`SequencesMeta.new_movie()` API function caused meta strip to change
length. Similar issue has been fixed in transform code by checking
if `MetaStack` exists. `MetaStack` is not used when changing data in
python.
Provide `seqbase` to `SEQ_time_update_sequence()` so the function can
check if change happens inside of meta strip.
This patch also merges `seq_time_update_sequence_bounds()` into
`SEQ_time_update_sequence()`. This is because same issue applies for
both functions and it is confusing to have more time update
functions.re if this will lead anywhere.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D12763
Previously the storage here was optimized to avoid indirections in BVH2
traversal. This helps improve performance a bit, but makes performance
and memory usage of Embree and OptiX BVHs a bit worse also. It also adds
code complexity in other parts of the code.
Now decouple triangle and curve primitive storage from BVH2.
* Reduced peak memory usage on all devices
* Bit better performance for OptiX and Embree
* Bit worse performance for CUDA
* Simplified code:
** Intersection.prim/object now matches ShaderData.prim/object
** No more offset manipulation for mesh displacement before a BVH is built
** Remove primitive packing code and flags for Embree and OptiX
** Curve segments are now stored in a KernelCurve struct
* Also happens to fix a bug in baking with incorrect prim/object
Fixes T91968, T91770, T91902
Differential Revision: https://developer.blender.org/D12766
Was just comparing this item's and the parent item's names. But if an item has
no parents, only its own name has to match for the check to return true. Make
sure that the number of parents also matches.
MSVC does not support variable size array definition.
Use maximum possible stack, similar to the GPU case.
Not expected to have user-measurable difference.
Now the icons to add or delete catalogs are only shown when mouse hovering a
catalog item in the tree. This is convenient for quick creation of catalogs,
and doesn't require activating a catalog to edit it first.
Determining if a tree item is hovered isn't trivial actually. The UI tree-view
code has to find the matching tree-row button in the previous layout to do so,
since the new layout isn't calculated yet.
The UI code to ensure consistent button state over redraws was just comparing
the name of the item, ignoring the parent names. So with multiple items of the
same name, there might have been glitches (didn't see any myself though).
There's a leftover to-do though, we don't check yet if the matched buttons are
actually from the same tree. Added TODO comment.
Make volume stack allocated conditionally, potentially based on the
actual nested level of objects in the scene.
Currently the nested level is estimated by number of volume objects.
This is a non-expensive check which is probably enough in practice
to get almost perfect memory usage and performance.
The conditional allocation is a bit tricky.
For the CPU we declare and define maximum possible volume stack,
because there are only that many integrator states on the CPU.
On the GPU we declare outer SoA to have all volume stack elements,
but only allocate actually needed ones. The actually used volume
stack size is passed as a pre-processor, which seems to be easiest
and fastest for the GPU state copy.
There seems to be no speed regression in the demo files on RTX6000.
Note that scenes with high nested level of volume will now be slower
but correct.
Differential Revision: https://developer.blender.org/D12759
This ended up being a copy of:
`toolsettings->workspace_tool_type == SCE_WORKSPACE_TOOL_FALLBACK`
requiring boiler plate assignment in gizmos refresh callbacks.
Remove this struct member and check `toolsettings->workspace_tool_type`
directly, since so far there has been no advantage in gizmo-groups
being able to control this themselves.
Resolve regression in c9d9bfa84a,
which added support for tools to be tagged as using a fallback too.
In these cases the "Active Tool" setting was ignored and the fallback
tool would be used (the spin tool would box select for example).
Catalogs can now be renamed by double clicking them in the Asset
Browser. This is mostly done through the tree-view API, the asset
specific code is very little.
There is some polish left to be done here, e.g. the double click
currently also collapses/uncollapses and activates the clicked item. And
the rename button takes the full width of the row. But addressing these
is better done as part of some other behavioral changes that are planned
anyway.
It's unclear why this code was added in the first place, but it seems
unnecessary, it can be restored if we find this breaks something.
The Embree docs mention that the same primitive may be hit multiple times, but
my understanding is that about e.g. curves where both the frontside and backside
may be hit. However those hits would be at different distances.
The context for this change is that we want to add an optimization where we
can immediately update throughput for transparent shadows instead of recording
intersections, and avoid duplicate would require extra work. However there is
an Embree example that does something similar without worrying about duplicate
hits either.
Makes things look more appealing visually. Plus it's a way to visually
group the tree rows together, which can be important if there are more
widgets surrounding the tree.
Liboverride references need a special handling during append, since
those pointers should never be made local, nor reampped to newly
localized data. And liboverride references should never be directly made
local either, to ensure their liboverride usages remain pointing to
linked data and not local one.
Issue was reported by the studio, and also probably as part of T91892.
If a renaming button was removed via `UI_but_active_only_ex()` and that
button was placed using the layout system, the button was still in the
layout.
So far this didn't cause issues, because all cases where the button may
be removed were not using the layout system.
Fixes:{T91064}
Caused by {rBcd118c5581f482afc8554ff88b5b6f3b552b1682}
- Applies `ensure_valid_reflection()` to the normal input on all BSDFs for CPU and GPU.
- This doesn't affect hair.
- Removes `ensure_valid_reflection()` from the output of Bump Map and Normal Map nodes for CPU/GPU as it is not needed.
- The fix doesn't touch OSL.
Reviewed By: brecht, leesonw
Maniphest Tasks: T91064
Differential Revision: https://developer.blender.org/D12403
- Support toggle/deselect/deselect_all options
(matching 3D viewport object selection).
- Support legacy selection behavior.
- Support selecting by the center in preview views (holding Ctrl).
Re-order common sequencer key-map to be at the top level
(shared by preview and sequence view).
Without this sequencer tools would be displayed at different levels
in the hierarchy which is confusing and doesn't represent the separation
between "Sequencer" and "SequencerPreview" key-maps.
Sequencer poll was succeeding outside of a preview region.
This meant it couldn't be used in tool key maps which are currently
shared between preview & sequencer regions.
Free strip `anim` data immediately after rendering. This doesn't affect
rendering performance, because each new loop would have to seek to start
of strip. Also strips are now freed anyway, but after rendering loop
ends.
With SF edit file, thumbnail rendering used around 60GB of memory.
Now it uses few hundreds MB (depends on movie file resolution, codec,
etc.)
Freeing of strips caused UI to be unresponsive for brief period. This
issue is not removed, but is more spread out so it is less noticable.
This effectively undoes some of the following commit:
rB4537e8558468c71a03bf53f59c60f888b3412de2
The tables in question were duplicated 5-6 times into the blender
executable due to the headers being used in multiple translation units.
This contributes ~6.3kb worth of duplicate data into the binary.
Some further details are in the below revision.
Differential Revision: https://developer.blender.org/D12724
Some software may export velocity as a different type than 3D vectors
(e.g. as colors or flat arrays or floats), so we need to explicitely
check for this.
A more robust attribute handling system allowing us to cope with other
software idiosyncrasies is on the way, so this fix will do for now.
This didn't belong on blenlib since it uses DNA data types
and included a bad-level call to BKE_curve.h.
It also meant linking in blenlib would depend on the freetype library,
noticeable for thumbnail extraction (see D6408).
The long term goal is completely removing DerivedMesh, and these
functions are making some refactoring of mesh normals (T91186) more
complicated. They are not used anywhere.
