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
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
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.
