happened because code could not find a valid base in current scene
after adding the object, added some checks for that.
Root of the issue was wrong assumptions in `BKE_object_add` logic, which
would pick the first valid ancestor collection in case initially
selected collection was not editable. In some case, this could pick a
collection not instanced in the current scene's view layer, leading to
not getting a valid base for the newly added object.
Addressed this by adding a new variant of `BKE_collection_object_add`,
`BKE_collection_viewlayer_object_add`, which ensures final collection is
in given viewlayer.
Code was using the loop [which is looping over the selection] index as
an index for the lookup into the edges directly, but needs to be a
lookupinto the IndexMask.
Also renamed the variable (as used elsewhere) to make this more clear.
If accepted, would be nice to still get this into 3.2.
Maniphest Tasks: T98536
Differential Revision: https://developer.blender.org/D15089
The tweak tool was toggling node selection twice, as the selection
key-map is already accounted for in the node key-map there is no need
to duplicate the actions in the tweak tool.
Although reusing the same patch coordinate for all corner pointing the
same vertex works for interpolation vertices, it does work for
interpolation face varying attributes. So we need to keep the original
patch coordinates around for face varying interpolation. This was caused
by the previous fix (a5dcae0c64).
The existing BUTTONS_OT_file_browse operator that's used for
uiTemplateImage layouts fails to work correctly with UDIM textures.
This is mainly due to it not realizing that it must tokenize the
filepath before signaling that an update has been made. It also doesn't
work correctly when executing its SHIFT-click behavior to open the image
in an external application. Lastly, it doesn't set the filters to Images
and Movies which is suboptimal for the user.
The new operator takes the unique features of BUTTONS_OT_file_browse
and creates a customized variant better suited for images.
Differential Revision: https://developer.blender.org/D14824
Also cleaned up the code a tad bit. Note that I
found two more bugs:
* GPU subdivision attribute interpolation
is producing visual artifacts.
* "Show on cage" mode for subdivision surface
just shows black colors.
Sculpt undo now detects if an attribute layer has
changed type/domain and unconverts it back. This
is a temporary workaround to a more fundamental
bug in the undo system.
Memfile undo assumes it can always rebuild the
application state from the prior undo step,
which isn't true with incremental undo systems.
The correct fix is to push an extra undo step prior
to running an operator if an incremental undo system
is active and the operator is using memfile undo.
Removed OPTYPE_UNDO flags from the swap brush colors and
sample color operators. These types of operators are
not supposed to be undoable in the first place. Also
memfile undo is too buggy for it.
The issue was that the extend socket (the last empty socket in
Input/Output nodes) was repeatedly removed and added again,
which caused more updates than necessary. Now, the extend
socket is kept if it existed already.
Differential Revision: https://developer.blender.org/D15084
Move MNEE to own kernel, separate from shader ray-tracing. This does introduce
the limitation that a shader can't use both MNEE and AO/bevel, but that seems
like the better trade-off for now.
We can experiment with bigger kernel organization changes later.
Differential Revision: https://developer.blender.org/D15070
Issues stems from the mesh not being watertight. This was caused by
floating point precision issues when evaluating patch coordinates at
patch boundaries (loops/corners in different patches pointing to the same
vertex). To fix this we ensure that all loops pointing to the same vertex
share the same patch coordinate. This keeps code simple, and does not
require to track precision issues in floating point math all over the
place.
See previous commit for an explanation of what went wrong. Similar to
the fix there, we also have to update the dragged data (e.g. the
data-block) referenced by the button.
Committing separately since this could cause further issues.
In Blender buttons are recreated over redraws, except of the active
button which is kept alive, and replaces the new version of itself in
the new redraw. In order to do that, the button needs to be recognized.
This process of recognizing and matching buttons from different redraws
isn't always bullet-proof. That's okay-ish, but we have to make sure
that the relevant data of the old active button is updated with the
newest data.
Here the matching would go wrong, and the new active button was
recognized as the old active button, which was in fact removed when the
asset was cleared. This patch makes sure the image buffer attached to
the buttons is updated when buttons were recognized as matching.
Note that the dragging will still use the wrong data-block, this will be
fixed in the following commit.
When multiple objects are in edit mode, UVs for the objects, except for
the first one (in rendering order) appear corrupted. The corruption is
because the UVs are not evaluated as the compute shader is not bound,
thus we read unitialized memory.
We keep track of the currently bound shader in the GPU context in order
to avoid unnecessary shader switches in case the same shader is used in
consecutive calls. However, the shader used by the OpenSubdiv evaluator
is not part of Blender and therefore not tracked via the GPU context.
When extracting UVs for multiple objects, we only ever run a single
shader (FVar evaluation). However, between the compute calls, we also
call the OpenSubdiv stencil evaluation shader, which uses `glUseProgram`
modifying the current program, outside of our control, which then also
unbinds the Blender compute shader making the compute dispatch fail ("No
active compute shader").
The fact that extracting the UVs for the first rendered object works is
because another (Blender) shader was bound in the GPU context prior to
our binding of our evaluation shader.
To fix this, we remember, in the OpenSubdiv evaluator, the current
program so that it can be reset after the stencil program is done.
Differential Revision: https://developer.blender.org/D15064
The operator assumed it was called on a mesh object, which has
mostly been the case because of lack of support for other object
types, but the new curves object is supported, which is the situation
in the report.
Differential Revision: https://developer.blender.org/D15063
Some of the tools in sculpt mode were still referring to the previously experimental sculpt vertex colors.
They should instead refer to color attributes.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D15073
This commits corrects the calculation of the Frame Node label size,
making it independent of the 'Line Width' user preference.
Since `U.dpi` is actually DPI divided by `U.pixelsize` and `U.pixelsize`
is calculated from line-width multiplying by `U.pixelsize` undoes
the connection between line-width and label size.
It now stays the same, regardless of the line-width setting.
Reviewed By: Julian Eisel, Harley Acheson
Differential Revision: https://developer.blender.org/D14338
These lines were not used now because the handling of copy data has changed.
Assigning the `eval` data can produce unexpected result, especially since everywhere ID_RECALC_TRANSFORM is used, we also do a copy on write. That should take care of `ob->data` for the eval object.
The problem was because the check was done with the total weights of the first element of the array and if this was null or 0, the weights were not duplicated.
As this bug was introduced fixing T97150 due a problem in the weight data, now instead to duplicate all stroke data to create the perimeter for the PDF/SVG, only the points are duplicated because the weights are not needed. This fix the original bug and also reduce the memory used by the export process.
Subdivision did not properly update when evaluating first without and then with
orco coordinates. Now update the subdivision evaluator settings every time, and
reallocate the vertex data buffer when needed.
there is an additional issue in this file where orco coordinates are not
available immediately on the first frame when they should be, and only appear
on the second frame. However that is an old limitation related to the depsgraph
not getting re-evaluated on viewport display mode changes, here we just fix the
crash.
I'm not sure what is causing this. Vertex normals get corrupted
for paint and mask sculpt brushes but not the normal geometric
ones. Since we don't actually need to recalculate normals
here to begin with I've just disabled it. The code now
calls the appropriate node mark update function based on
the sculpt tool.
Previously, when there were multiple curve points at the same or
almost the same position, the computed tangent was unpredictable.
Now, the handling of this case is more explicit, making it more
predictable. In general, when there are duplicate points, it will just use
tangents from neighboring points now.
Also fixes T98209.
Differential Revision: https://developer.blender.org/D15016
The original fix for T97366 was too restrictive and breaks real-world
cases of single-file UDIM textures. See D13297 for an example.
This patch effectively reverts the original fix and instead fixes the
downstream code to accept single-file ranges as necessary.
Note: This means it is very important for users to make use of the
"UDIM detection" option during `image.open` or drag n' drop scenarios in
order to declare their intent when loading their files.
Differential Revision: https://developer.blender.org/D14853
After recent changes to Nishita sky to clamp negative colors, the pixels ended
up a bit brighter which lead to them exceeding the half float max value. The
CUDA float to half function seems to need clamping.
After removing the default 'Home' shortcut for "Frame All", a NDOF (3D
Mouse) default shortcut was still available for the operator. The event
filtering introduced in 4357fb63db was missing the NDOF filtering
logic. So while the context menu correctly found the NDOF keymap item,
its actual shortcut change/removal code incorrectly filtered out the
NDOF keymap items and thus failed to find the item.
This is a regression caused by a230445cae.
The internal cause of the issue was that the synchronization component
was no longer tagged as visible (and hence not evaluated) as it not
connected to any directly visible IDs.
Changed the logic in a way that if any component of an ID is evaluated
the synchronization component will be evaluated as well.
The naming of the flag in the component node is a bit confusing, but
for the simplicity of the change for the upcoming release left it
unchanged.
`BKE_id_delete` should only check for consistency of user count with
regards to the tags and flags of the ID, not 'protect' nor even warn in
case a 'fake user' ID is deleted (such higher-level checks are to be
handled by higher-level code).
Also replace the assert + debug print by a CLOG error, this avoids
'assert crash' while still failing tests, and always producing a useful
message.
Fixes T98374 and T98260.
Callbacks used in the gizmo operator template don't support updating
while being dragged, set the EXCLUDE_MODAL flag so the offsets
aren't accumulated. Also fix the offset being applied twice to the
move gizmo.
A global variable was mistakenly used here which would accumulate the
vertex attributes (leading to an assertion failure after a while), use
the wrong number of components depending on the attribute data type,
among other issues.
Faces, edges, and vertices are still shown when GPU subdivision is
actived. This is because the related edit mode flags were ignored by the
subdivision code.
The flags are now passed to the various compute shaders mostly as part of
the extra coarse data, also used for e.g. selection. For loose edges, a
temporary buffer is created when extracting them. Loose vertices are
already taken into account as it reuses the routines for coarse mesh
extraction, although `MeshRenderData.use_hide` was not initialized,
which is fixed now.
When appending an already linked data, `BKE_blendfile_append` would not
properly substitute the `item->new_id` pointer of link/append context
items with the local duplicate of the linked ID.
This would cause drag'n'drop of assets to work incorrectly in some
cases. Fixes part of T95706 and T97320.
Regression in [0] which removed the call to BVH-tree recalculation
before calculating the selection.
Instead of recalculating the BVH-tree, postpone recalculating mesh data
until after the selection has been calculated.
[0]: 6e77afe6ec
Added in [0] but isn't needed as all needed variables are in the
ViewContext. Avoid passing in the context is it makes debugging
issues with MESH_OT_knife_project more difficult to investigate since
it's possible values written to the ViewContext are ignored.
[0]: 6e77afe6ec
When no image user is known the last used frame of the image is used to
read a frame. When partial updating an image there is always an image user
that would use a zerod out image user, meaning the frame number is set to 0
when using the clone tool.
This fix syncs the frame with the last used frame of the image to ensure
that the buffer exists. There is a bailout in the overlay_edit_uv.c.
8f79fa9c67 was an attempt to fix this already, but it didn't quite
work. Problem was that the tooltip was messing with the drop-box and
operator properties, setting the name property for its own internal
logic. This name property would then be used rather than the session
UUID to query the material, which gave the wrong material (linking can
result in multiple IDs of the same type with the same name). A followup
commit will further sanitize this.
More and more of the drop operations are being switched to use the ID's
session UUID rather than the name, but the cleanup after a drop operator
was cancelled assumed they would set the name. They will now first
attempt to use the session UUID and fallback to the name if needed.
Add notifier such that the Motion Paths panel in the Object Properties tab
gets redrawn, after using the Create Motion Path button.
The reason it didn't update was that the button actually triggers a popup,
and then executes in the context of that popup. It now actually emits a
`ND_DRAW_ANIMVIZ` notifier, and ensures that the panel redraws on that.
It is a know issue that split normals aren't supported when using high
quality normals in the viewport. Some AMD platforms were pushed to use
high quality normals to work around a driver bug where 1010102 texture
formats `GL_INT_2_10_10_10_REV` wasn't uploaded to the GPU.
This change will remove commonly used polaris platforms from the
work-around. This has been tested with a RX480 against the latest AMD
whql drivers (22.5.1). Users need to ensure that they use the latest
drivers that are available on their platform.
Although this change doesn't fix the underlying issue to support edit
normals when high quality normals are enabled. It will not force that
common platforms cannot use a feature as their platform is forced into
using a work-around.
Since rBb47c5505aa37, Batchs containing GPUIndexBuf initialized via
a PyObject with buffer protocol no longer work.
This was because of an unsafe optimization in the GPUIndexBuf module
for Python.
So remove this micro-optimization.
When saving from the menu the region was not set,
causing the last region in `area->regionbase` to be used
as the region was assigned before comparison.
* Port over new code tables from Cycles
* Convert Rec.709 to scene linear for lookup table.
* Move code for wavelength and blackbody to IMB so they can access the
required transforms, which are not in blenlib.
* Remove clamping from blackbody shader to bypass the texture read.
Since it's variable now easiest to just always read from the texture
than pass additional parameters.
* Fold XYZ to RGB conversion into the wavelength table.
Ref T68926
Regenerate blackbody RGB curve fit to not clamp values, and extend down to
800K since it does now change below 965K.
Note that as before, blackbody is only defined in the range 800K to 12000K
and has a fixed value outside of that. But within that range there should
be no more unnecessary gamut clamping.
917c096be6 applied to objects only, this also applies the same fix for
the general 3D View drop operations, e.g. used for dragging materials,
images, worlds, etc.
This is needed to fix T95706, but apparently something else is still
going on. Needs further investigation.
The operators to handle object drag and drop (from the asset browser,
outliner, etc.) used the object name to find the object to add and
place. This is problematic with linking and/or library overrides, since
this may lead to multiple objects with the same name. So the wrong
object would be used by the drop operators.
Partially fixes T97320. T95706 probably needs the same fix but for
materials.
As a side-effect, the "Name" button won't show up in the Adjust Last
Operation panel anymore. This isn't really useful, and I doubt this was
ever intentionally exposed in the UI.
Steps to reproduce were:
- Open Clip Editor
- Call "Open Clip" (e.g. Alt+O)
- Select video file
The file wouldn't be loaded into the Clip Editor.
Caused by 7849b56c3c.
This is part of a fix for T88570, where the file selector would crash
when activated multiple times.
Calling save multiple times would free the operator, leaving a dangling
pointer which was used when panels were visible that accessed the
"active_operator".
Reviewed By: Severin
Ref D14905
Existing code to replace the file operation was failing when done from
the window for the file operation itself.
Basically, this patch does two things:
- Implement a well defined window context to use as the "owner" or
"root" of the File Browser. This will be used for managing the File
Browser and to execute the file operation, even after the File Browser
was closed.
- Ensure the context is valid when dealing with file File Browser event
handlers.
Previously the window context just wasn't well defined and just happened
to work well enough in most cases. Addressing this may unveil further
issues, see T88570#1355740.
Differential Revision: https://developer.blender.org/D13441
Reviewed by: Campbell Barton
`BKE_nlastrip_find_active()` and `update_active_strip()` now do a more
thorough search for the active strip, by also recursing into
meta-strips.
There were some assumptions in the NLA code that the active strips is
contained in the active track. This assumption turned out to be false:
when there is a meta-strip, the active strip could be inside that, and
then it's not contained in `active_track.strips` directly.
Apart from the above, there are other situations in which the track
pointed to by `AnimData::act_track` does *not* contain the active strip
(even indirectly).
Both these cases can happen when the transform system is moving a strip
that's in tweak mode. Entering tweak mode doesn't just search for the
active track/strip, but also falls back to the last-selected ones. This
means that the track/strip pointers & flags can be out of sync with
what's being tweaked/transformed. Because of this, the assumption that
there is any relation between "active strip" and "active track" has been
removed from the `update_active_strip()` function.
All this searching could be prevented by passing the `AnimData` to the
code that duplicates tracks & strips. That way, when the active
track/strip is dup'ed, the `AnimData` pointers can be updated as well.
That would require more refactoring and thus could introduce other bugs,
so that's left for later.
Found those missing casts while looking into a crash report made in
the Blender Chat. Was unable to reproduce the crash, but the casts
should totally be there to avoid integer overflow.
* Rename ambiguous rgb to scene_linear in some places
* Precompute matrices to directly go to scene instead of through XYZ
* Make function signatures more consistent
Between scene linear and sRGB, XYZ, linear Rec.709 and ACES2065-1.
And add some clarifications about color spaces in the docs.
Fixes T98267
Ref T68926
Differential Revision: https://developer.blender.org/D14989
Some RNA property update callbacks can already generate such modifier,
and only one is allowed per object, so had to adapt code actually adding
new modifiers in liboverride apply codebase.
Also fixed an unreported issue of incorrect interpolation of thickness.
Reviewed By: Aleš Jelovčan (frogstomp), Antonio Vazquez (antoniov)
Differential Revision: https://developer.blender.org/D15005
The final normalization step of sculpt normal calculation iterates over
all unique vertices in each node and marks them as done. However,
storing the done mask in a bitmap meant that multiple threads could
write to a single byte at the same time, because the bits for separate
threads could be stored in the same byte. This is not threadsafe
Fixing this issue seems to improve performance as well. First I tested
just clearing the entire bitmap after the normal calculation. Then I
tested using an array of booleans instead, which turned out to be
slightly better, and simplifies code a bit.
I tested on a Ryzen 3800x, on an 8 million polygon subdivided
Suzanne by using the grab brush with a radius large enough to
affect most of the mesh.
| Original | Clear Entire Bitmap | Boolean Array |
| --------- | ------------------- | ------------- |
| 67.9 ms | 59.1 ms | 57.9 ms |
| 1.0x | 1.15x | 1.17x |
Differential Revision: https://developer.blender.org/D14985
When labels in popups (typically headings) matched the name of a button,
the label would be activated instead of the button.
This caused the unwrap menu in the UV editor not to re-select "Unwrap"
when opening a second time.
The change was kind of intentional on {rB21e72496a629}.
That commit made mouse movement to "select" the contraint in Auto
Constraint a requirement.
This deduplicated the code a bit, but this requirement is not
comfortable for the first "selection" of the contraint.
So the constraint "selection" is now done in two ways:
- If there is no contraint, the "selection" is done immediately;
- If there is already a constraint, the "selection" is delayed by 1 event to simulate a constraint cancellation if there is no mouse movement.
Existing code for the `Move` operator, and some `Collections` panel
operations (Object properties) was absolutely not override-safe, and
sometimes not even linked-data safe.
That max number of `10000` level of recursivity was a typo (should have
been `1000`), but even that is way too high, typical sane situation
should not lead to more than a few tens of levels, so reducing the max
level to 200.
Also improve error message with more context info about the issue.
Found while investigating issues for the Blender Studio's Heist production.
That max number of `10000` level of recursivity was a typo (should have
been `1000`), but even that is way too high, typical sane situation
should not lead to more than a few tens of levels, so reducing the max
level to 200.
Also improve error message with more context info about the issue.
Found while investigating issues for the Blender Studio's Heist production.
In some cases (when there is an evaluated curve), the conversion code
would try to free the evaluated data-block twice, because freeing the
object would free it from `data_eval` and then the data-block was freed
again explicitly. Now check if the data-block is stored in `data_eval`
before freeing `object.data` manually. This is another area that's made
more complex by the fact that we change the meaning of `object.data`
for evaluated objects. The solution is more complicated than it should
be, but it works whether or not an evaluated mesh or curve exists.
It wasn't possible to temporarily orbit the view, then set back to an
axis-aligned view.
Details:
- It was possible to change RegionView3D.view_rotation while the view
kept the axis alignment value (Top, Left, Front .. etc) which
displayed in the viewport overlay.
Now changing the view rotation directly or via "view_matrix" resets
the axis-alignment - clearing when the view is no longer axis-aligned
or assigning the newly aligned axis.
- RegionView3D.is_orthographic_side_view added in [0] could be assigned
but wasn't useful as it treated an enum as a boolean only setting the
RegionView3D.view to RV3D_VIEW_USER or RV3D_VIEW_FRONT.
Now enabling this aligns the viewport rotation to it's closest
axis-aligned orientation setting RegionView3D.view & view_axis_roll
accordingly. Note that the "orthographic" term is misleading as the
property only relates to axis-alignment, not to the
perspective/orthographic setting. We could consider deprecating the
current naming.
[0]: 63bae864f4
View roll checked if the resulting roll was close to a view axis
but didn't write the aligned quaternion back to the final result.
Add ED_view3d_quat_to_axis_view_and_reset_quat since most callers
to ED_view3d_quat_to_axis_view will reset the quaternion when a view
aligned axis is found.
Issue revealed by rB90e12e823ff0. Hidden objects may not be fully
evaluated by the despgraph, do not attempt to restore edit/sculpt/etc.
modes for those.
Should also be backported to 2.93 LTS release.
Sometimes integers are mixed using float weights. In those cases
the mixed result has to be converted from into float again.
Previously, this was done using a simple cast, which was unexpected
because e.g. 14.999 would be cast to 14 instead of 15.
Now, the values are rounded properly.
This can affect existing files unfortunately without a good option
for versioning. Gladly, very few files seem to depend on the details
of the old behavior.
Differential Revision: https://developer.blender.org/D14892
This makes changes to the opensubdiv module to support additional vertex data
besides the vertex position, that is smootly interpolated the same way. This is
different than varying data which is interpolated linearly.
Fixes T96596: wrong generated texture coordinates with GPU subdivision. In that
bug lazy subdivision would not interpolate orcos.
Later on, this implementation can also be used to remove the modifier stack
mechanism where modifiers are evaluated a second time for orcos, which is messy
and inefficient. But that's a more risky change, this is just the part to fix
the bug in 3.2.
Differential Revision: https://developer.blender.org/D14973
Previously, the depsgraph assumed that every node tree might contain
a reference to a video. This resulted noticeable overhead when there
was no video.
Checking whether a node tree contained a video was relatively expensive
to do in the depsgraph. It is cheaper now due to the structure of the new
node tree updater.
This also adds an additional run-time field to `bNodeTree` (there are
quite a few already). We should move those to a separate run-time
struct, but not as part of a bug fix.
Differential Revision: https://developer.blender.org/D14957
If making liboverride of an empty collection, this (root of override
hierarchy) collection would get untagged in code when checking for
collections that do not need to be overridden, leading to not overriding
this root collection, and later in code to using NULL pointer.
Regression caused by [0] which changed node selection to use
PRESS for selection and CLICK_DRAG to transform the selection.
This conflicted with Alt-LMB which would select the node then
pass-though to node.background_sample, preventing the drag event
from being activated.
Resolve by only activating background-sample when the cursor
isn't over a node or socket.
[0]: 4c3e91e5f5
Issue was caused by treating such strip as meta and using
`seq->channels` directly, which is not valid for scene strips.
Pointer to channels is now provided by function
`SEQ_get_seqbase_from_sequence`.
Ensure the correct total/diffuse/transmission depth is set when evaluating
shaders for MNEE, consistent with regular light shader evaluation.
Differential Revision: https://developer.blender.org/D14902
We need to ensure `Main.filepath` is consistent with the current path
where we are saving the .blend file, otherwise some path processing code
can produce invalid results (happens with e.g. the code syncing the two
path storages in Library IDs).
The problem is that depsgraph evaluation happens before the OpenGL context
is initialized, and so modifier evaluation happens without GPU subdivision.
Later the BKE_subsurf_modifier_can_do_gpu_subdiv test in the draw code gives
a different result.
This just checks if the mesh has information for GPU subdivision in the draw
code, and if so uses it. This is only set if the test for supported GPU
subdivision passes in the modifier evaluation.
Additionally it may be good to perform OpenGL context initialization earlier
so background render can take advantage of GPU subdivision, but this is more
complicated.
Differential Revision: https://developer.blender.org/D14969
This patch fixes long standing issue of reverse mode only performing loop 2 times and then stops displaying.
Differential revision: https://developer.blender.org/D14921
A regression since 113b8030ce: circle selection operators does not
define properties like deselect_all and a special name callback is to
be used for those.
This is what was already done for circle select in the 3D viewport.
Some other spaces were using the generic pick operator for the circle
selection which causes error prints in the console.
Regression in tests from [0] tests were written to assume a newline was
added to the result of Text.as_string which is no longer the case.
[0]: f4ff36431c
Refactoring event click-drag detection broke click detection for
simulated events. Resolve this by sharing logic for update previous
values in `wmWindow.eventstate` for regular event handling
(no functional changes for non-simulated events). Failure to detect
clicks for simulated events broke the undo test
`test_undo.view3d_multi_mode_select` in `../lib/tests/ui_simulate/run.py`.
All undo tests now pass.
Moving Text.as_string() from Python to C [0] added an extra new-line
causing a round-trip from_string/to_string to add a new-line,
this also broke the undo test `test_undo.text_editor_simple` in
`../lib/tests/ui_simulate/run.py`.
[0]: 231eac160e
When texture painting, brush textures and brush texture masks were not
transformed to account for UDIM tiles.
Differential Revision: https://developer.blender.org/D14671
Fix a crash when a driver variable targets an object and uses
`data.shape_keys.key["name"].value` in its expression.
The fix consists of adding an extra relation from the targeted object's
`GEOMETRY` component to the driver evaluation. This ensures that its
`data` pointer has been evaluated by the depsgraph and is safe to
follow.
This also resolves the concern raised on rB56407432a6aa.
Reviewed by: brecht
Differential Revision: https://developer.blender.org/D14956
The 'OVERRIDE_HIDDEN' extra collection would often be mistakenly added
to a linked collection, which is totally forbidden and guaranteed to
crash on undo/redo.
Reworked the code instantiating that extra collection in a more generic
and hopefully robust way now.
Made tangent frame consistent across the surface regardless of the sample,
which was not the case with the previous algorithm. Previously, a tangent
frame would stay consistent for the same sample throughout the walk, but not
from sample to sample for the same triangle. This actually resulted in code
simplification.
Also includes additional fixes:
* Fixed an important bug that manifested itself with multiple lights in the
scene, where caustics had abnormally low amplitude: The final light pdf did
not include the light distribution pdf.
* Removed unnecessary orthonormal basis generation function, using cycles'
native one instead.
* Increased solver max iteration back to 64: It turns out we sometimes need
these extra iterations in cases where projection back to the surface takes
many steps. The effective solver iteration count, the most expensive part,
is actually much less than the raw iteration count.
Differential Revision: https://developer.blender.org/D14931
We really need to fix how unprojected radius (scene unit) works.
What happened is the paint code updates the brush's normal radius
with the current unprojected pixel radius, which was then
used by texture brush tiled mode.
To fix this I just cached the pixel radius at stroke start in
UnifiedPaintSettings->start_pixel_radius.
This was due to using `BKE_scene_has_object` function, which uses the
cache of bases of the viewlayers, which do not have entries for the
content of excluded collections... Now use
`BKE_collection_has_object_recursive` instead.
Steps to reproduce were:
- Factory startup
- Right-click in 3D View
- Move the mouse over "Shade Flat", wait for the tooltip
The changed logic in 4680331749 to lookup an active button was
incorrect. It didn't respect the priority of active button candidates.
Issue is that the operator acts on the active button, and also uses that in the
poll. So the actually active button would affect the poll of a different
button. For the superimposed icons we need to be able to execute these polls
properly for non-active buttons.
This enables temporarily overriding the active button for lookups via context.
While a bit of a hack it makes sense conceptually.
Reviewed By: Campbell Barton
Maniphest Tasks: T97518
Differential Revision: https://developer.blender.org/D14880
Regression caused by [0] which made `ui_but_is_interactive` consider
label buttons with tool-tips to be interactive. This prevented
the clicks to pass through to the nodes for selecting/dragging.
Resolve this by allowing buttons to be activated for the purpose
of showing tool-tips but otherwise considering them disabled
(as if the UI_BUT_DISABLED is set when handling events).
[0]: 484a914647
Reviewed By: Severin
Ref D14932
When extracting UV point indices, only the vertex points coming from the
original geometry should be drawn. For this, the routines (for subdivision
and coarse meshes) would only consider a vertex to be real if the extraction
type is `MAPPED`, and that an origin index layer on the vertices exist
with a valid origin index for the current vertex.
However, if the extraction type is `MESH`, which can happen with for
example an empty Geometry Node modifier, or with deferred subdivision,
this would consider every vertex to not be "real" and therefore hidden from
the UV editor.
This reworks the condition for "realness" to also consider a vertex to be
real if there is no origin layer on the vertices. The check on the extraction
type is removed as it becomes redundant.
This only modifies the check in the UV data extraction for point indices,
however similar checks exist throughout the extraction code, these will
be dealt with separately in master.
Differential Revision: https://developer.blender.org/D14773
Since rB2a7a01b339ad, `lineSmooth` has lost its default value of true.
rBa0a99fb25284 only fixed the problem on master.
But thanks to @hitrpr for spotting the bug in version 3.2 too.
Differential Revision: https://developer.blender.org/D14876
Disable the new more accurate timing code, this is not needed for Blender.
In USD itself this code is disabled on macOS anyway, so it should operate fine
without it.
Ref T97950, T95206
Differential Revision: https://developer.blender.org/D14928
This abuse of one one size value to handle another allocated array of a
different size is bad in itself, but at least now read/write code of
this modifier should not risk invalid memory access anymore.
NOTE: invalid memory access would in practice only happen in case endian
switch would be performed at read time I think (those switches only check
for given length being non-zero, not for a NULL data pointer...).
This was because the main `surface_vert.glsl` was changed to accomodate the
needs of the `ShaderCreateInfo` but was still used by the cryptomatte
shader. The fix is to include the same libraries as the material shaders
and bypass `attrib_load()`.
This was caused by the `mb_data->obmat[]` being wrong because they are
now shared between the particle system and the object.
But Hair need the dupli parent matrix instead of the object matrix.
Disabling `Show Emitter` option fixes the bug.
To avoid this problem, request a different `EEVEE_ObjectMotionData`
for particle systems using a different key pointer in the hash.
This is a bit dirty but there is less code polution using this workaround.
Differential Revision: https://developer.blender.org/D14911
There are two problems when adding a paint slot to an object without an
existing material. First, the `invoke` method creates a material on the
object. This modifies the object even if the operation is not executed.
Second, the fill color defaults to black when there is no existing
material (even when adding a normal, bump, or displacement layer).
This patch moves the material creation to the `exec` method.
When no material exists on the object, a default Principled BSDF is
referenced for default colors in the `invoke` method.
Differential Revision: https://developer.blender.org/D14828
If merging is enabled, the mesh might be recreated before
the dirty flag can be cleared, which means the normals aren't
valid anymore. To fix this, clearing the dirty flag should happen
before the merging. This is an existing bug, just exposed by
more recent explicit dirty normal tagging.
In rare cases the mesh has not been evaluated when snapping, this fix
just prevents the crash as is done elsewhere in Blender when the
evaluated mesh isn't available, there is a separate report (T96536)
about evaluation not working properly.
Now add a default ".usdc" file extension if no (or the wrong) extension
is given instead of presenting the user with the error that "no suitable
USD plugin to write is found".
This is in line with how other exporters do this.
Maniphest Tasks: T97947
Differential Revision: https://developer.blender.org/D14895
GLSL has different max number of ssbo per glsl stage.
This patch checks if the number of compute ssbo blocks matches
our requirements for the GPU Subdiv, before enabling it.
Some platforms allow more ssbo bindings then blocks per stage.
Geometry nodes can generate color attributes that aren't on point or corner domain.
When not found in these domains it will be processed as a common attribute.
Manually revert commit [0] as it caused problems macOS (reported T96435).
- Includes fixes from [1] & [2].
- T98037 TODO has been created to keep track of this feature.
Thanks to @jbakker & @sergey for investigating this issue as I wasn't
able to reproduce the bug.
[0]: 0cb5eae9d0
[1]: cb986446e2
[2]: cc8fe1a1cb
Knife projection BVH-tree lookup could use invalid indices since the
mesh being cut is also used for BVH intersection tests.
Solve by storing triangle indices when knife project is used so a
triangle index can always be used to look up original coordinates of a
triangle.
This is caused by the geometry shader used by the edit mode line drawing.
If the drawcall uses indexed drawing and if the index buffer only contains
restart indices, it seems the result is 1 glitchy invocation of the
geometry shader.
Workaround by tagging these special case index buffers and bypassing
their drawcall.
Assume geometry is always potentially animated, since we can't use our heuristic
to detect if the object is potentially animated by looking at modifiers on the
object.
The main original reason for this check was to avoid evaluating subdivision
surfaces for many static objects, which is not happening here anyway.
New OBJ exporter is missing "Path Mode" setting for exporting .mtl
files. The options that used to be available were: Auto, Absolute,
Relative, Match, Strip Path, Copy. All of them are important. The new
behavior (without any UI option to control it) curiously does not match
any of the previous setting. New behavior is like "Relative, but to the
source blender file, and not the destination export file".
Most of the previous logic was only present in Python based code
(bpy_extras.io_utils.path_reference and friends). The bulk of this
commit is porting that to C++.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14906
Code was not really designed to hanlde corrupted (e.g. local ID) root
hierarchies, now it should handle better those invalid cases and restore
proper sane situation as best as possible.
Fixes crashes with some corrupted files from Blender studio.
Use the same method for node selection and dragging that is used
in the 3D viewport and UV editor. Instead of relying on a modal
operator - use the keymap to handle click/drag events.
Details:
Failure to transform unselected nodes was caused by [0] & [1] however
detecting drag relied on specific behavior which I don't think we should
be depending on.
This error happened when selection was defined both in the key-map for
the tool and for the node-editor.
- The left mouse button would activate selection in both the tool
and "Node Editor" keymap.
- The first selection would return `FINISHED | PASS_THROUGH` when
selecting a previously unselected node.
- The same PRESS would trigger a second selection would return
`RUNNING_MODAL | PASS_THROUGH`,
(starting a NODE_OT_select as a modal operator).
- In 3.1 (with tweak events) the modal operator would then exit and
fall-back to the tweak event which would transform the selected
nodes.
- In 3.2 (as of [0]) the PRESS that starts the modal operator is
considered "handled" and prevents drag event from being detected.
The correct behavior in this case isn't obvious:
If a modal operator starts on pressing a button, using that same the
release to generate drag/click events is disputable.
Even in the case or 3.1 it was inconsistent as tweak events were
generated but click events weren't.
Note: after investigating this bug it turns out a similar issue already
existed in 2.91 and all releases afterwards. While the bug is more
obscure, it's also caused by the tweak event being interrupted as
described here, this commit resolves T81824 as well.
[0]: 4d0f846b93
[1]: 4986f71848
Reviewed By: Severin
Ref D14499
The acquire locking of the draw manager introduced other issues.
The current implementation was a hacky solution as we know that the
final solution is something totally different {T98016}.
Related issues:
* {T97988}
* {T97600}
Viewports where cleared explicitly due to compatibility reasons with Intel iGPUs.
This slowed down other platforms as well, this wasn't noticeable on all platforms.
This patch will be more selective when to enable the workaround.
Currently only for iGPUs on Mac + Linux.
Introduced by {35594f4b92fa4cbb5b848f447b7a3323e572b676}.
Some platforms do not support temp variables to be used as inout parameter.
Detected on Mac with Intel iGPU.
During UV unwrapping, Cube Projection, Sphere Projection, Cylinder
Projection and Project From View (in the 3D Viewport), when "Correct
Aspect" toggle is active, it now uses a query cache to perform a
per-face aspect ratio ("per_face_aspect") correction for the active
image of each face.
Reviewed By: campbellbarton
Ref D14852
Some drivers completely forbid quote characters even in unused
preprocessor directives.
This patch adds a debug build check for all `.glsl` files that need to
be manually handled. For shared headers with `#include` directives, we
need to do runtime patching of the source to remove the quote.
Also fix an instance of the quotes check failing in `eevee_next`.
This was caused by the `Closure` members being added to the final contribution
more than once. The workaround is to clear the members once a closure has
been added to the final contribution. I used `inout` on `Closure` inputs
so that the render engine implementation of mix and add closure nodes
can do its own thing. The nodegraph handling of inout was changed for this
to work.
OCIO could build before pystring and imath due to
OCIO missing the dependencies on these two projects
No rebuild required as the build would have failed
during the libs build if you ran into this issue.
Curve tangent was correctly mistaken with curve normal.
This patch fixes the name of the output in the glsl function and make curve
attributes more explicit (with `curve_` prefix).
This also improve the normal computation by making it per pixel to match
cycles.
Also ports the changes to eevee-next.
The previous 3.1 libraries (accidentally) used glApi instead of GLEW and were
working for GPU subdivision, so revert to that. There's a suspected conflict
with Blender's own bundled GLEW or other issue with GLEW, causing the crash in
T97737.
The current GPU subdivision implementation does not need OpenCL, CUDA or GLFW.
So also remove libraries needed for that. It's simpler to stick to compute
shaders in OpenGL/Vulkan/Metal and not involve additional APIs.
Ref T95206
Differential Revision: https://developer.blender.org/D14898
The mesh drawing code used a different mesh to check whether or not to
draw face dots and to actually retrieve them. The fix is moving the
responsibility of determining whether to use subsurf face dots to the
creation of `MeshRenderData` where the mesh used for drawing is
known, rather than doing it at a higher level.
Differential Revision: https://developer.blender.org/D14855
`TEMPERATURE` type was also missing, not only the new-ish
`TIME_ABSOLUTE` one...
Added a static assert on the size of the `bpyunits_ucategories_items`
array, and a comment on anonymous enum of `B_UNIT_`, in the hope this
won't happen again in the future.
On certain systems when eevee is used in a 3d viewport could crash. It
happened more often on slower systems or systems with slower glsl compilers.
For example an Intel Mac Mini. The cause was that even if a GPUMaterial
was in used it could be freed.
For some reasons the mipmap count was not set to max during creation.
Probably it was mistaken with the UDIM tilemap texture which is only
1D and has only one mipmap.
This is because some drivers / GPU actually still do double buffer swapping
but others don't. Adding this do ensure the background color of the first
redraw.
Note that this fix was not tested on the problematic hardware and might not
solve the issue.
This was because the alpha clip thresholding was previously done in the
material nodes codegen. Now it is the responsibility of the engine to
implement it.
This adds a loose uniform that is set by EEVEE itself to control the clip
behavior.
Makes the File Browser filter by .obj and .mtl files by default again. Note
that this commit focuses on fixing this specific bug, further
refactors/tweaks/fixes are planned (see D14863).
Differential Revision: https://developer.blender.org/D14862
Reviewed by: Aras Pranckevicius
When resizing a viewport all engine instance data was cleared.
This wasn't the intended design and lead to performance regressions
in the image engine.
This patch makes sure that the instance data isn't cleared when
the viewport size changes. When using instance data, draw engines
are responsible to update the textures accordingly.
This could also reduce flickering/stalling when resizing the viewport
in eevee-next.
Fixes T95428.
Reviewed By: fclem
Maniphest Tasks: T95428
Differential Revision: https://developer.blender.org/D14874
This patch implements T97613, switching viewport shading when using Color Filter and Mask By Color
ALSO, this patch makes it so viewport shading color switches only when SOLID mode is chosen, to prevent color switching when using other shading modes, without user noticing it.
{F13049889}
Reviewed By: JulienKaspar, joeedh, jbakker
Maniphest Tasks: T97613
Differential Revision: https://developer.blender.org/D14765
Related to T95616, the relationship between Image ID and ImBuf 'cached'
buffers can be fairly confusing when using the RNA API.
Reviewed By: campbellbarton, jbakker
Differential Revision: https://developer.blender.org/D14833
Regression caused by [0] which flushed selection from vertices -> edges,
causing additional edges to be selected. Now selected is flushed based
on the mode, instead of all elements. Note that these function names
could be improved to make it clearer how these flushing functions are
different.
Also skip flushing unless selection is performed.
[0]: 55c82d8380
Regression in [0] which is useful when applying modifiers as a shape-key
but not when applying modifiers which keeps the existing shape-keys.
[0]: 65c5ebf577
It was wrongly writing passes twice, for both the surface entry and exit points.
We can skip code for filtering closures, emission and holdout also, as these do
nothing with only a subsurface diffuse closure present.
The root of the issue is that compositor is using refresh mechanism
to handle recalc.
The safest fix which does not require deep refactor is to check to
whether node space was tagged for refresh from listener (currently
it is listeners which are responsible for tackling compositor tree
recalc).
Differential Revision: https://developer.blender.org/D14856
While possible extra whitespace after all OBJ/MTL keywords was properly
skipped, it was not done for the "f" (face definition) keyword.
While at it, also support indented keywords, i.e. extra whitespace at
the beginning of the line.
There's a tiny bit of performance drop while importing (e.g. importing
blender 3.0 splash scene: 53.38sec -> 54.21sec on my machine). But
correctness is more important.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14854
Support merging UV's that share the same vertex and are very close when
applying modifiers.
This is needed to prevent UV's becoming "detached" which can happen when
applying the subdivision surface modifier.
This regression was caused by [0] which removed selection threshold for
nearby coordinates. While restoring the UV selection threshold could be
done - some selection operations that walk around connected UV fans
wouldn't behave in a deterministic way (such as select shortest path).
There are also other cases where UV's may be compared without a
threshold such as tangent calculation and exporters which have their own
logic to handling UV's.
Also resolves T86896, T89903.
[0]: b88dd3b8e7
Reviewed By: sergey
Ref D14841
Recently `gpu_shader_3D_smooth_color_frag.glsl` had the uniform declarations removed.
For shaders without `ShaderCreateInfo` that require this source this results in the error:
```
ERROR (gpu.shader): hair_refine_shader_transform_feedback_workaround_create FragShader:
|
54 | fragColor = finalColor;
|
| gpu_shader_3D_smooth_color_frag.glsl:5:0: Error: Use of undeclared identifier 'fragColor'
| gpu_shader_3D_smooth_color_frag.glsl:5:0: Error: Use of undeclared identifier 'finalColor'
|
55 | fragColor = blender_srgb_to_framebuffer_space(fragColor);
|
| gpu_shader_3D_smooth_color_frag.glsl:6:0: Error: Use of undeclared identifier 'fragColor'
| gpu_shader_3D_smooth_color_frag.glsl:6:0: Error: Use of undeclared identifier 'fragColor'
```
So port that shader to use `ShaderCreateInfo`.
evice_update_preprocess is supposed to detect modified attributes and flag the
device_vector for a copy through device_update_flags. However, since object
attributes are only created in device_update_attributes afterwards, they can't
be included in that check.
Change the function that actually updates the device_vector to tag it as
modified as soon as its content gets updated.
Differential Revision: https://developer.blender.org/D14815
A shader node setup accidentally used the bump normal as emission. Bump
mapping nodes are excluded from light shader evaluation to reduce kernel size
and register pressure, but in that case should write zero instead of leaving
memory uninitialized.
Thanks to Lukas for helping identify the cause.
The coarse polygon count was set to the one of the BMesh instead of
the the one of the mesh used for subdivision, which caused the
compute shaders to output wrong data.
Similar issue/solution as in rB5188c14718c5 from this Monday actually,
there may be more of those still lurking around... Quite surprising they
all get reported now, this behavior has been in Blender since years.
The problem was the layer transformation was already applied in the layer and if we apply in the bake, we are doing double transformation.
Differential Revision: https://developer.blender.org/D14844
The issue was that the `NodeTreeRef` acceleration data structure was
rebuild much more often than necessary. That happened because the
Map Range node accidentally tagged the node tree for change even
though it did not actually change.
Differential Revision: https://developer.blender.org/D14842
Fixes T97794 (which is a reintroduction of an older issue T67266 that
has been fixed in the python importer, but the fix was not in the C++
one). Some software produces OBJ files with mtllib statements like
mtllib "file name in quotes.mtl", and the new importer was not stripping
the quotes away.
While at it, I noticed that MTLParser constructor was taking a StringRef
and treating it as a zero-terminated string, which is not necessarily
the case. Fixed that by explicitly using a StringRefNull type.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14838
Fix regression described in T97799.
Apply layer transform and layer parenting to all visible frames, i.e. active frame + onion skinning frames.
Reviewed By: #grease_pencil, antoniov
Maniphest Tasks: T97799
Differential Revision: https://developer.blender.org/D14829
When the profile only has one control point, its segment count was
determined incorrectly. There needs to be a special case for a single
control point, when there are no segments even if the curve is cyclic.
Recalc selection count at the end of the circle selection.
This was removed from circle selection as it became a performance
bottleneck, helas some operators rely on it.
These kinds of depsgraph evaluations should not be marked as low priority as
this could negatively affect playback performance. Low priority should mainly
be used for background tasks.
Isolate frame writing task so that multithreaded image operations don't cause
the thread to start writing another frame. If that happens we may reach the
MAX_SCHEDULED_FRAMES limit, and cause the render thread and writing threads to
deadlock waiting for each other.
Additionally, don't set task priority to low because this may cause the task
scheduler to be slow in scheduling the write and color management tasks.
Overwriting UV map or vertex color data in Geometry nodes will move the
layers to another CustomData channel, and as such, will make attribute
lookup fail from the UVMap and Vertex Color nodes in EEVEE as the
CustomDataType will also be modified (i.e. no longer `CD_MTFACE` or
`CD_MCOL`).
As discussed in T93179, the solution is to use `CD_PROP_AUTO_FROM_NAME`
so that the render engine is able to find the attributes. This also makes
EEVEE emulate Cycles behaviour in this regard. `attr_load_uv` and
`attr_load_color` are also removed in favor of the generic attribute
API in the various GLSL shaders.
Although `CD_PROP_AUTO_FROM_NAME` is now used even for UV maps, the
active UV map is still used in case the attribute name is empty, to
preserve the old behavior.
Differential Revision: https://developer.blender.org/D13730
The crash is caused as the data for the UV editor is requested before
the data for the mesh as a separate draw update. Since building the UV
stretch angle buffer requires the position buffer, the latter is not
created yet in this case.
To fix this, create a local position buffer from the subdivision data. An
alternate fix was considered to remove the dependency on the position
buffer by interpolating on the GPU the coarse stretch angle buffer but
this did work. Maybe this will be revisited.
- Fix T97793: when a UV coordinate after vt is missing, use zero. While
at it, also use zeroes for positions & normals, since "maximum
possible float" is very likely to cause issues in imported meshes.
- Fix T97795: use 1.0 default if -bm value is missing, instead of zero.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14826
Fix several correctness issues where the new OBJ/MTL importer was not
producing the same results as the old one, mostly because the code for
some reason had slightly different logic. Fixes T97757:
- When .obj file tries to use a material that does not exist, the code
was continuing to use the previous material, instead of creating new
default one, as the previous importer did.
- Previous importer was always searching/parsing "foo.mtl" for a
"foo.obj" file, even if the file itself does not contain
"mtllib foo.mtl" statement. One file from T97757 repros happens to
depend on that, so resurrect that behavior.
- When IOR (Ni) or Alpha (d) are not specified in .mtl file, do not
wrongly set -1 values to the blender material.
- When base (Kd) or emissive (Ke) colors are not specified in the .mtl
file, do not set them on the blender material.
- Roughness and metallic values used by viewport shading were not set
onto blender material.
- The logic for when metallic was set to zero was incorrect; it should
be set to zero when "not using reflection", not when "mtl file does
not contain metallic".
- Do not produce a warning when illum value is not spelled out in .mtl
file, treat as default (1).
- Parse illum as a float just like python importer does, as to not
reintroduce part of T60135.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14822
Regression from rB1a7757b0bc69/rBa0acb9bd0cc0. Special handling
(averaging) of weights on merged center vertices also requires to be
'reversed' when new correct merge order is used, compared to previous
behavior.
Run autopep8 as well as clang-format when calling "make format",
the PATHS argument is passed to both utilities which will only operate
on files they support.
For example: `make format PATHS=release/scripts` formats Python scripts,
`make format PATHS=source/blender/blenlib` would format C/C++.
If users really want they can format C/C++ & Python files at the same
time since both formatting utilities filter on file extension.
`make format PATHS="release/scripts/startup/nodeitems_builtins.py source/creator/creator.c"`
A LIBDIR variable has been added to the GNUmakefile to simplify
references to this directory which can be one of 3 possible values.
Reviewed By: sybren, brecht
Ref D14789
Regression caused by [0] that caused the error message to be
created based on a normalized exception (which hid line numbers).
PyC_ExceptionBuffer{_Simple} & BPy_errors_to_report
no longer clears the exception.
This could have been resolved by changing python_script_error_jump
however that would involve changes to reference counting that are more
risky (noted in code-comment).
[0]: 2d2baeaf04
Leak was caused because output image buffer was initialized twice. Once
in speed effect strip and then by cross effect strip used for
interpolation feature.
This was reported explicitly for originally being in face selectmode,
but could also crash when in vertex selectmode (when doing multiple
cuts).
The reason here is that `MESH_OT_loopcut` switches to edge select mode
(needed for `TRANSFORM_OT_edge_slide`, done in `ringsel_finish`), but was
only doing this on the active object's editmesh, all other participating
meshes would keep their selectmode which would now be out of sync with
both the active object's editmesh and scene settings for these.
This causes problems later in 'Select Linked'. Here, a mixture of
objects are used. First the viewcontext is set up with the active
object, then all participating objects are iterated (changing the
viewcontext to another object), then `unified_findnearest` would use that
changed viewcontext which would now contain the last object iterated. To
repeat: this could now have a different selectmode than the active
object which is later **again** used to get the nearest `BMElem` from in
`EDBM_elem_from_selectmode`. So in the failing case, we could get an
edge (but no face because of edge selectmode) from `unified_findnearest`,
`EDBM_elem_from_selectmode` would return NULL though (edge provided, but
in face selectmode), leading to the crash.
To solve this I assume it is best to change selectmode on all
participating meshes in multi-object editmode loopcut if necessary so
these are always in sync for following operations.
Alternatively, `Select Linked` (and probably lots more operators) would
have to be tweaked to pay closer attention which object is really used
to get selectmode from.
Note the selectmode is actually set back from edge selectmode in certain
cases (see `USE_LOOPSLIDE_HACK`), this patch changes that as well to act
on all participating meshes.
Maniphest Tasks: T95752
Differential Revision: https://developer.blender.org/D14791
This commit implements copying of materials and material indices from
all of the boolean node's input meshes. The materials are added to the
final mesh in the order that they appear when looking through the
materials of the input meshes in the same order of the multi-socket
input node.
All material remapping is done with mesh-level materials. Object-level
materials are not considered, since the meshes don't come from objects.
Merging all materials rather than just the materials on the first mesh
requires a change to the boolean-mesh conversion. This subtly changes
the behavior for object linked materials, but in a good way I think;
now the material remap arrays are respected no matter the number
of materials on the first mesh input.
Differential Revision: https://developer.blender.org/D14788
The thickness in the deform mode was just correct if the radius when
drawing a stroke was set to 20. Now all factors that influence the
stroke thickness are considered in deform mode.
Differential Revision: https://developer.blender.org/D14691
Relinking code would weirdly enough allow clearing of extra/fake user
status on IDs used by affected ID, which would be utterly wrong.
Fairly unclear why this was working OK in reported case before rBa71a513def20,
could not spot any obvious reason just from reading code...
Also, in `libblock_remap_data_update_tags`, only transfer fake user
status if `new_id` is not NULL (otherwise that would have removed that
falg from `old_id`, without actually transferring it to anything).
This patch adds allows the user to select the initial fill color when
adding a new color attribute layer.
---
{F13035372}
Reviewed By: JulienKaspar, joeedh
Differential Revision: https://developer.blender.org/D14743
* BLENDER_VERSION_CYCLE set to beta
* Update pipeline_config.yaml to point to 3.2 branches and svn tags
* Update and uncomment BLENDER_VERSION in download.cmake
This commit introduce back support for all geometry types and all nodetree support.
Only the forward shading pipeline is implemented for now.
Vertex Displacement is automatically enabled for now.
Lighting & Shading is placeholder.
Related Task: T93220
# Conflicts:
# source/blender/draw/engines/eevee_next/eevee_engine.cc
# source/blender/gpu/CMakeLists.txt
This should have no functional changes.
This reduce the complexity of the shader by only supporting 2 colors.
We never use more than 2 color in practice and this makes usage not require
a UBO.
The old python importer had a "if do_transparency, set blend_method to
BLEND" type of logic. This bit was missing in the new importer; it was
only setting the eevee blend method when a transparency texture was
present, but not in other cases of transparency (as driven by MTL
"illum" mode).
Reviewd By: Howard Trickey
Differential Revision: https://developer.blender.org/D14783
Even if available OBJ/MTL format documentations don't explicitly specify
which characters can possibly separate keywords & arguments, turns out
some files out there in the wild use TAB character after the line
keywords. Which is something the new 3.2 importer was not quite
expecting (T97417).
Fix this by factoring out a utility function that checks if line starts
with a keyword followed by any whitespace, and using that across the
importer. Also fix some other "possible whitespace around name-like
parts" of obj/mtl parser as pointed out by the repro files in T97417.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14782
- Fix default_value initialization of custom node tree interface:
This was crashing when adding a custom interface socket to a tree.
The node_socket_set_typeinfo function was called too early, creating a
default float socket, which then doesn't match the socket type after
changing to the custom type.
The node_socket_set_typeinfo only allocates and initializes
default_value when it isn't already set. That is because the function is
used either when creating new sockets or to initialize typeinfo after
loading files. So default_value has to be either null or has to be
matching the current type already.
- Fix RNA flag for string return value of the valid_socket_type callback:
String return values of registerable RNA functions need a
PROP_THICK_WRAP flag since they don't have a fixed buffer to write into.
This should have (mostly) no functional changes.
Support for clipping was added to the shaders but it is not enabled for now
as we wait for the Gpencil engine to support clipping.
Simple port.
Also a description of how each shader is used has been added.
NOTE: The shader created using `OVERLAY_shader_paint_vertcol` cannot be tested.
Apparently it is created but not used.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D14812
The Multiscatter GGX code was missing the same-side checks for incoming and
outgoing directions w.r.t. to shading and geometry normal.
Should not be needed for the Glass variant since it intentionally has both
reflection and transmission.
This allows object extras such as image-empties to be shown in the VR
viewport/headset display. Being able to see reference images in VR can
be useful for architectural walkthroughs and 3D modeling applications.
Since users may not want to see all object extras (lights, cameras,
etc.), per-object-type visibility settings are also added as session
options.
By slightly refactoring the definition of the 3D View object types
visibility panel (note: no functional changes), the VR Scene Inspection
add-on can show a similar panel without duplicating code. When VR
selection is possible in the future, the object type select options can
also be enabled.
Reviewed By: Severin
Differential Revision: https://developer.blender.org/D14220
A change proposed in T97697, using existing saturation as a multiplier for the filter operation
{F13038602}
Review By: Joseph Eagar, Julian Kaspar
Differential Revision: https://developer.blender.org/D14808
Ref D14808
Add "stageMetersPerUnit" render setting for USD files that have that set to
something other than the default (e.g. exported by Blender).
And fix a crash when an application creates a Hydra render pass on a thread
that does not have an OpenGL context current.
Long term, this should replace the XML format. This reuses the Hydra render
delegate implementation, and so supports the same features. The same command
line options and GUI work for both XML and USD also.
The implementation of this is still disabled, waiting for some refactoring of
USD library linking. However we want the Cycles code to be in sync between
repositories for the 3.2 release.
Ref T96731
* Leave code for building the render delegate against other applications and
their USD libraries to the Cycles repository, since this is not a great fit.
In the Blender repository, always use Blender's USD libraries now that they
include Hydra support.
* Hide non-USD symbols from the hdCycles shared library, to avoid library
version conflicts.
* Share Apple framework linking between the standalone app and plugin.
* Add cycles_hydra module, to be shared between the standalone app and plugin.
* Bring external libs code in sync with standalone repo, adding various missing
libraries.
* Move some cmake include directories to the top level cycles source folder
because we need to control their global order, to ensure we link against the
correct headers with mixed Blender libraries and external USD libraries.
In the current code we do not render any curves if they have not been
converted to meshes. This change makes the custom bone drawing try to
render mesh objects first and then falls back to curve objects if there
is no mesh data available.
Reviewed By: Clement
Differential Revision: http://developer.blender.org/D14804
While the reference would never be used in case of NULL pointer, this
bit of code was not really clear and nice, so make it less ambiguous
now. Also add early return in case a NULL idv pointeris actually passed.
While code deleting old (relocated to new ones) IDs would work fine in
typical cases, it would fail badly in others, when e.g. drivers would
create 'reversed' dependency from the obdata ID to the object ID.
This commit uses a less efficient, but much safer method. It also
ensures no relocated old IDs is left over in the file (previous version
could easily leave some old IDs from the old library until a full
save/reload cycle happened).
While relatively minor issue, it could become fairly annoying in a big
project, since once an ID is tagged as directly linked it tend to remain
as such.
Now also force indirect tagging when loading new IDs as part of a lib
relocation process, since the ones actually directly linked will be
tagged as such later on.
While this only had minor potential effect, both code incrementing
usercount of newly remapped IDs were wrong.
Original one would by-pass any 'ensured user' handling, newer one would
systematically make the ID directly linked...
`id_us_plus_no_lib` is to be used here.
Fact that those options are only used in a specific case, and that the
same behavior is ensured in a different part of the code in other cases,
is fairly confusing and unfortunate... At least document it.
This was caused by the compilation job being created suspended (to avoid
UI slowdown because of the material Preview Icons). The suspended job
wasn't passing the `WM_jobs_test` in `DRW_deferred_shader_remove` and
the material would still be in the compile queue with its status equal to
`GPU_MAT_QUEUED`. This would block the main thread in the waiting loop.
But since the job manager timer needs to execute in the main thread, the
compilation job was never being pushed out of its suspended state.
This lead to a complete lock situation.
The solution is to use `WM_jobs_customdata_from_type` which does exactly
what we need.
Also fixed a nullptr free.
* Float/double promotion warnings were mainly meant for avoiding slow
operatiosn in the kernel. Limit it to that to avoid hard to fix warnings
in Hydra.
* Const warnings in Hydra iterators.
* Unused variable warnings when building without glog.
* Wrong camera enum comparisons in assert.
* PASS_UNUSED is not a pass type, only for pass offsets.
There were two calls to access job's custom data. One of them
ignored job type, the other one ignored job owner.
Now there is a single function to access job's custom data.
If the job type or owner is not relevant NULL or WM_JOB_TYPE_ANY
can be passes explicitly.
Differential Revision: https://developer.blender.org/D14803
When pushing down an Action onto an NLA track, set the new Strip's
influence to the Action's influence. This is done by setting a key due
to the way the NLA Strip influence works (it's either animated, or
ignored).
Reviewed By: sybren, RiggingDojo
Differential Revision: https://developer.blender.org/D14719
For a single day in 2015 between rBff3d535bc2a6309 and rB945f32e66d6ada,
custom data structs could be written with an incorrect maxlayer field.
This means that custom data structs read from those files would think
they have more space to add new layers than they actually did, causing
a crash if more layers were added. This was found while investigating
a crash from D14365 which adds new face corner layers in versioning.
The fix is to reset all maxlayer integers to totlayer, which is
done when writing files in current Blender anyway.
The file tests/render/motion_blur/camera_zoom_blur_perspective.blend
has this problem as it was added on 2015-07-21, right between the two
commits. Adding three custom data layers in versioning code would crash.
The problem was originally found and investigated by Martijn Versteegh
(@Baardaap), thanks!
Differential Revision: https://developer.blender.org/D14786
Add the Discontinuity (Euler) Filter operator to the Dope Sheet->Key
menu, so it's not only available from the Graph Editor->Key menu. On
request of @pablico, see T95386.
This required changing a poll function which is used by a bunch of other
operators, which seemed scary at first, but my thinking is that if an
operator can execute in the Graph Editor, then it should also be able to
execute in the Dope Sheet. I think the only reason this wouldn't be true
is if we were storing animation data in the UI itself, which of course
we don't. So I hope this is okay.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D14015
Simple port with a few cosmetic changes:
- Attribute named "color" for indices VBO is now called "index"
- The indices VBO is now composed of `int`s instead of `uint`s (this simplifies the source)
Differential Revision: https://developer.blender.org/D14800
This patch has also been contributed upstream, so will not be needed anymore
soon. Also automatically clear cached variables for new nanovdb location in
libs.
NanoVDB is now bundled with OpenVDB (since rBb9c37608a9e) instead of a
separate package.
This still doesn't include our patch to support AMD HIP; that'll be
committed separately soon.
The bitwise XOR used to compute the delta (`changed`) might produce NaN and
thus produce undefined behavior when comparing to another float (because of
float promotion).
shader_builder had no manifest set, leading
to the classic common control version being
loaded which in turn caused an import error
and made the executable fail to initialize.
After rB47276b847017, for certain socket types such as object or
material, the name of the item is not obtained correctly leading
the tooltip to display random non-character memory values as text.
Differential Revision: https://developer.blender.org/D14762
When the object's position depends on the geometry and the geometry
depends on the object's position, we can't count on the object's
evaluated geometry to be available. Lattices and mesh objects have
equivalent checks in this vertex parenting function.
Differential Revision: https://developer.blender.org/D14781
Along with the port to createInfo this also:
- Packs constant uniforms in a UBO.
- Share enum declaration and unify names
- Makes codeflow easier to undestand.
- Split grid data to its own struct.
# Conflicts:
# source/blender/draw/engines/overlay/overlay_grid.c
For properties exposed to the geometry nodes modifier, decorators didn't
work at all for colors and it only worked on the X component of vectors.
The fix is to use -1 for the RNA index of the decorator button instead
of 1, which lets the UI code figure out what to do with arrays.
This patch modifies tooltips of attributes and UV maps to resolve
inconsistencies. It also restores the vertex color icon that went
missing from the UI lists when color attributes replaced vertex colors.
Fixes T97614
Differential Revision: https://developer.blender.org/D14768
Recent cleanups 9a8669ac81 and 1c790555a0
incorrectly interpereted the bitfield width syntax as a default
value. Also resolve two other compilation warnings.
Currently, the `eval` and `pdf` are not explicitly set to zero when a BSDF sample is invalid (e.g., below the upper hemisphere), when calling
`bsdf_sample` or `bsdf_eval`. It is assumed that `eval` and `pdf` are set to zero before these functions are called, which can cause problems if not.
This patch fixes this potential problem by explicitly setting `eval` and `pdf` to zero when the sampled direction is invalid.
I also added a sanity check if `eval` and `pdf` are valid (i.e., >= 0.f).
The check is activated when build in debug mode and with the `WITH_CYCLES_DEBUG` set to `ON`.
Reviewed By: brecht, sergey
Differential Revision: https://developer.blender.org/D14776
When converting from XYZ to RGB it can happen, in some sky models, that the resulting RGB values are negative.
Atm, this is not considered and the returned values for the sky model can be negative.
This patch clamps the returned RGB values to be `= 0.f`
Reviewed By: brecht, sergey
Differential Revision: https://developer.blender.org/D14777
Allow the session to save the choice for domain and data type since
it is highly likely that the user intends to use the same settings
for subsequent color attributes.
Use expanded button selectors for domain and data type since there
are only two options for each.
Differential Revision: https://developer.blender.org/D14785
A snap state can be replaced by another snap state of a gizmo or cursor.
The snap gizmo should only change its state, not the current state.
It's not really a problem currently.
D14686 added autopep8 which implicitly dragged in
toml and pycodestyle which were not versioned, this
diff adds explicit versions of these deps so there
won't be any version changes if we rebuild in the
future.
Reviewed By: brecht, sybren
Differential Revision: https://developer.blender.org/D14793
If the measure gizmo is enabled, its snap state is used in the snap update.
As it does not require a plane, the orientation of the plane is not
calculated.
However, the calculation of the plane's orientation must prevail over
the states.
`PXR_ENABLE_OSL_SUPPORT=OFF`: OpenShadingLanguage is an optional
dependency of the Imaging module. However, since that module was
included for its support for converting primitive shapes (sphere, cube,
etc.) to geometry, OSL is not necessary. Disabling it will make it
simpler to build Blender; currently only Cycles uses OSL.
`PXR_ENABLE_GL_SUPPORT=OFF`: GL support on Linux also links to X11
libraries. Enabling it would break headless or Wayland-only builds.
OpenGL support would be useful if someone wants to work on a Hydra
viewport in Blender; when that's actually being worked on, we could
patch in a new PXR_ENABLE_X11_SUPPORT option (to separate OpenGL from
X11) and contribute it upstream.
`PXR_BUILD_OPENIMAGEIO_PLUGIN=OFF`: It's used for loading image textures
in Hydra Storm / Embree renderers which we don't use.
Reviewed By: LazyDodo, brecht, makowalski
Differential Revision: https://developer.blender.org/D14792
Split 'ED_view3d_cursor_snap_data_get' into 'update' and 'get' functions
Sometimes we just want to update and sometimes we just get the result.
Make it clear.
When height is limited, it is defined by space occupied by strips,
but at least channels 1 to 7 will be always visible. This allows it to
easily overview timeline content by zooming out to maximum extent in Y
axis and panning in X axis.
More channels can be "created" on demand by moving strip to higher
channel. When strip is removed and highest channel becomes empty, view
will stay as is until it is moved down. Then new highest point is
remembered and it is not possible to pan upwards until strip is moved to
higher channel.
Limiting takes into account height of scrubbing and markers area as
well as scrollers. This means that when zoomed out to maximum extent,
no strips are obstructed by fixed UI element.
Fixes T57976
Reviewed By: Severin
Differential Revision: https://developer.blender.org/D14263
This complex comment was hard to parse visually.
There was some odd line breaking going on, and together with no
indentation for the continued lines, it was just a blob of text with no
visual structure. You wouldn't see easily where the description of an
argument started or ended.
Geometry node group inputs and outputs get a new property that controls
the attribute name used for that field input/output when assigning the
node group to a modifier for the first time. If the default name is assigned
to an input, the default "Use attribute name" is true .
In order to properly detect when a node group is first assigned,
the modifier now clears its properties when clearing the node group.
Ref T96707
Differential Revision: https://developer.blender.org/D14761
This patch adds the drag and drop strip previews in the VSE.
It also adds two new functions to the drag and drop API.
1. "draw_in_view" for callbacks that wants to draw elements in local viewport coordinates
2. "on_drag_start" that can be used for prefetching data only once at the start of the drag.
Reviewed By: Julian, Campbell
Differential Revision: http://developer.blender.org/D14560
When using an offset modifier on a grease pencil object, the performance
could be impacted due to the randomize option introduced by rB6a2bc40e0131.
Even if the option was not used (offset, rotation, scale set to zero), the modifier
would still compute the random transformation matrix.
The patch checks if the randomization is used and only then caluclates the matrix.
Reviewed By: #grease_pencil, antoniov
Differential Revision: https://developer.blender.org/D14553
As the default handle type in Blender is 'Auto Clamped' the Equalize
Handles operator will often appear to have no affect on the selected
keyframes on which it is run. If either of the keyframes' handle types
are 'Auto', 'Auto Clamped', or 'Vector', this patch will convert the
handles to 'Aligned'.
Reviewed By: sybren
Maniphest Tasks: T96476
Differential Revision: https://developer.blender.org/D14345
Since rBfa43c47c7cb8446b632a4c0f712162ba615fe51f the progress bar do not
show the compilation progress. This was misleading as users could think
it could be canceled or prevent rendering.
Now we just show how many shaders are still in the compilation queue inside
each viewport. This number is more accurate than the percentage that was
previously displayed in the progress bar.
This uses refcounter instead of double thread mutexes. This should be
more robust and avoir use after free situation.
Also remove redundant structures and the use of scene as the job owner.
Removal of the `WM_redraw_windows` call in `wm_file_write` introduced
in rB7a9cfd08a8d7415ff004809cf62570be9152273e as that can cause
crashing while saving from a script.
See D14780 for more details.
Differential Revision: https://developer.blender.org/D14780
Reviewed by Campbell Barton
At least on the Mac, `std::sort` sometimes passes the same value in the
`a` and `b` parameters.
The `true` return is only for cases where `a` is less than `b`.
This is to avoid use after free when the `GPUPass` gets compiled after the
original `GPUMaterial` used to create it was freed.
The issue was introduced by rBfa3bd17ae873
The subdivision is always recomputed on the CPU when displaying stats
if the mesh is animated which leads to bad performance.
This caches the subdivision topology counters from the draw code in the
mesh runtime and uses them for the viewport statistics.
Differential Revision: https://developer.blender.org/D14774
The crash is caused as the subdivision wrapper does not have loop
normals, which are generally computed at the end of the modifier stack
evaluation via `mesh_calc_modifier_final_normals`. (Note that they are
initially computed, but deleted by the subdivision wrapper creation.)
This records in the mesh runtime whether loop normals should have been
computed and computes them alongside the subdivision wrapper.
Differential Revision: https://developer.blender.org/D14489
There are a number of shaders, most notably grid_frag.glsl, which rely on default assignments to uniform values within shaders. This is not currently supported by the shader uniform push model implemented for the Metal backend, wherein uniform updates are pushed as a singular block of data. Any default assignment would become over-written.
As such, adding assignments of these default values in the high-level, to ensure the correct value is written for all APIs. This likely impacts Vulkan push-constants as well.
Authored by Apple: Michael Parkin-White
Ref T96261
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D14555
This covers implementation of the GPUTexture abstraction for the Metal backend, with additional utility functionality as required.
Some components have been temporarily disabled pending dependencies on upcoming Metal backend components, and these will be addressed as the backend is fleshed out.
One core challenge addressed in the Metal backend is the requirement for read/update routines for textures. MTLBlitCommandEncoders offer a limited range of the full functionality provided by OpenGLs texture update and read functions such that a series of compute kernels have been implemented to provide advanced functionality such as data format conversion and partial/swizzled component updates.
This diff is provided in full, but if further division is required for purposes of code review, this can be done.
Authored by Apple: Michael Parkin-White
Ref T96261
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D14543
When more than one, consecutive, subdivision modifier is used on a Mesh,
the last subsurf modifier is used for GPU subdivision even though it
might be disabled. This is because retrieving the last subsurf modifier
in the draw code did not check whether the modifier was disabled or not.
To fix this, the session UUID of the modifier which delegated evaluation
to the GPU code is cached and used in the draw to select the right subsurf
modifier.
Differential Revision: https://developer.blender.org/D14488
This is a stripped down version of D14645 without the scene specialisation optimisations.
The two major changes in this patch are:
- Enables more aggressive inlining on Apple Silicon resulting in a 1.1x speedup and 10% reduction in spill, at the cost of longer pipeline build times
- Revival of shader binary archives through a new ShaderCache which is shared between MetalDevice instances using the same physical MTLDevice. This mitigates the extra compile times via explicit caching (rather than, as before, relying on the implicit system shader cache which can be purged without notice)
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D14763
Adds a column to the right in the Library Overrides Hierarchies view
mode to toggle editability of library overrides.
Note that making a library override non-editable currently involves
clearing all overridden properties. This is an arguable design choice,
we should probably at least warn the user before doing this.
Part of T95802.
Reviewed by: Bastien Montagne
Differential Revision: https://developer.blender.org/D14653
The code to compare buttons from the previous to the current frame, to
see if they match (an thus should keep the same state) was quite
generic, and didn't allow much flexibility/customization. For some
cases this isn't enough, and a more specific comparison is needed. Say
if some buttons don't actually store comparable data themselves, only
via the button context. This was the case in D14653.
It was possible to set context pointers for buttons via the layout, but
not for buttons in places where the layout system wasn't used (where
buttons are placed manually). This is needed for buttons in the
Outliner, see D14653.
/maxcpucount:1 and /m are the same option with the latter
one using all cores available, leading to the situation
where msbuild would start N side by side project builds
that all tried to use N cores as well. leading to severe
memory and compute starvation during the deps build.
Goals:
* Better high level control over where devirtualization occurs. There is always
a trade-off between performance and compile-time/binary-size.
* Simplify using array devirtualization.
* Better performance for cases where devirtualization wasn't used before.
Many geometry nodes accept fields as inputs. Internally, that means that the
execution functions have to accept so called "virtual arrays" as inputs. Those
can be e.g. actual arrays, just single values, or lazily computed arrays.
Due to these different possible virtual arrays implementations, access to
individual elements is slower than it would be if everything was just a normal
array (access does through a virtual function call). For more complex execution
functions, this overhead does not matter, but for small functions (like a simple
addition) it very much does. The virtual function call also prevents the compiler
from doing some optimizations (e.g. loop unrolling and inserting simd instructions).
The solution is to "devirtualize" the virtual arrays for small functions where the
overhead is measurable. Essentially, the function is generated many times with
different array types as input. Then there is a run-time dispatch that calls the
best implementation. We have been doing devirtualization in e.g. math nodes
for a long time already. This patch just generalizes the concept and makes it
easier to control. It also makes it easier to investigate the different trade-offs
when it comes to devirtualization.
Nodes that we've optimized using devirtualization before didn't get a speedup.
However, a couple of nodes are using devirtualization now, that didn't before.
Those got a 2-4x speedup in common cases.
* Map Range
* Random Value
* Switch
* Combine XYZ
Differential Revision: https://developer.blender.org/D14628
The new name is `Internal Dependencies`.
* Is more inline with the general goal of the panel.
* Can also show which external objects are used in the node tree in the future.
This name was choosen in the geometry nodes submodule meeting (2022-04-25).
Differential Revision: https://developer.blender.org/D14752
This macro allows defining a default argument for when the translation
unit is compiled in C++. Otherwise (in C), the argument has to be passed
explicitly.
A couple of benefits:
* Default arguments are a nice quality-of-life feature in C++. It's
annoying if these can't be used in C++ files, just because the header
with the function declaration still needs to be C compatible.
* Adds useful information to the API declaration. E.g. that an argument
can be nullptr.
* Should help us to move to using default arguments more, helping
readability (arguably)
Used in D14653.
Reviewed By: LazyDodo
Differential Revision: https://developer.blender.org/D14654
This commit adds support for the curves object to the apply modifier
operator. A warning is added when the evaluated result of the modifier
doesn't contain any curves data.
Differential Revision: https://developer.blender.org/D14730
Motion paths can now be initialised to more sensible frame ranges,
rather than simply 1-250:
- Scene Frame Range
- Selected Keyframes
- All Keyframes
Reviewed By: sybren, looch, dfelinto, pablico
Maniphest Tasks: T93047
Differential Revision: https://developer.blender.org/D13687
Add missing check as the context override dict may have been copied
since it was assigned, also initialize the context manager with
PyType_Ready, while it didn't cause any errors - it's expected
that all types are initialized.
Aggressive needs to be enabled for many useful edits to take effect,
the ignore list is currently used to disable edits that may need
further checks/investigation before they're enabled.
Note that aggressive was already enabled for the:
source/tools/utils/autopep8_clean.py script which was previously used
when applying autopep8.
Failure to clip automatic-names meant named could end with a "." for e.g.
Error in [0] meant the clipped text was copied then
immediately overwritten.
[0]: 354e6b9c18
- De-duplicate txt_new_linen & txt_new_line.
- Don't accept NULL as input for txt_new_linen & txt_new_line
(callers can pass in an empty string instead).
- Avoid duplicate strlen calls in txt_new_linen.
- Use memcpy when the length of the string is known in txt_delete_sel &
BKE_text_load_ex.
`attribute_try_create` didn't understand that the vertex group
attribute already existed because it only looks for names in custom
data layers when the domain matches. Using `attribute_exists`
unfortunately requires another loop through all attribute names,
but that should be optimized separately in the future anyway.
Differential Revision: https://developer.blender.org/D14756
Added a fallback path to compute the
cursor radius for when the stroke
starts over a blank area of space
(in which case SCULPT_cursor_geometry_update
fails).
This commit makes the `MeshVertex.normal` property read-only.
In practice it already was, since the value could be overwritten at any
time if code requests updated normals, so this just clarifies that
position and avoids misleading users.
Differential Revision: https://developer.blender.org/D14696
auto-iteration property in mask filter
Note: Auto-iteration is still set manually
for increase/decrease contrast. These should
probably become their own operators.
by showing redo panel.
The A hotkey has "auto iteration" enabled by default,
which calculates the number of times to run the filter
using a heuristic based on vertex count.
To make clear to the user what is going on the redo
panel is now shown for the mask filter operator.
NOTE: I discovered the source of the bug where sculpt
operators' redo panels were greyed out. The name
fed to SCULPT_undo_push_begin must match the operator
name. I've added a comment in sculpt_intern explaining
this.
Remembering the number of curves of every type makes it fast to know
whether processing specific to a single curve type has to be done.
This information was accessed in quite a few places, so this should be
an overall reduction in overhead for the new curves type.
The cache is computed eagerly, in other words every time after changing
the curve types. In order to reduce verbosity I added helper functions
for some common ways to set the types.
Differential Revision: https://developer.blender.org/D14732
Changes:
- Remove `BLI_memarena` (Use `MEM_cnew` and `MEM_delete` to allocate cached data)
- Implement `snap_object_data_mesh_free_ensure` and `snap_object_data_editmesh_free_ensure` and skip need to get original key Object for editmesh data
- Use `BMEditMesh` as key for editmesh `Ghash`
- Make a better distinction between `SnapObjectData`s. (`SnapData_Mesh` and `SnapData_EditMesh`)
The editing data of a `SURF`s is similar to that of Curves and should be supported for snapping.
But unlike Curve objects, for snapping, only support the nurb points if the object is in edit mode.
This matches the solution for Meshes and avoids having to create a kind
of "boundbox" for the SURF nurb points.
* Adjust width based on node width, necessary to make the longer name below work.
* Show "X Named Attributes" in the overlay.
* Use "Accessed named attributes" in the tooltip.
Differential Revision: https://developer.blender.org/D14751
* If removing an attribute failed (even though it exists), don't log it as being used.
* Make warning message a bit more informative.
Differential Revision: https://developer.blender.org/D14750
View moves faster with two active directions.
This is probably because `dvec` is not normalized when moving in two directions.
Normalizing this direction vector will fix the problem.
Differential Revision: https://developer.blender.org/D14729
Since cbf033c055 the `matte_id` will be allocated in the node
storage for the forward compatibility purposes. However, this
field was never freed, leading to memory leak.
Causes annoying popup on macOS when running Cycles tests,
for example render_passes_cryptomatte_asset.blend
Differential Revision: https://developer.blender.org/D14728
Add Bake Action to the NLA edit menu to aid discoverablity and allow
people to understand that Bake Action is part of working with the NLA.
Part of the NLA road map improvement project for the Animation Module.
This was a community request to add access to the Bake without needing
to turn on developer tools in the preferences and then use search in the
NLA for bake.
It seems this was always intended, as the operator is called `nla.bake`.
Reviewed By: sybren
Differential Revision: https://developer.blender.org/D14575
- Add missing doxy-section for Apply Parent Inverse Operator
- Use identity for None comparison in Python.
- Remove newline from operator doc-strings.
- Use '*' prefix multi-line C comment blocks.
- Separate filenames from doc-strings.
- Remove break after return.
While the ability to run `autopep8 .` to format Blender's Python code
is handy, the settings to perform this can conflict with other uses
of autopep8 (which editors may use to auto-format on save).
Some arguments now need to be passed in, e.g:
autopep8 . --in-place --recursive --jobs=0
We'll likely include this in `make format` convenience target so the
details for invoking autopep8 shouldn't be an issue in the long term.
The current code for computing tangents is not exactly fast.
This has been a long-standing issue, and recently came up again with T97378.
The main bottleneck is fetching the mesh data, since it's handled through a callback system and each vertex might have its data queried dozens of times.
I've tried a lot of things to optimize `mikktspace.c`, but unfortunately most weren't that useful:
- Vectorizing SVec3 gives a ~5% speedup, but I'm not sure if the additional ~70 lines of code are worth it
- Keeping an internal copy of the data instead of re-querying all the time helps a lot (~50-60% time reduction), but requires a lot of extra memory (~100 byte per face)
- Going C++ and replacing the internal quicksort with std::sort shows no difference
- Restructuring the entire file to be a header-only library so that the callbacks can be inlined gives ~10% reduction, but is a major change and deviation from the original library
In the end, two simple fixes that actually help remain:
- Don't re-query the number of faces in each loop iteration
- Don't bother looking for identical vertices if there's only one vertex with that hash
With this, time for the test case in T97378 goes from 6.64sec to 4.92sec. It's something I guess.
I feel like completely refactoring this library would not be a bad idea at some point, but for now it does the job...
Differential Revision: https://developer.blender.org/D14675
The uv fix just submitted had a bug where I forgot to wrap around
after adding 1. This apparently worked anyway in a debug build
but not in release build, hence the buildbot tests were failing.
Works for both Cycles and multires bake. Triangles are baked to multiple
UDIM images if they span across them, though such UV layouts are generally
discouraged as there is no filtering across UDIM tiles.
The bake margin currently only works within UDIM tiles. For the extend method
this is logical, for the adjacent faces method it may be useful to support
copying pixels from other UDIM tiles, though this seems somewhat complicated.
Fixes T95190
Ref T72390
Remove need for shadow caustic caster geometry to have a UV layout. UVs were
useful to maintain a consistent tangent frame across the surface while
performing the walk. A consistent tangent frame is necessary for rough
surfaces where a normal offset encodes the sampled h, which should point
towards the same direction across the mesh.
In order to get a continuous surface parametrization without UVs, the
technique described in this paper was implemented:
"The Natural-Constraint Representation of the Path Space for Efficient
Light Transport Simulation" (Supplementary Material), SIGGRAPH 2014.
In addition to implementing this feature:
* Shadow caustic casters without smooth normals are now ignored (triggered
some refactoring and cleaning).
* Hit point calculation was refactored using existing utils functions,
simplifying the code.
* The max number of solver iterations was reduced to 32, a solution is
usually found by then.
* Added generalized geometry term clamping (transfer matrix calculation can
sometimes get unstable).
* Add stop condition to Newton solver for more consistent CPU and GPU result.
* Add support for multi scatter GGX refraction.
Fixes T96990, T96991
Ref T94120
Differential Revision: https://developer.blender.org/D14623
This is fairly tricky to perform, since there is often very limited
contextual information available about the 'active' hierarchy to
override.
This commit is a first step, it is expected to handle decently well
cases like objects and obdata (recreating necessary object and
collection hierarchy, and/or hooking it to a potential existing
hierarchy), at least in most common cases.
Ref: {T95707}.
Extends the changes started in f31c3f8114 to completely separate
much of the DRW curves code from the particle hair drawing. In the short
term this increases duplication, but the idea is to simplify development
by making it easier to do larger changes to the new code, and the new
system will replace the particle hair at some point.
After this, only the shaders themselves are shared.
Differential Revision: https://developer.blender.org/D14699
Add macros to get/set boolean attributes, to set float2/float3
attributes and to get float2/float3 attributes via pointer access.
Needed for D14365 and further generic attribute integration.
Differential Revision: https://developer.blender.org/D14708
This substantially redoes the logic by which bevel chooses, for
the middle segment when there are an odd number of segments,
which face to interpolate in, and which vertices to snap to which
edges before doing that interpolation. It changes the UV layouts
of a number of the regression tests, for the better.
An example, in the reference bug, is a cube with all seams, unwrapped
and then packed with some margin around them, now looks much
better in UV space when there are an odd number of segments.
The original mistake I made in b9febb54a4 was thinking
that the input curve object data to `BKE_displist_make_curveTypes`
was already copied from the original. I think I misread some of its
`ID` flags. This commit places the result of curves evaluation in a
duplicated curve instead, and copies the edit mode pointers
necessary for drawing overlays. `Curve` needs to know not to
free those pointers.
I still don't have a full understanding of why some of the tactics I've
used work and others don't. I've probably tried around 8 different
solutions at this point, and this is the best I came up with.
The dependency graph seems to have some handling of edit mode
pointers that make the edit mode overlays work if the evaluated
result is only an empty curve created by the evaluated geometry set.
This doesn't work with the current method and I need to set the
edit mode pointers at the end of evaluation explicitly.
We're constrained by the confusing duality of the old curves system
combined with the new design using the evaluated geometry set.
Older areas of Blender expect the evaluated `Curve` to be a copy
of the original, even if it was replaced by some arbitrary evaluated mesh.
Differential Revision: https://developer.blender.org/D14561
Added some regex magic in i18n py module to also extract UI names from
all of our units definitions.
Those enum values are fully dynamically generated, so they cannot be
extracted from RNA introspection.
Caused by {rBcfa53e0fbeed}
Above commit mixed up source and destination meshes causing bad lookups
on calculated normals.
Now make sure we get normals from our destination mesh to project along.
Note this was only reported for Projected Face Interpolated mode, but
same was true for Projected Edge Interpolated mode (though that one is a
bit weird to test since I think there is generally something wrong with
that mode -- with or without rBcfa53e0fbeed).
Fixes T97528
Maniphest Tasks: T97528
Differential Revision: https://developer.blender.org/D14726
When in mesh editmode, attributes point to bmesh customdata, the
attribute data is empty since custom data is stored per element instead
of a single array there (same es UVs etc.).
Opposed to e.g. UVs, general attributes were not setting their data
length/size to zero in case of editmode though, which could lead to
- crash in Outliner Data Api view [that was reported in T95922]
- RuntimeError such as the following:
```
RuntimeError: bpy_prop_collection[index]: internal error, valid index 0
given in 8 sized collection, but value not found
```
Now check for mesh editmode in `BKE_id_attribute_data_length` (and
return zero in that case).
Alternatively, the check could also be done in
`rna_Attribute_data_length` only (such as UVs do in
`rna_MeshUVLoopLayer_data_length`).
Ref D11998
Fixes T95922
Maniphest Tasks: T95922
Differential Revision: https://developer.blender.org/D14714
Regression in 4d0f846b93,
passing selection through to drag relied on tweak events running
even when the press event was handled which is not the case for drag.
Fix for T84962
Before the patch, edit voxel size always displayed voxel size without units, just as a number in meters.
Now it changes like in the voxel remesh panel and shows correct units
Video:
{F13009428}
In adaptive mode:
{F13009435}
Reviewed By: JulienKaspar
Maniphest Tasks: T84962
Differential Revision: https://developer.blender.org/D14682
This adds pyproject.toml, needed to configure defaults for autopep8.
The file is auto-discovered in a similar way to .clang-format, other
tools could be configured here too. For now just configure autopep8 so
this can be enabled in IDE's without causing unexpected edits such as
wrapping lines over 80 columns in width.
Now autopep8 can be used from the root directory by running: autopep8 .
This uses multiple-jobs to run autopep8 over all Python scripts except
paths that are explicitly ignored in exclude defined by pyproject.toml.
Reviewed By: mont29, brecht, sybren
Ref D14686
This enables building of HIP binaries for AMD RDNA and RDNA2 GPUs.
This requires the 22.10 / ROCm 5.1 driver.
Ref T91571
Differential Revision: https://developer.blender.org/D14360
If the attribute already existed, but had a different domain or data type
than the user tried to write to it with (i.e. writing to `position` with
a float), then the data from field evaluation wasn't freed.
This restructures the geometry nodes modifier attribute writing process
to solve that problem and remove some of the nested if statements
that made the process confusing.
One case that still doesn't work is writing to a builtin attribute from
a different domain. If `OutputAttribute` gets that feature then that
could be supported here too.
Differential Revision: https://developer.blender.org/D14706
The problem is old.
rB52be06301257 (fixed by rB4b35d6950d4f) just masked it.
`Object->data`, on evaluated objects, is not a safe pointer to get
objects with the same `BMEditData`.
Use `Object->runtime.data_orig` instead.
As `GPU_PRIM_LINE_LOOP` is not supported on Vulkan or Metal and
`GPU_PRIM_TRI_FAN` is not supported on Metal, they will be removed in
future releases.
So it is important to inform users that they are obsolete and may not
be supported for a long time.
Release Notes: https://wiki.blender.org/wiki/Reference/Release_Notes/3.2/Python_API
Reviewed By: campbellbarton
Differential Revision: https://developer.blender.org/D14679
Property collection functions were not really in sync which could result
in crashes in UI_LISTS.
The reason is that attributes of type color defined on domains other than
Points or Corners are still valid, but should really be skipped for the
special "color_attributes".
`rna_AttributeGroup_color_length` itself was fine here, it skips these,
but the iterator (`rna_AttributeGroup_color_iterator_begin` /
`rna_Attributes_noncolor_layer_skip`) wasnt, so when a UI_LIST
filter_items() would iterate the collection it would actually get
results were it shouldnt.
Now check a suiting domain in `rna_Attributes_noncolor_layer_skip` as
well.
Maniphest Tasks: T97502
Differential Revision: https://developer.blender.org/D14717
Since {rBeae36be372a6b16ee3e76eff0485a47da4f3c230} the distinction
between float and byte colors is more explicit in the ui. So far, geometry
nodes couldn't really deal with byte colors in general. This patch fixes that.
There is still only one color socket, which contains float colors. Conversion
to and from byte colors is done when read from or writing to attributes.
* Support writing to byte color attributes in Store Named Attribute node.
* Support converting to/from byte color in attribute conversion operator.
* Support propagating byte color attributes.
* Add all the implicit conversions from byte colors to the other types.
* Display byte colors as integers in spreadsheet.
Differential Revision: https://developer.blender.org/D14705
This adds a new subpanel to the geometry nodes modifier which is just
used to display information about used attributes.
* A new panel is used because adding this information anywhere else
clutters the ui too much imo.
* The general layout is similar to that in the tooltip. I found it to be more
trouble than it's worth to share this code.
Possible future improvements:
* Don't show the panel if there are no used named attributes.
* Add some heuristics to determine which named attributes the user does
not have to care about because they are only used in the node group
and don't affect anything else.
Differential Revision: https://developer.blender.org/D14701
Very easy fix, the bug seemed to be a result of a typo on the right-most margin.
Old: {F13013777}
New: {F13013782}
Maniphest Tasks: T97497
Differential Revision: https://developer.blender.org/D14711
This was projecting the unnormalized z scale axis onto the plane defined
by the view vector. If object scale was very small, this made the empty
still visible at viewing angles far from the object axis.
Now use the normalized z scale axis to make this work the same at all
object scales.
Fixes T97004.
Maniphest Tasks: T97004
Differential Revision: https://developer.blender.org/D14557
This was possible with the gizmo, but no using the operator (e.g. from
the sidebar).
Now store original offsets and reset to these on calcel (ESC or RMB).
Maniphest Tasks: T97465
Differential Revision: https://developer.blender.org/D14704
Text inserted via TEXT_OT_insert would always have auto-close
logic applies which interfered any insertion of literal strings
containing brackets.
Now auto-closing brackets is restricted to characters read from
events from the operators invoke functions.
Sculpt paint tools now pop up an error message if
dynamic topology or multires are enabled.
Implementation notes:
* SCULPT_vertex_colors_poll is now a static function in sculpt_ops.c.
It is now used solely by the legacy color attribute conversion
operators (SCULPT_OT_vertex_to_loop_colors and SCULPT_OT_loop_to_vertex_colors)
and should be deleted when they are.
* There is a new method, SCULPT_handles_colors_report, that returns true if
the sculpt session can handle color attributes; otherwise it returns false
and displays an error message to the user.
- Verrtex paint mode has been refactored into C++ templates.
It now works with both byte and float colors and point
& corner attribute domains.
- There is a new API for mixing colors (also based
on C++ templates). Unlike the existing APIs byte
and float colors are interpolated identically.
Interpolation does happen in a squared rgb space,
this may be changed in the future.
- Vertex paint now uses the sculpt undo system.
Reviewed By: Brecht Van Lommel.
Differential Revision: https://developer.blender.org/D14179
Ref D14179
Python 3.10 has added "soft keywords" [0] to their list of identifiers.
This patch adds these soft keywords to the list of builtin functions
that the text editor searches for when highlighting Python code.
The only soft keywords that Python 3.10 current has are: `match`,
`case`, and `_`, but more will likely be added in future versions.
Currently, the `_` soft keyword is ignored from highlighting. It is a
wildcard matching pattern when used with `case` (similar to `default`
for `switch`es in C/C++), but `_` is far more often used in other
contexts where highlighting the `_` might seem strange. For example,
ignoring elements when unpacking tuples (`_, g, _, a = color`).
This patch also updates the commented Python code for creating the list
of keywords, for convenience.
Before:
{F13012878}
After:
{F13012880}
Example from PEP-636 [1]
Note: These soft keywords are only reserved under specific contexts.
However, in order for the text editor to know when the keywords are used
in the appropriate contexts, the text editor would need a full-blown
Python grammar [2] parser. So, for now, these keywords are simply added
in along with the other keywords in order to highlight them in the text
editor.
[0]: https://docs.python.org/3/reference/lexical_analysis.html#soft-keywords
[1]: https://peps.python.org/pep-0636/#matching-specific-values
[2]: https://docs.python.org/3/reference/grammar.html
Ref D14707
Currently the "exclude" option for autopep8 isn't as reliable as it
should be since passing in absolute paths to autopep8 causes
the paths not to match. See: D14686 for details.
So explicitly disable autopep8 in this generated file (the generator
has already been updated).
Also use spaces for indentation otherwise autopep8 re-indents them.
This seems like a bug in autopep8 since it's changing lines with
autopep8 disabled. Use a workaround instead of looking into a fix since
it's simpler for all our Python files to use spaces instead of tabs and
there isn't much benefit mixing indentation for scripts.
Sometimes, when moving strip with 2 input effect attached and causing
strip overlap, this overlap is resolved incorrectly - too big offset
is applied. This is because effects are not taken into consideration
when "shuffeling" to resolve overlap.
To fix usual cases (transitions), overlap between strip and it's effect
is considered to be impossible.
There are edge cases, but these would be much more complicated to
implement and could have negative impact on performance.
Reviewed By: ISS
Differential Revision: https://developer.blender.org/D14578
Crash was caused by incorrectly assuming, that rendered strip is meta
when editing code in bulk, but this wasn't true and strip had no
channels initialized, which caused crash on NULL dereference.
Correct approach is to find list of channels where on same level as
rendered strip, similar to how `SEQ_get_seqbase_by_seq` function works
for list of strips.
This reverts commit 390b9f1305. It seems to
break things on Linux for unknown reasons, so leave it out for now. A solution
to this will be required for Vega cards though.
There is not much point in having those editable in overrides, and since
those are pointers, their value always differs from ref linked ID vs.
local override one, generating 'noise' in Outliner's override property
view.
Building against the existing 3.1 libraries should continue to work, until
the precompiled libraries are committed for all platforms.
* Enable WebP by default.
* Update Windows for new library file names.
* Automatically clear outdated CMake cache variables when upgrading to new
libraries.
* Fix static library linking order issues on Linux for OpenEXR and OpenVDB.
Implemented by Ray Molenkamp, Sybren Stüvel and Brecht Van Lommel.
Ref T95206
The "PROP" in the name reflects its generic status, and removing
"LOOP" makes sense because it is no longer associated with just
mesh face corners. In general the goal is to remove extra semantic
meaning from the custom data types.
For some reason, the rework of liboverride handling of Collection items
insertion (rB33c5e7bcd5e5) completely missed to update accordingly the
default liboverride apply code...
Many thanks to Wayde Moss (@GuiltyGhost) for the investigation and
proposed solution.
This was caused by the use of a reserved keyword macro that is not
directly used but causes an error on some compiler.
Change the occurences to not match the macros.
This file was skipped by source/tools/utils/autopep8_clean.py
since it doesn't have a .py extension, running the autopep8 tool
recursively detects Python scripts without extensions.
- Increase the stack level so the reported line number references
script authors code (not Blender's wrapper function).
- Include the operator name and poll/call usage in the warning.
Support a way to temporarily override the context from Python.
- Added method `Context.temp_override` context manager.
- Special support for windowing variables "window", "area" and "region",
other context members such as "active_object".
- Nesting context overrides is supported.
- Previous windowing members are restored when the context exists unless
they have been removed.
- Overriding context members by passing a dictionary into operators in
`bpy.ops` has been deprecated and warns when used.
This allows the window in a newly loaded file to be used, see: T92464
Reviewed by: mont29
Ref D13126
These functions can be used with PyArg_ParseTupleAndKeywords
(and related functions) to take typed RNA arguments without
having to extract and type-check them separately.
No functional changes, extracted from D13126.
There's a small typo in the tool tip for applying the Parent Inverse. This patch fixes that typo
old:
{F13010751}
new:
{F13010749}
Reviewed By: Blendify
Maniphest Tasks: T97437
Differential Revision: https://developer.blender.org/D14693
This is mostly a cleanup to avoid hardcoding the eager calculation of
normals it isn't necessary, by reducing calls to `BKE_mesh_calc_normals`
and by removing calls to `BKE_mesh_normals_tag_dirty` when the mesh
is newly created and already has dirty normals anyway. This reduces
boilerplate code and makes the "dirty by default" state more clear.
Any regressions from this commit should be easy to fix, though the
lazy calculation is solid enough that none are expected.
This adds support for rendering motion blur for volumes, using their
velocity field. This works for fluid simulations and imported VDB
volumes. For the latter, the name of the velocity field can be set per
volume object, with automatic detection of velocity fields that are
split into 3 scalar grids.
A new parameter is also added to scale velocity for more artistic control.
Like for Alembic and USD caches, a parameter to set the unit of time in
which the velocity vectors are expressed is also added. For Blender gas
simulations, the velocity unit should always be in seconds, so this is
only exposed for volume objects which may come from external OpenVDB
files.
These parameters are available under the `Render` panels for the fluid
domain and the volume object data properties respectively.
Credits: kernel advection code from Tangent Animation's Blackbird based
on earlier work by Geraldine Chua
Differential Revision: https://developer.blender.org/D14629
The fix ensures that the reference count for `IShellItem *pSI` is decremented,
preventing a memory leak. For `IFileOperation *pfo` the decrement of the
reference count is only attempted when `CoCreateInstance` is successful.
Additionally, the gotos have been replaced with nested if/else statements.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D14681
Since shader sources are now parsed on demand via `GPUShaderCreateInfo`,
sources are not available to be read via
`GPU_shader_get_builtin_shader_code`.
Currently this results in a crash as the code tries to read `NULL`
pointers.
`GPU_shader_get_builtin_shader_code` was created with the intention of
informing the user how a builtin shader works, thus "replacing"
detailed documentation.
Therefore this function doesn't really have a practical use in an addon.
So, instead of updating the function (which would require several
changes to the gpu module), remove it and improve the documentation.
Release Notes: https://wiki.blender.org/wiki/Reference/Release_Notes/3.2/Python_API#Breaking_Changes
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D14678
Noted as part of T94775 investigation by Wayde Moss (@GuiltyGhost),
thanks!
NOTE: this mistake probably did not have any pratical impact in current
code, at least for overrides.
This is to make the codegen and shading nodes object type agnostic. This
is essential for flexibility of the engine to use the nodetree as it see
fits.
The essential volume attributes struct properties are moved to the
`GPUMaterialAttribute` which see its final input name set on creation.
The binding process is centralized into `draw_volume.cc` to avoid
duplicating the code between multiple engines. It mimics the hair attributes
process.
Volume object grid transforms and other per object uniforms are packed into
one UBO per object. The grid transform is now based on object which simplify
the matrix preparations.
This also gets rid of the double transforms and use object info orco factors
for volume objects.
Tagging @brecht because he did the initial implementation of Volume Grids.
Noticed while looking into oneAPI patch.
Seems to be unused, without clear indication why/when it might be
needed. Removing the function simplifies adding the new backend.
Differential Revision: https://developer.blender.org/D14652
Continuing the refactors described in T93602, this commit moves
the face dot tag set by the subdivision surface modifier out of
`MVert` to `MeshRuntime`. This clarifies its status as runtime data
and allows further refactoring of mesh positions in the future.
Before, `BKE_modifiers_uses_subsurf_facedots` was used to check
whether subsurf face dots should be drawn, but now we can just check
if the tags exist on the mesh. Modifiers that create new new geometry
or modify topology will already remove the array by clearing mesh
runtime data.
Differential Revision: https://developer.blender.org/D14680
The following methods weren't included in API docs.
- BlendDataLibraries.load
- BlendDataLibraries.write
- Text.region_as_string
- Text.region_from_string
This doesn't cause any functional change as the RNA property
of Bone wasn't overridden by the _GenericBone's property,
however the `children` property was documented twice, causing a warning.
The UI description for the `bpy.types.ActionFCurves.remove` was incorrect;
seemingly a copy-paste typo from the `rna_Action_groups_remove` function.
Reviewed By: sybren, Blendify
Differential Revision: https://developer.blender.org/D14659
Preserve multi socket link order when copying nodes or adding a new
group input sockets by linking directly to multi inputs from the group
input node's extension socket.
This is done by also copying the `multi_input_socket_index` when
the new links are created by copying existing or temporary links.
Reviewed By: Hans Goudey
Differential Revision: http://developer.blender.org/D14535
Keep the existing Rec.709 fit and convert to other colorspace if needed, it
seems accurate enough in practice, and keeps the same performance for the
default case.
Reimplement copy geometry node groups in C. The version implemented in
Python could also manually copy the animation data, but it's more
standard to do this with `BKE_id_copy_ex` and `LIB_ID_COPY_ACTIONS`.
Differential Revision: https://developer.blender.org/D14615
`rna_NodeSocket_refine` and `rna_Node_refine` take significant time
when building the `NodeTreeRef` acceleration data structure, but they
aren't used at all. This commit removes their eager calculation and
instead creates them on-demand in the `rna()` functions. They also
aren't inlined to avoid including `RNA_prototypes.h` in the header.
Differential Revision: https://developer.blender.org/D14674
Continued improvements to the new C++ based OBJ importer.
Performance: about 2x faster.
- Rungholt.obj (several meshes, 263MB file): Windows 12.7s -> 5.9s, Mac 7.7s -> 3.1s.
- Blender 3.0 splash (24k meshes, 2.4GB file): Windows 97.3s -> 53.6s, Mac 137.3s -> 80.0s.
- "Windows" is VS2022, AMD Ryzen 5950X (32 threads), "Mac" is Xcode/clang 13, M1Max (10 threads).
- Slightly reduced memory usage during import as well.
The performance gains are a combination of several things:
- Replacing `std::stof` / `std::stoi` with C++17 `from_chars`.
- Stop reading input file char-by-char using `std::getline`, and instead read in 64kb chunks, and parse from there (taking care of possibly handling lines split mid-way due to chunk boundaries).
- Removing abstractions for splitting a line by some char,
- Avoid tiny memory allocations: instead of storing a vector of polygon corners in each face, store all the corners in one big array, and per-face only store indices "where do corners start, and how many". Likewise, don't store full string names of material/group names for each face; only store indices into overall material/group names arrays.
- Stop always doing mesh validation, which is slow. Do it just like the Alembic importer does: only do validation if found some invalid faces during import, or if requested by the user via an import setting checkbox (which defaults to off).
- Stop doing "collection sync" for each object being added; instead do the collection sync right after creating all the objects.
Cleanup / Robustness:
This reworking of parser (see "removing abstractions" point above) means that all the functions that were in `parser_string_utils` file are gone, and replaced with different set of functions. However they are not OBJ specific, so as pointed out during review of the previous differential, they are now in `source/blender/io/common` library.
Added gtest coverage for said functions as well; something that was only indirectly covered by obj tests previously.
Rework of some bits of parsing made the parser actually better able to deal with invalid syntax. E.g. previously, if a face corner were a `/123` string, it would have incorrectly treated that as a vertex index (since it would get "hey that's one number" after splitting a string by a slash), instead of properly marking it as invalid syntax.
Added gtest coverage for .mtl parsing; something that was not covered by any tests at all previously.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14586
- Was not exporting "Poly" curves at all,
- Had a crash when a single object contains multiple curves of different types -- it had a check for "is this nurbs compatible?" only for the first curve, and then proceeded to treat the other curves as nurbs as well, without checking for validity.
Fixed both issues by doing the same logic as in the old python exporter:
- Poly curves are supported,
- Treat object as "nurbs compatible" only if all the curves within it are nurbs compatible.
Added test coverage in the gtest suite. While at it, made "all_curves" test use the "golden obj file template" style test, instead of a manually coded test that checks intermediate objects but does not check the final exported result.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14611
The new 3.1 OBJ exporter code had incorrect code to determine which vertex group a polygon belongs to -- for each vertex, it was only looking at the first vertex group it has, and not using the group weight either.
This 99% fixes T96824, but not 100% on the user's submitted mesh -- exactly two faces from that mesh get assigned a different group compared to the old exporter. Either choice is "correct" given that on these two faces there are two vertex groups with equal contribution. The old Python exporter was picking the group based on internal python group name map order, whereas the new C++ exporter is picking the group with the lowest index, in case of ties. I'm not sure if it's possible to fix this TBH, will have to wait until the importer is also C++.
While at it, the new vertex group calculation code was doing a lot of redundant work for each and every face (traversing group lists several times, allocating & freeing memory), so I fixed that. Exporting a 6-level subdivided Monkey mesh with 30 vertex groups was taking 810ms, now takes 330ms.
Reviewed By: Howard Trickey
Differential Revision: https://developer.blender.org/D14500
This adds a basic unit test to check USD has been correctly
build with imaging components to support building both with
the old and new libs, it automatically adds the test when it
detects a library with imaging enabled. (platform devs will
have to pay attention it runs the test to validate the libs
build correctly)
For future use in the code it also defines a USD_HAS_IMAGING
define one could check if we're building against an USD lib
that has it (just because we build/ship with it, doesn't
mean downstream builds will ship with it, so we'll have
to be a little pro-active there)
Reviewed By: sybren
Differential Revision:https://developer.blender.org/D14456
In some circumstances singular files with numbers in their name (like
turntable-1080p.png or frame-1042.png) might be detected as a UDIM.
The root cause in this particular instance was because `BKE_image_get_tile_info`
believed this file to be a tiled texture and replaced the filename with
a tokenized version of it. However, later on, the code inside `image_open_single`
did not believe it was tiled because only 1 file was detected and our
tiled textures require at least 2. This discrepancy lead to the broken
filename situation.
This was a regression since rB180b66ae8a1f as that introduced the
tokenization changes.
Differential Revision: https://developer.blender.org/D14667
Also, as a suggestion, this patch changes Mask By Color and Color Filter to be the same shade of green as paint and smear tool icons
{F12998856}
{F12998857}
{F12998858}
Reviewed By: Julian Kaspar & Joseph Eagar
Differential Revision: https://developer.blender.org/D14632
Ref D14632
If the `wpoly` vector was small, the `wpoly_new` pointer could point
to part of its inline buffer on the stack, which becomes invalid out of
that scope. Instead, store `wpoly_new` as a span, and assign it properly
from the moved vector.
The HIG mentions that redundant words like "Enables" or "Activates"
shouldn't be used for tooltips of boolean properties. In this case
"When checked" was the redundant language that was implied by
the checkbox itself-- convention is to just state what the property
does when it's on.
Also change a few conjugations to the imperative and simplify
wording slightly, in order to be more consistent with language
elsewhere in Blender, and to be a bit more direct.
Differential Revision: https://developer.blender.org/D14644
Introduced by my recent commit: {rB3acbe2d1e933}
Lead to crash when insert_keyframe_direct() was called. Keyframing
crashed for NLA special properties (influence, animated_time),
driven properties, etc.
This commit changes the Curve to Mesh node to work with `Curves`
instead of `CurveEval`. The change ends up basically completely
rewriting the node, since the different attribute storage means that
the decisions made previously don't make much sense anymore.
The main loops are now "for each attribute: for each curve combination"
rather than the other way around, with the goal of taking advantage
of the locality of curve attributes. This improvement is quite
noticeable with many small curves; I measured a 4-5x improvement
(around 4-5s to <1s) when converting millions of curves to tens of
millions of faces. I didn't obverse any change in performance compared
to 3.1 with fewer curves though.
The changes also solve an algorithmic flaw where any interpolated
attributes would be evaluated for every curve combination instead
of just once per curve. This can be a large improvement when there
are many profile curves.
The code relies heavily on a function `foreach_curve_combination`
which calculates some basic information about each combination and
calls a templated function. I made assumptions about unnecessary reads
being removed by compiler optimizations. For further performance
improvements in the future that might be an area to investigate.
Another might be using a "for a group of curves: for each attribute:
for each curve" pattern to increase the locality of memory access.
Differential Revision: https://developer.blender.org/D14642
This is a partial fix to the fact that rendering with EEVEE or other GL
render engines is currently blocking the whole UI when asking to redraw
a viewport.
This patch just bypasses the viewport bind (containing the Draw Context
lock) and the following drawing. There is an update tagging to not
loose a viewport update if there was one asked.
Other queries other than view redraw (such as selection depth drawing or
offscreen drawing) will still block the whole UI as they need immediate
data feedback.
Ping @Severin for the change in `WM_draw_region_viewport_bind()`.
I'm assuming this is not an issue because it's highly unlikely to
bring up this operator during rendering. But in this case, it would just
lock as usual.
The bypassing in `DRW_notify_view_update` might be a bit overparanoid.
The ported normal calculation from ceed37fc5c neglected to
use the tilt attribute to rotate the normals around the tangents.
This commit adds that behavior back, adding a new math header file
to avoid duplicating the rotation function for normalized axes.
Differential Revision: https://developer.blender.org/D14655
This patch contains an initial pixel extractor for PBVH and an initial paint brush implementation.
PBVH is an accelleration structure blender uses internally to speed up 3d painting operations.
At this moment it is extensively used by sculpt, vertex painting and weight painting.
For the 3d texturing brush we will be using the PBVH for texture painting.
Currently PBVH is organized to work on geometry (vertices, polygons and triangles).
For texture painting this should be extended it to use pixels.
{F12995467}
Screen recording has been done on a Mac Mini with a 6 core 3.3 GHZ Intel processor.
# Scope
This patch only contains an extending uv seams to fix uv seams. This is not actually we want, but was easy to add
to make the brush usable.
Pixels are places in the PBVH_Leaf nodes. We want to introduce a special node for pixels, but that will be done
in a separate patch to keep the code review small. This reduces the painting performance when using
low and medium poly assets.
In workbench textures aren't forced to be shown. For now use Material/Rendered view.
# Rasterization process
The rasterization process will generate the pixel information for a leaf node. In the future those
leaf nodes will be split up into multiple leaf nodes to increase the performance when there
isn't enough geometry. For this patch this was left out of scope.
In order to do so every polygon should be uniquely assigned to a leaf node.
For each leaf node
for each polygon
If polygon not assigned
assign polygon to node.
Polygons are to complicated to be used directly we have to split the polygons into triangles.
For each leaf node
for each polygon
extract triangles from polygon.
The list of triangles can be stored inside the leaf node. The list of polygons aren't needed anymore.
Each triangle has:
poly_index.
vert_indices
delta barycentric coordinate between x steps.
Each triangle is rasterized in rows. Sequential pixels (in uv space) are stored in a single structure.
image position
barycentric coordinate of the first pixel
number of pixels
triangle index inside the leaf node.
During the performed experiments we used a fairly simple rasterization process by
finding the UV bounds of an triangle and calculate the barycentric coordinates per
pixel inside the bounds. Even for complex models and huge images this process is
normally finished within 0.5 second. It could be that we want to change this algorithm
to reduce hickups when nodes are initialized during a stroke.
Reviewed By: brecht
Maniphest Tasks: T96710
Differential Revision: https://developer.blender.org/D14504
* Curves objects now support the geometry nodes modifier.
* It's possible to use the curves object with the Object Info node.
* The spreadsheet shows the curve data.
The main thing holding this back currently is that the drawing code
for the curves object is very incomplete. E.g. it resamples the curves
always in the end, which is not expected for curves in general.
Differential Revision: https://developer.blender.org/D14277
**Relevant to Artists:** This patch adds an option to the Parenting
menu, `Object (Keep Transform Without Inverse)`, and Apply menu, `Parent
Inverse`. The operators preserve the child's world transform without
using the parent inverse matrix. Effectively, we set the child's origin
to the parent. When the child has an identity local transform, then the
child is world-space aligned with its parent (scale excluded).
**Technical:** In both cases, the hidden parent inverse matrix is
generally set to identity (cleared or "not used") as long as the parent
has no shear. If the parent has shear, then this matrix will not be
entirely cleared. It will contain shear to counter the parent's shear.
This is required, otherwise the object's local matrix cannot be properly
decomposed into location, rotation and scale, and thus cannot preserve
the world transform.
If the child's world transform has shear, then its world transform is
not preserved. This is currently not supported for consistency in the
handling of shear during the other parenting ops: Parent (Keep
Transform), Clear [Parent] and Keep Transform. If it should work, then
another patch should add the support for all of them.
Reviewed By: sybren, RiggingDojo
Differential Revision: https://developer.blender.org/D14581
The new created strokes were not setting the right `orig` pointers, so the select operator could not select the strokes and points.
Now, the original pointers are set in the new strokes. To set the values is necessary assign the original pointer from the reference stroke because in modifiers the stroke is evaluated, so we need back two levels: Eval->Eval->Orig
Issue introduced in {rB80859a6cb272}
Only seen on Windows.
The `keyword_parse` lambda function code did not consider `\r` as a
whitespace char, resulting in a wrong parse.
(More investigation needs to be done).
Assume that only one layer matches the id and return instead
of continuing to iterate over attributes after the layers have
been potentially reallocated.
This commit removes all EEVEE specific code from the `gpu_shader_material*.glsl`
files. It defines a clear interface to evaluate the closure nodes leaving
more flexibility to the render engine.
Some of the long standing workaround are fixed:
- bump mapping support is no longer duplicating a lot of node and is instead
compiled into a function call.
- bump rewiring to Normal socket is no longer needed as we now use a global
`g_data.N` for that.
Closure sampling with upstread weight eval is now supported if the engine needs
it.
This also makes all the material GLSL sources use `GPUSource` for better
debugging experience. The `GPUFunction` parsing now happens in `GPUSource`
creation.
The whole `GPUCodegen` now uses the `ShaderCreateInfo` and is object type
agnostic. Is has also been rewritten in C++.
This patch changes a view behavior for EEVEE:
- Mix shader node factor imput is now clamped.
- Tangent Vector displacement behavior is now matching cycles.
- The chosen BSDF used for SSR might change.
- Hair shading may have very small changes on very large hairs when using hair
polygon stripes.
- ShaderToRGB node will remove any SSR and SSS form a shader.
- SSS radius input now is no longer a scaling factor but defines an average
radius. The SSS kernel "shape" (radii) are still defined by the socket default
values.
Appart from the listed changes no other regressions are expected.
This adds a structure, `ABCReadParams`, to store some parameters passed
to `ABC_read_mesh` so we avoid passing too many parameters, and makes it
easier to add more parameters in the future without worrying about
argument order.
Differential Revision: https://developer.blender.org/D14484
Box, Circle and Lasso select were not taking into account if a
mask(point) was parented; selection was only succeeding in the original
place.
Now check coordinates from evaluated mask (points) instead while
setting selection flags and DEG tagging still happens on the original ID.
Maniphest Tasks: T97135
Differential Revision: https://developer.blender.org/D14651
Extremely subttle bug that would only appear in some specific
circumstances, would cause memfile undo writing code to falsely detect
some ID as changed because it would get the wrong 'starting point' of
comparison with existing previous memfile step.
See T85756 for detailed explanation and reproducible case.
This adds a new node editor overlay that helps users to see where
named attributes are used. This is important, because named
attributes can have name collisions between independent node
groups which can lead to hard to find issues.
Differential Revision: https://developer.blender.org/D14618
When GPU subdivision is enabled the mesh objects remain unsubdivided on
the CPU side, and subdivision should be requested somewhat manually (via
`BKE_object_get_evaluated_mesh`).
When referencing an object, the Object Info node calls
`bke::object_get_evaluated_geometry_set` which first checks if a Geometry
Set is present on the object (unless we have a mesh in edit mode). If so
it will return it, if not, the object type is discriminated, which will
return a properly subdivided mesh object via `add_final_mesh_as_geometry_component`.
The unsubdivided mesh is returned because apparently the check on the
Geometry Set always succeeds (as it is always set in `mesh_build_data`).
However, the mesh inside this Geometry Set is not subdivided.
This adds a check for a MeshComponent in this Geometry Set, and if one
exists, calls `add_final_mesh_as_geometry_component` which will ensure
that the mesh is subdivided on the CPU side as well.
Differential Revision: https://developer.blender.org/D14643
Originally was noticed when using a linked background scene and a scene
camera from another (local) scene.
The root issue was that relation from view layer to object's base flags
evaluation was using wrong view layer. This is because the relation was
created between object and currently built view layer, and it only was
happening once (since the object-level relations are only built once).
Depending on order in which `build_object` was called it was possible
that relation from a wrong view layer was used.
Now the code is better split to indicate which parts of object relations
are built when object comes from a base in the view layer, and which ones
are built on indirect linking of object to the dependency graph.
This patch makes relations correct in the cases when the same object is
used as a base in both active and set scenes. But, the operation which
handles object-level flags might not behave correctly as there is no
known design of what is the proper thing to do in this case. Making a
clear design and implementation of case when object is shared between
active and set scene is outside of the scope of this patch.
Differential Revision: https://developer.blender.org/D14626
Prefer using immVertex3f when 3D shaders are used for 2D rendering due to overhead of vertex padding in hardware. CPU overhead is negligible.
Authored by Apple: Michael Parkin-White
Ref T96261
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D14494
MSL follows C++ standard convention, and such variable declarations within switch statements must have their scope localised within each case. Adding braces in all cases ensures correct behaviour and avoids 'case ... bypass initialization of local variable' compilation error.
struct initialisation to follow C++ rules for Metal. Implicit constructors replaced with either explicit constructors or list-initialization where appropriate.
Ref T96261
Authored by Apple: Michael Parkin-White
Reviewed By: fclem
Maniphest Tasks: T96261
Differential Revision: https://developer.blender.org/D14451
Add a new operator, "Start Tweaking Strip Actions (Full Stack)", which
allows you to insert keyframes and preserve the pose that you visually
keyed while upper strips are evaluating,
The old operator has been renamed from "Start Tweaking Strip Actions" to
"Start Tweaking Strip Actions (Lower Stack)" and remains the default for
the hotkey {key TAB}.
**Limitations, Keyframe Remapping Failure Cases**:
1. For *transitions* above the tweaked strip, keyframe remapping will
fail for channel values that are affected by the transition. A work
around is to tweak the active strip without evaluating the upper NLA
stack.
It's not supported because it's non-trivial and I couldn't figure it
out for all transition combinations of blend modes. In the future, it
would be nice if transitions (and metas) supported nested tracks
instead of using the left/right strips for the transitions. That
would allow the transitioned strips to overlap in time. It would also
allow N strips to be part of the (previously) left and right strips,
or perhaps even N strips being transitioned in sequence (similar to a
blend tree). Proper keyframe remapping through all that is currently
beyond my mathematical ability. And even if I could figure it out,
would it make sense to keyframe remap through a transition?
//This case is reported to the user for failed keyframe insertions.//
2. Full replace upper strip that contains the keyed channels.
//This case is reported to the user for failed keyframe insertions.//
3. When the same action clip occurs multiple times (colored Red to
denote it's a linked strip) and vertically overlaps the tweaked
strip, then the remapping will generally fail and is expected to
fail.
I don't plan on adding support for this case as it's also non-trivial
and (hopefully) not a common or expected use case so it shouldn't be
much of an issue to lack support here.
For anyone curious on the cases that would work, it works when the
linked strips aren't time-aligned and when we can insert a keyframe
into the tweaked strip without modifying the current frame output of
the other linked strips. Having all frames sampled and the strip
non-time aligned leads to a working case. But if all key handles are
AUTO, then it's likely to fail.
//This case is not reported to the user for failed keyframe
insertions.//
4. When using Quaternions and a small strip influence on the tweaked
Combine strip. This was an existing failure case before this patch
too but worth a mention in case it causes confusion. D10504 has an
example file with instructions.
//This case is not reported to the user for failed keyframe insertions. //
5. When an upper Replace strip with high influence and animator keys to
Quaternion Combine (Replace is fine). This case is similar to (4)
where Quaternion 180 degree rotation limitations prevent a solution.
//This case is not reported to the user for failed keyframe insertions.//
Reviewed By: sybren, RiggingDojo
Differential Revision: https://developer.blender.org/D10504
Undefined behaviour for divergent control-flow fixes, replacement for partial vector references, and resolution of a number of calculation precision issues occuring on macOS.
Authored by Apple: Michael Parkin-White
Ref: T96261
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D14437
This patch fixes a typo in commit e59f754c16 which incorrectly uses `GPU_TEXTURE_ARRAY` instead of `GPU_FORMAT_COMPRESSED`.
`GPU_FORMAT_COMPRESSED` and `GPU_TEXTURE_ARRAY` both currently evaluate to 16, so this patch does not change anything functionally; however, this patch will prevent issues from arising in the future.
Reviewed By: fclem
Differential Revision: https://developer.blender.org/D14384
Add operator to select markers left/right of the current frame
(including the current frame).
`bpy.ops.marker.select_leftright(mode='LEFT', extend=False)`
`mode` can be either 'LEFT' or 'RIGHT'.
The naming and defaults of the above variables match similar operators
(e.g., `bpy.ops.nla.select_leftright`)
This also adds a new sub-menu to the Marker menu found in animation
editors, exposing both the new `bpy.ops.marker.select_leftright`
operator as well as the `bpy.ops.marker.select_all` operator.
Despite the name "Before Current Frame" and "After Current Frame", it
also selects a marker that falls on the current from for both of the
modes. This is to match the behavior found in the `nla.select_leftright`
operator.
RCS: https://blender.community/c/rightclickselect/OgmG/
Reviewed by: sybren, looch
Differential Revision: https://developer.blender.org/D14176
F2 allows renaming lots of different types of active items, and now it
also works for markers.
Before, Ctrl+M was used, and it's context-sensitive: you often get
"Mirror Keys" instead, when your cursor isn't on the markers region, and
that operator has nothing to do with either renaming or markers.
**What this commit does:**
- Replace Ctrl+M shortcut with F2.
- Adds the `TOPBAR_PT_name_marker` panel which is implemented similar
to the global rename panel. This having to press enter twice to
confirm or escape twice to cancel, which would happen if the
`marker.rename` operator was called directly.
- Replace usages of `marker.rename` in the UI with `wm.call_panel`.
- To make the Industry Compatible keymap consistent with Blender
Default, the rename shortcut only works when hovering the markers
area.
Reviewed By: ChrisLend, sybren
Differential Revision: https://developer.blender.org/D12298
There are no real difference from the previous icons, but since some
manual changes were introduced in the icons file, we still needed
a final sync between them.
Unlike regular selection cycling that is activated when clicking again
in the same location, object mode would cycle to another object
if the object that was selected happened to already be active.
This made it impossible to click-drag to tweak the active object
if there were other objects behind it as those would be activated first.
Resolves T96752.
Particles baked into memory would never load the final frame because
of an off-by-one error calculating the particles `dietime`.
This value indicates the frame which the particle ceases to exist but
was being set to the end-frame which caused this bug as the scenes
end-frame is inclusive.
While the last frame was properly written and read from memory,
the `dietime` was set to the last frame causing all the particles to be
considered dead when calculating the cached particle system.
The GPU evaluation for curves will have to change significantly from the
current particle hair drawing code, due to its more general use cases
and support for more curve types. To simplify that process and avoid
introducing regressions for the rendering of hair particle systems,
this commit splits drawing functions for the curves object and
particle hair.
The changes are just inlining of functions and copying code
where necessary.
Differential Revision: https://developer.blender.org/D14576
Problem is that the orco layer was not taken care of by the GPU
subdivision routines. This only handles the issues for EEVEE/Workbench.
For Cycles, this would need to be handled at the wrapper level somehow.
Use Alt-Slash to remove objects from local-view (was M prior to [0]),
following the convention of using Alt to perform the reverse of an
action. Also remove the confirmation menu as this key as it can be
undone and it's not likely to be pressed by accident.
This can be useful to quickly subtract items from a complex selection
with items that only become visible when entering local-view.
The M key was originally used in 2.4x since moving between layers wasn't
possible. Now moving between collections is possible in local-view
the keys collided.
[0]: cf5d582b77
edges
When wireframe mode is turned on, the subdivision edges not originating
from coarse edges were also drawn as regular edges, which would confuse
users trying to select them. These should not be drawn in edit mode,
only in object mode when optimal display is turned off (matching the CPU
subdivision case).
Fix word-wrapped tooltip text not showing by aligning to pixel grid.
See D14639 for more details.
Differential Revision: https://developer.blender.org/D14639
Reviewed by Campbell Barton
The "dir" argument to `BKE_where_on_path` was only actually
used in a few places. It's easier to see where those are if there
isn't always a dummy argument.
When displaying a deform modifier in edit mode, a cached
array of positions is used. Parallelizing bounds calculation when
that array exists can improve the framerate when editing slightly
(a few percent). I observed an improvement of the min/max itself
of about 10x (4-5ms to 0.4ms).
If all of the curves are poly curves, the evaluated positions are the
same as the original positions. In this case just reuse the original
positions span as the evaluated positions.
Addresses T97257, to make it consistent with regular attributes tab
Review by: Julian Kaspar
Differential Revision: https://developer.blender.org/D14631
Ref D14631
`GPU_shader_get_uniform_block` is marked as deprecated and the value
returned does not match what `GPU_uniformbuf_bind` expects.
Also, small typo fix in python error message.
Differential Revision: https://developer.blender.org/D14638
This behaviour was introduced in a687d98e67 to bring the old obscure
"M" operator to remove objects from the local view. In order to avoid
the keymap clash with the Move to Collection operator, the Move to
Collection was artificially restricted to work in local view.
In retrospect, the "Remove from Local View" operator is in the menu anyways,
so it didn't even need to have a shortcut (back in 2.79 the operator was
not in a menu).
The changes introduced here are:
* No shortcut for "Remove from Local View"
* No more restrictions to "Move/Link to Collection" from local view.
Thanks for Philipp Oeser for digging the old commit that introduced this
and for the rationale on the changes.
This can be useful to match transforms to what native Cycles
would see in Blender, as USD typically uses centimeters, but
Blender uses meters. This patch also fixes the hardcoded focal
length multiplicator, which is now using the same units as
everything else. Default of "stageMetersPerUnit" is 0.01 to match
the USD default of centimeters.
Differential Revision: https://developer.blender.org/D14630
Print it as a "%s" so that possible percentage symbols in the
error message does not cause issues.
Use proper assert (assert(true) is a no-op).
Also use `empty()` instead of `length()`.
Reviewed with Clement in real life.
Solves compilation warning with Clang, and moves manipulation with
DNA structures to the designed way for C++.
The tests and few other places are update to the new code by Jacques.
Ref T96847
Maniphest Tasks: T96847
Differential Revision: https://developer.blender.org/D14625
After appending, new link/append code would delete linked IDs, even if
those where pre-existing. Note that this would actually lead to invalid
memory access later in append code (ASAN crash).
This was one of multiple placeholder brushes to simplify development.
Having it is not necessary anymore.
It was a brush that could add new curves according to a specific density.
This functionality will be brought back as a new brush later.
Ref T97255.
Implements T97163
Newly created meshes have all voxel remesher checkboxes aside from Fix Poles enabled.
Startup files updated with versioning.
Reviewed By @JulianKaspar
Differential Revision: https://developer.blender.org/D14608
Ref D14608
Regression in [0] which caused canceled PRESS events not to generate
CLICK_DRAG.
Resolve by checking for an active brush tool in poll instead of the
PARTICLE_OT_brush_edit invoke function.
[0]: 4d0f846b93,
- Add logging for CLICK_DRAG event handling to debug drag events.
- Use logging API for reporting the key-map, operator and event.
This command now prints useful information for investigating
key-map and event handling issues:
blender --log "wm.handler.*" --log-level 4
Internally many offsets for BLF were integers but exposed as floats,
since these are used in pixel-space, many callers were converging them
back to integers. Simplify logic by using ints.
Support sub-pixel kerning and hinting for future support for improved
character placement. No user visible changes have been made.
- Calculate sub-pixel offsets, using integer maths.
- Use convenience functions to perform the conversions and hide the
underlying values.
- Use `ft_pix` type to distinguish values that use sub-pixel integer
values from freetype and values rounded to pixels.
This was originally based on D12999 by @harley with the user visible
changes removed so they can be applied separately.
In order to allow GLSL Cross Compilation across platforms, expose in
Python the `GPUShaderCreateInfo` strategy as detailed in
https://wiki.blender.org/wiki/EEVEE_%26_Viewport/GPU_Module/GLSL_Cross_Compilation
The new features can be listed as follows:
```
>>> gpu.types.GPUShaderCreateInfo.
define(
fragment_out(
fragment_source(
push_constant(
sampler(
typedef_source(
uniform_buf(
vertex_in(
vertex_out(
vertex_source(
>>> gpu.types.GPUStageInterfaceInfo.
flat(
name
no_perspective(
smooth(
>>> gpu.shader.create_from_info(
```
Reviewed By: fclem, campbellbarton
Differential Revision: https://developer.blender.org/D14497
Propagate the fp settings from the main thread to all the worker threads (the fp settings includes the FZ settings among other things) - this guarantees consistency in execution of floating point math regardless if its executed in tbb thread arena or on main thread
Add FZ mode to arm64/aarch64 in parallel to the way its been done on intel processors, currently compiling for arm target does not set this mode at all, hence potentially runs slower and with possible results mismatch with intel x86.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D14454
We better handle NULL object pointers before doing layer collections
resync, otherwise said resync process has to deal with those NULL
pointers. By the look of it this mistake has been there since the origin
of the remapping/relinking code.
Also for safety (and optimization), do not perform layer collection
resync from `libblock_remap_data_postprocess_object_update` when
`libblock_remap_data_postprocess_collection_update` is called
immediately afterwards.
Also added same 'skip on NULL collection object pointer' check to
`layer_collection_local_sync` as the one in
`layer_collection_objects_sync`, since it's fairly hard to always
guaranty there is no such NULL pointer when calling that code.
Animation would still play in the viewport.
There are two ways to unlink an action from the Outliner:
[1] `Unlink Action` on the Animation Data context menu.
This does `outliner_do_data_operation` / `unlinkact_animdata_fn` and has
the correct DEG update.
[2] `Unlink` on the Action context menu
This does `outliner_do_libdata_operation` / `unlink_action_fn` and was
missing the DEG update.
Now add the missing DEG update to the second case.
Maniphest Tasks: T95679
Differential Revision: https://developer.blender.org/D14089
The issue reported was that the recently introduced manual framerange of
an action (see rB5d59b38605d6) was not having an immediate effect in
case the action was pushed down from the Action Editor and only showed
its effects after e.g. saving and reloading the file. However doing the
same thing (pushing down the action) was working fine when done from the
NLA.
Now bring pushdown in sync (in terms of DEG update tagging) between the
Action Editor and the NLA, meaning that now both the owner and the
action are tagged when pushdown happens from the Action Editor as well.
Fixes T96964.
Maniphest Tasks: T96964
Differential Revision: https://developer.blender.org/D14564
The new tab in the properties panel for "Color Attributes" shows the
data type and domain, just like the Attributes tab.
The issue is that the UI does not scale well and can only display the
full names when dragged to an extreme width.
This is due to the inclusion of the render icon in between the
attribute name and attribute type description.
This patch changes the order that items are displayed in the Color
Attributes Panel. Moving the render icon to the very right.
The result is consistent with other parts of the Blender UI and does
not take as much space to display the full text.
Reviewed By: @jbakker
Differential Revision: https://developer.blender.org/D14567
Ref D14567
Color filter sharpening now clamps the output.
The sharpening delta is now calculated from the
difference of two levels of vertex averaging instead
of one smooth iteration and the base color.
TODO: Sharpen in a different color space;
SRGB-linear has saturation artifacts. I
tried HSL but it had value artifacts. I'd
like to try LAB but we don't seem to have
conversion functions for it (at least as far
as I could see).
- Replace SPACE_TYPE_LAST with SPACE_TYPE_NUM (adding 1).
- Rename RGN_TYPE_LEN to RGN_TYPE_NUM
This makes it possible to tag space-type/region-type combinations
with `bool tag[SPACE_TYPE_NUM][RGN_TYPE_NUM]` which reads more clearly
than `bool tag[SPACE_TYPE_LAST + 1][RGN_TYPE_LEN]`.
The type of the key was changed from string to integer in 66da2f537a, but
the GSet was still being created for string keys.
As long as the integers stay small enough, this even kind of works on
little-endian systems, but of course it should use an integer hash instead.
Currently, hovering over a socket itself shows no tooltip at all, while
hovering over its value field shows "Default value", which is not helpful.
This patch therefore implements socket tooltips following the proposal at
https://blender.community/c/rightclickselect/2Qgbbc/.
A lot of the basic functionality was already implemented for Geometry Nodes,
where hovering over the socket itself shows introspection info.
This patch extends this by:
- Supporting dynamic tooltips on labels, which is important for good tooltip
coverage in a socket's region of the node.
- Adding a function to setting a dynamic tooltip for an entire uiLayout, which
avoids needing to set it manually for a wide variety of socket types.
- Hiding the property label field in a tooltip when dynamic tooltip is also
provided. If really needed, this label can be restored through the dynamic
tooltip, but in all current cases the label is actually pointless anyways
since the dynamic tooltip gives more accurate and specific information.
- Adding dynamic tooltips to a socket's UI layout row if it has a description
configured, both in the Node Editor as well as in the Material Properties.
Note that the patch does not add any actual tooltip content yet, just the
infrastructure to show them. By default, sockets without a description still
show the old "Default value" tooltip.
For an example of how to add socket descriptions, check the Cylinder node
in the Geometry Nodes.
Differential Revision: https://developer.blender.org/D9967
When using proportional editing, the 'project onto self' snap setting
is ignored since proportional editing does not allow snapping to
self. The UI should reflect this fact. This patch makes 'project onto
self' active only when proportional editing is off.
Reviewed By: mano-wii
Differential Revision: https://developer.blender.org/D14496
Simply add the few lines so that the Cycles renderable New Curves
(with a small temporary patch to output the new type to render engines)
would get assigned a shader (in particular a hair shader).
Differential Revision: https://developer.blender.org/D14622
* Don't assume the display colorspace name fully defines the transform
to display space, this is not true in OpenColorIO 2 where view transforms
may be defined in more complexs ways than just specifying a colorspace.
* In places where we need to store the display colorspace name, resolve
<USE_DISPLAY_NAME> token manually.
Ref T96590
Adding nodes via `NodeAddOperator` could fail if the node's poll fails
in `rna_NodeTree_node_new`. Examples are trying to add a RenderLayers
node or a Cryptomatte node to a nodegroup.
Now except the python error and use blender error reporting only instead
of throwing full python stacktraces at the user.
Differential Revision: https://developer.blender.org/D14584
For some reasons(c) some base classes (like `bpy.types.Modifier`) are
not listed anymore by a call to
`bpy.types.ID.__base__.__subclasses__()`, unless they are first accessed
explicitely (e.g. by executing `bpy.types.Modifier` etc.).
Coordinate checks in `spline_under_mouse_get` need to take place with
the evaluated mask to get the right center. This is now more in line to
how this is done in `ED_mask_point_find_nearest`.
Note: similar issues are reported with box/circle/lasso selection in
T97135, will tackle these separately though.
Maniphest Tasks: T85467
Differential Revision: https://developer.blender.org/D14598
This implements the icon for snake hook, as well as tweak the
growth brush to make it bigger and with the triangles more
distinct for better reading of the icon.
Differential Revision: https://developer.blender.org/D14616
It was not possible to assign a shortcut to menu items in the insert
key-frame menu going back to version 2.7x. Doing so would replace the
current key that opens the insert keyframe menu (I-key by default),
instead of binding a key to insert a key-frame for the keying-set
referenced by the menu item.
Now each menu item can be bound to a key or added to the "Quick
Favorites" menu, directly inserting a key-frame for the corresponding
keying-set.
Note that users must use the operator `anim.keyframe_insert_by_name`
when setting up key-shortcuts as `anim.keyframe_insert` is only intended
to launch the menu.
Keymap Editor:
When editing these key-map items in the key-map editor, the keying-set
identifier must be used. At the moment the key-map editor doesn't
support showing a drop-down list. The identifiers can be used from the
tool-tip or the info editor.
{F12994924}
Details:
Use `ANIM_OT_keyframe_insert_by_name` instead of
`ANIM_OT_keyframe_insert_menu` for the insert keyframe popup menu to
resolve the following issues binding keys to keying sets:
- The index of the keying set isn't stable (adding/removing keying sets
may change it).
- Binding a key to items in the popup menu triggers a popup instead of
inserting a key using the keying set from the menu item.
While support for using the current operator could be improved, it will
still only work for built-in keying sets, so I'd prefer to use an
operator that is intended for key-bindings.
Besides supporting binding keys to menu items there are no functional
changes.
Reviewed By: sybren
Ref D14289
This simply adds access to the vertex crease layer from Python
code. Documentation was modified to remove "Edge" as it is
shared between edges and vertices, similarly to bevel weights.
Differential Revision: https://developer.blender.org/D14605
- Missing star prefix.
- Unnecessary indentation.
- Blank line after dot-points
(otherwise doxygen merges with the previous dot-point).
- Use back-slash for doxygen commands.
- Correct spelling.
These operators build a list of all lightgroups that are used by the view layer's
scene and either add all used lightgroups that are not part of the view layer yet
or remove all lightgroups in the view layer that are not being used.
Differential Revision: https://developer.blender.org/D14596
As far as I can see, it makes a lot of sense to have the alpha channel here, it matches the 2.x behavior and also matches what Eevee is doing.
Differential Revision: https://developer.blender.org/D14595
Use the Extend method for these, as these do not work correctly. For UVs
it's better to extend the UVs from the same face, and for tangent space
the normals should be encoded in a matching tangent space.
Later the Adjacent Faces method might be improved to support these cases.
Ref T96977
Differential Revision: https://developer.blender.org/D14572
Port the "Normal" and "Curve Tangent" nodes to the new curves data-block
to avoid the conversion to `CurveEval`. This should make them faster by
avoiding all that copying, but otherwise nothing else has changed.
This also includes a fix to move the normal mode as a built-in curve
attribute when converting to and from `CurveEval`. The attribute is
needed because the option is used implicitly in many nodes currently.
Differential Revision: https://developer.blender.org/D14609
Since effect strips can't be transformed directly, their selection had
to be forced in order to process them. This often failed in more
complicated scenarios, because there was no attempt to parse hierarchy
completely. In worst case only one effect in chain would be selected.
This code was marked by `XXX_DURIAN_ANIM_TX_HACK`.
Instead solution described above, a collection of strips that depend
on non effect strip position is built by function
`query_time_dependent_strips_strips` and it is stored in `TransSeq`.
In `flushTransSeq` this collection is iterated and transformation offset
is applied to effect strip animation. Strips in collection should be
consistent with true state of dependency and should be complete.
Functional changes:
- When 2-input effect strip changes position, animation is offset even
if only handles are moved. This only applies to 2-input effect however.
- Selection is not extended to include effect strips anymore. If effects
are to be moved, they must be selected manually. This is because
previously it was very hard to reorganize effects in chain, since moving
first strip in chain would always select anywhere from 1 to n effects.
So creating or filling gap in channel would almos always result in
collision especially if their order in timeline doesn't perfectly
represent their order in chain.
Previously, the new attributes were zero-initialized. However, sometimes
the default has to be something else. The old behavior led to unexpected
behavior in the Snap Curves to Surface operator in Deform mode, when the
curves were not attached to the surface before.
Differential Revision: https://developer.blender.org/D14588
This change moves the grid panel UI from the View tab up into the
Overlay panel.
Reasons to move to the Overlay panel include:
- Consistency with the grid options in the 3D viewport
- The grid has been drawn as an Overlay for quite some time already
Additional changes that now make sense to have:
- The grid responds to the main Overlay show/hide toggle
- Adds a toggle to show/hide the grid which is consistent with overlays in general
As before, these grid controls are only available for active UV edit
sessions.
Differential Revision: https://developer.blender.org/D11862
- Use "curve" instead of "spline" in comments
- Use non-plural variable names
- Tag topology dirty after resolution modified rather than positions
- Reorder enum values to change which value is zero (and the default)
- Remove a duplicate unused variable
draw_common.h was included in a C++ file
leading to the linker looking for the
decorated name for `G_draw` which lead
to a linker error.
adding an extern "C" for C++ fixes
the issue.
lengths along a set of points. This can be used for the sample curves
node, or finding new points along a curve when extending
or shrinking it.
This commit uses it in the snake hook brush as an example.
The logic is similar to the uniform length sampling, but the next
sample length is retrieved from the input instead of multiplication.
For the sample node in the future, though this sort of sampling can be
potentially done more efficiently for specific curve types besides
poly curves, it's simpler, at least as a start, to work on a set of
evaluated points that can be treated like a poly curve.
Differential Revision: https://developer.blender.org/D14571
Avoid installing all the ogg, theora, xvid etc. codec lib dev packages
unless we actually build ffmpeg itself. Otherwise they are not necessary
for Blender build itself.
This adds support to show dots for the curves points when in edit mode,
using a specific overlay.
This also adds `DRW_curves_batch_cache_create_requested` which for now
only creates the point buffer for the newly added `edit_points` batch.
In the future, this will also handle other edit mode overlays, and
probably also replace the current curves batch cache creation.
Maniphest Tasks: T95770
Differential Revision: https://developer.blender.org/D14262
The menu with the options was not visible because the tool checked must be the sculpt, not draw.
This was broken in old version, but I cannot determine when or if never worked at expected.
The `frame_offset` used for creating `TimeSamplings` when exporting was
being clamped, which would make subframe sampling potentially fail, or
get out of sync.
Both the Alembic and USD libraries use double precision floating
point numbers internally to store time. However the Alembic I/O
code defaulted to floats even though Blender's Scene FPS, which is
generally used for look ups, is stored using a double type. Such
downcasts could lead to imprecise lookups, and would cause
compilation warnings (at least on MSVC).
This modifies the Alembic exporter and importer to make use of
doubles for the current scene time, and only downcasting to float
at the very last steps (e.g. for vertex interpolation). For the
importer, doubles are also used for computing interpolation weights,
as it is based on a time offset.
Although the USD code already used doubles internally, floats were used
at the C API level. Those were replaced as well.
Differential Revision: https://developer.blender.org/D13855
This patch adds color attributes to TexPaintSlot. This allows an easier selection
when painting color attributes.
Previously when selecting a paint tool the user had to start a stroke, before the
UI reflected the correct TexPaintSlot. Now when switching the slot the active
tool is checked and immediate the UI is drawn correctly.
In the future the canvas selector will also be used to select an image or image texture node
to paint on. Basic implementation has already been done inside this patch.
A limitation of this patch is that is isn't possible anymore to rename images directly from
the selection panel. This is currently allowed in master. But as CustomDataLayers
aren't ID fields and not owned by the material supporting this wouldn't be easy.
{F12953989}
In the future we should update the create slot operator to also include color attributes.
Sources could also be extended to use other areas of the object that use image textures
(particles, geom nodes, etc... ).
Reviewed By: brecht
Maniphest Tasks: T96709
Differential Revision: https://developer.blender.org/D14455
Avoid Blender overwriting artist's choices. The automatic change from
"Hold" (i.e. bidirectional extrapolation) to "Hold Forward" (i.e. only
extrapolate forward in time) has been removed.
This patch does not change strip evaluation. Between two strips, the
first with `None` extrapolation and the next with `Hold`, neither strip
will evaluate, which matches previous behavior. A future patch can
change the evaluation behavior.
Reviewed By: RiggingDojo, sybren
Maniphest Tasks: T82230
Differential Revision: https://developer.blender.org/D14230
Previous commit (rBrB59681a7ccdcf) was effectively doing nothing, due to
weird hacks we had to do with OpenVDB 8.0 to 'integrate' NanoVDB.
Now OpenVDB 9.0 natively includes NanoVDB, which allows us to greatly
simplify that part of the code in install_deps.
The all_objects.blend test scene (in subversion tests repo) contained an
object with a subdivision surface. Which changes vertex positions
slightly, depending on used OpenSubDiv version and the compile flags. It
seems that the intent of the test was "test export of meshes that use
modifiers", so I changed that object to be a cube with a simple "taper"
modifier instead.
While at it, changed OBJ exporter test code to always print the
"expected and what we got" text difference details, when a test fails.
Much easier to see than just "the files are different" output. The code
to print that was behind an off by default flag for some reason.
This diff should get comitted together with updated all_objects templates
in subversion tests repo.
Reviewed By: Sebastian Parborg
Differential Revision: https://developer.blender.org/D14597
The problem was the original file had some vertex weight information, but the weights array was empty, so the duplication was not done and the free memory crashed.
To avoid this type of errors, now before duplicate weights the function checks the pointer and also the number of weights elements in the array to avoid the duplicatiopn of empty data.
With the increased use of multi-character format units and keyword-only
arguments these are increasingly difficult to make sense of.
Split the string onto multiple lines, one per argument.
While verbose it's easier to understand and add new arguments.
Currently, only Lightgroups that exist in the current view layer can be
selected from object or world properties.
The internal UI code already has support for search fields that accept
unknown input, so I just added that to the API and use it for lightgroups.
When a lightgroup is entered that does not exist in the current view layer
(e.g. because it's completely new, because the view layer was switched or
because it was deleted earlier), a new button next to it becomes active and
adds it to the view layer when pressed.
Differential Revision: https://developer.blender.org/D14540
* Add missing GLEW and hgiGL libraries for Hydra
* Fix wrong case sensitive include
* Fix link errors by adding external libs to static Hydra lib
* Work around weird Hydra link error with MAX_SAMPLES
* Use Embree by default for Hydra
* Sync external libs code with standalone
* Update version number to match Blender
* Remove unneeded CLEW/GLEW from test executable
None of this should affect Cycles in Blender.
Ref T96731
This seems to serve no purpose anymore, I don't see anywhere that
CD_MFACE is requested for modifier evaluation, and it's confusing
to have this in this final normals computation function.
Found while looking into D14579.
Differential Revision: https://developer.blender.org/D14580
The mechanism to instance meshes when there are no modifiers did not take
into account that modifiers might get re-evaluated from an operator that
requests loop normals. Now check for that case and no longer use the
instance then.
In the future, a better solution may be to compute loop normals on demand
as is already done for poly and vertex normals, but that would be a big
change.
Differential Revision: https://developer.blender.org/D14579
This frequently showed up in profiling but shouldn't.
This also updates the code to use atomics for more correctness and
adds multi-threading for better performance.
This implements two optimizations:
* Reduce virtual function call overhead when a non-standard virtual
array is used as input.
* Use a lambda in `type_conversion.cc`.
In my test setup, which creates a float attribute filled with the index,
the running time drops from `4.0 ms` to `2.0 ms`.
Differential Revision: https://developer.blender.org/D14585
I observed a 4-5x performance improvement (from 50ms to 12ms)
with five million points, though obviously the change depends on
the hardware.
In the future we may want to disable the parallelization in
`parallel_invoke` when there is a small amount of points.
Differential Revision: https://developer.blender.org/D14590
This patch adds an option to only use every n-th segment of the
envelope result. This can be used to reduce the complexity of the
result.
Differential Revision: http://developer.blender.org/D14503
Method which overrides a base class's virtual methods are expetced to
be marked with `override`. This also gives better idea to the developers
about what is going on.
The first element of the iterator was not being tested against the flag.
So in some cases it would lead to more objects been made into
single-user than the active (or selected) ones.
The goal is to make the Add menu more convenient for the new curves object.
The following changes are done:
* Add `curves` submenu.
* Add an `Empty Hair` operator that also sets the surface object.
* Rename the old operator to `Random`. It's mostly for testing at this point.
Differential Revision: https://developer.blender.org/D14556
This operator snaps the first point of every curve to the corresponding
surface object. The shape of individual curves or their orientation is
not changed.
There are two different attachment modes:
* `Nearest`: Move each curve so that the first point is on the closest
point on the surface. This should be used when the topology of the
surface mesh changed, but the shape generally stayed the same.
* `Deform`: Use the existing attachment information that is stored
for curves to move curves to their new location when the surface
mesh was deformed. This generally does not work when the
topology changed.
The purpose of the operator is to help setup the "ground truth"
for how curves are attached to the surface. When the ground
truth surface changed, the original curves have to be updated
as well. Deforming curves based on an animated surface will be
done with geometry nodes independent of the operator.
In the UI, the operator is currently exposed in curves sculpt mode
in the `Curves > Snap Curves to Surface` menu.
Differential Revision: https://developer.blender.org/D14515
Avoid errors in the legacy Pose Library panel (in Armature properties)
when the Pose Library add-on is disabled.
It's unfortunate that a built-in panel now has knowledge of an add-on.
Then again, it's temporary (one or two Blender releases), and it now uses
feature detection instead of just assuming the add-on is enabled.
The operator could crash in case the context "object" was overridden
from python, but the "active_object" wasnt (and the active object was
not a mesh).
Reason for the crash is a mismatch in the operators poll function
`data_transfer_poll` vs. `dt_layers_select_src_itemf` -- in the former,
the overriden "object" was respected (and if this was a mesh, the poll
was permissive), in the later it wasnt and only the "active_object" was
used (if this was not a mesh, a crash would happen trying to get an
evaluated mesh).
Now rectify how the object which is used is being fetched -> use
`ED_object_active_context` everywhere (see also rBe560bbe1d584).
Maniphest Tasks: T96888
Differential Revision: https://developer.blender.org/D14552
This does two things:
* Introduce new `materialize_compressed` methods. Those are used
when the dst array should not have any gaps.
* Add materialize methods in various classes where they were missing
(and therefore caused overhead, because slower fallbacks had to be used).
This improves performance e.g. when creating an integer attribute
based on an index field. For 4 million vertices, I measured a speedup
from 3.5 ms to 1.2 ms.
Add the ability to get/set the selected text.
**Calling the new methods:**
- `bpy.data.texts["Text"].region_as_string()`
- `bpy.data.texts["Text"].region_from_string("Replacement")`
Expose the "Connected" mode from the weld modifier in the
"Merge by Distance" geometry node. This method only merges
vertices along existing edges, but it can be much faster
because it doesn't have to build a KD Tree of all selected
points.
Differential Revision: https://developer.blender.org/D14321
Change the modifier name in the modifier stack to "Curvature 3D"
to be consistent with the modifier name in the drop-down.
Differential Revision: https://developer.blender.org/D14476
If all islands had a size of zero, a division by zero would occur in
`GEO_uv_parametrizer_pack`, causing the UV coordinates to be set to
NaN. An alternative approach would be to skip packing islands with a
zero size, but If UV coordinates are for example outside the 0-1 range,
it's better if they get moved into that range.
Differential Revision: https://developer.blender.org/D14522
Set the curve resolution to Bezier and Nurbs curves when converting
data using `curves_to_curve_eval`. This was missed in 9ec12c26f1.
Differential Revision: https://developer.blender.org/D14577
Add "for_write" on function names that retrieve mutable data arrays.
Though this makes function names longer, it's likely worth it because
it allows more easily using the const functions in a non-const context,
and reduces cases of mistakenly retrieving with edit access.
In the long term, this situation might change more if we implement
attributes storage that is accessible directly on `CurvesGeometry`
without duplicating the attribute API on geometry components,
which is currently the rough plan.
Differential Revision: https://developer.blender.org/D14562
Change uses of "Hair" in Render Settings UI in the property editor
and the "Hair Info" node to use the "Curves" name to reflect the
design described in T95355, where hair is just a use case of a more
general curves data type.
While these settings still affect the particle hair system,
the idea is that if we have to choose one naming scheme to align
with, we should choose the option that aligns with future plans
and current development efforts, especially since the particle
system is considered a legacy feature.
A few notes:
- "Principled Hair BSDF" is not affected since it's meant for hair.
- Python API property identifiers are not affected.
Differential Revision: https://developer.blender.org/D14573
The last length value was not initialized, and all length values were
moved one position towards the front of each curve incorrectly.
Also fix an assert when a curve only had a single point.
This revision allows to specify CUDA host compiler (nvcc's -ccbin command
line option) when configuring the build. It addresses the case where the
C/C++ compiler to be used in CUDA toolchain should be different from the
default C/C++ compiler, for instance in case of compilers versions conflicts
or multiple installed compilers.
The new CMake option is named `CUDA_HOST_COMPILER` and can be used as follows:
`cmake -DCUDA_HOST_COMPILER=<path-to-host-compiler>`
If the option is not specified, the build configuration behaves as previously.
Differential Revision: https://developer.blender.org/D14248
When showing an action data-block added to a library overridden object
in the Graph Editor, the visibility toggles would be disabled.
Toggling the visibility should be possible still and works with the
shortcuts, just the button was incorrectly disabled.
Also added the usual disabled hint for the tooltip.
Differential Revision: https://developer.blender.org/D14568
Reviewed by: Bastien Montagne
2022-04-06 18:30:51 +02:00
1715 changed files with 61552 additions and 29158 deletions
The fast_float library provides fast header-only implementations for the C++ from_chars
functions for `float` and `double` types. These functions convert ASCII strings representing
decimal values (e.g., `1.3e10`) into binary types. We provide exact rounding (including
round to even). In our experience, these `fast_float` functions many times faster than comparable number-parsing functions from existing C++ standard libraries.
Specifically, `fast_float` provides the following two functions with a C++17-like syntax (the library itself only requires C++11):
```C++
from_chars_result from_chars(const char* first, const char* last, float& value, ...);
from_chars_result from_chars(const char* first, const char* last, double& value, ...);
```
The return type (`from_chars_result`) is defined as the struct:
```C++
struct from_chars_result {
const char* ptr;
std::errc ec;
};
```
It parses the character sequence [first,last) for a number. It parses floating-point numbers expecting
a locale-independent format equivalent to the C++17 from_chars function.
The resulting floating-point value is the closest floating-point values (using either float or double),
using the "round to even" convention for values that would otherwise fall right in-between two values.
That is, we provide exact parsing according to the IEEE standard.
Given a successful parse, the pointer (`ptr`) in the returned value is set to point right after the
parsed number, and the `value` referenced is set to the parsed value. In case of error, the returned
`ec` contains a representative error, otherwise the default (`std::errc()`) value is stored.
The implementation does not throw and does not allocate memory (e.g., with `new` or `malloc`).
It will parse infinity and nan values.
Example:
``` C++
#include "fast_float/fast_float.h"
#include <iostream>
int main() {
const std::string input = "3.1416 xyz ";
double result;
auto answer = fast_float::from_chars(input.data(), input.data()+input.size(), result);
std::cout << "parsed the number " << result << std::endl;
return EXIT_SUCCESS;
}
```
Like the C++17 standard, the `fast_float::from_chars` functions take an optional last argument of
the type `fast_float::chars_format`. It is a bitset value: we check whether
`fmt & fast_float::chars_format::fixed` and `fmt & fast_float::chars_format::scientific` are set
to determine whether we allow the fixed point and scientific notation respectively.
The default is `fast_float::chars_format::general` which allows both `fixed` and `scientific`.
The library seeks to follow the C++17 (see [20.19.3](http://eel.is/c++draft/charconv.from.chars).(7.1)) specification.
* The `from_chars` function does not skip leading white-space characters.
* [A leading `+` sign](https://en.cppreference.com/w/cpp/utility/from_chars) is forbidden.
* It is generally impossible to represent a decimal value exactly as binary floating-point number (`float` and `double` types). We seek the nearest value. We round to an even mantissa when we are in-between two binary floating-point numbers.
Furthermore, we have the following restrictions:
* We only support `float` and `double` types at this time.
* We only support the decimal format: we do not support hexadecimal strings.
* For values that are either very large or very small (e.g., `1e9999`), we represent it using the infinity or negative infinity value.
We support Visual Studio, macOS, Linux, freeBSD. We support big and little endian. We support 32-bit and 64-bit systems.
## Using commas as decimal separator
The C++ standard stipulate that `from_chars` has to be locale-independent. In
particular, the decimal separator has to be the period (`.`). However,
some users still want to use the `fast_float` library with in a locale-dependent
manner. Using a separate function called `from_chars_advanced`, we allow the users
to pass a `parse_options` instance which contains a custom decimal separator (e.g.,
std::cout << "parsed the number " << result << std::endl;
return EXIT_SUCCESS;
}
```
## Reference
- Daniel Lemire, [Number Parsing at a Gigabyte per Second](https://arxiv.org/abs/2101.11408), Software: Pratice and Experience 51 (8), 2021.
## Other programming languages
- [There is an R binding](https://github.com/eddelbuettel/rcppfastfloat) called `rcppfastfloat`.
- [There is a Rust port of the fast_float library](https://github.com/aldanor/fast-float-rust/) called `fast-float-rust`.
- [There is a Java port of the fast_float library](https://github.com/wrandelshofer/FastDoubleParser) called `FastDoubleParser`.
- [There is a C# port of the fast_float library](https://github.com/CarlVerret/csFastFloat) called `csFastFloat`.
## Relation With Other Work
The fastfloat algorithm is part of the [LLVM standard libraries](https://github.com/llvm/llvm-project/commit/87c016078ad72c46505461e4ff8bfa04819fe7ba).
The fast_float library provides a performance similar to that of the [fast_double_parser](https://github.com/lemire/fast_double_parser) library but using an updated algorithm reworked from the ground up, and while offering an API more in line with the expectations of C++ programmers. The fast_double_parser library is part of the [Microsoft LightGBM machine-learning framework](https://github.com/microsoft/LightGBM).
## Users
The fast_float library is used by [Apache Arrow](https://github.com/apache/arrow/pull/8494) where it multiplied the number parsing speed by two or three times. It is also used by [Yandex ClickHouse](https://github.com/ClickHouse/ClickHouse) and by [Google Jsonnet](https://github.com/google/jsonnet).
## How fast is it?
It can parse random floating-point numbers at a speed of 1 GB/s on some systems. We find that it is often twice as fast as the best available competitor, and many times faster than many standard-library implementations.
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.
