The issue was that the attribute propagation in the Fillet Curve node seems
pretty broken. I couldn't really make sense of the old code. It changed the
size of the point attribute domains on splines to 1 for some reason which
led to a crash in the next node.
Differential Revision: https://developer.blender.org/D13362
The mirror modifiers merge option caused unnecessary re-ordering
to the vertex array with original vertices merging into their copies.
While this wasn't an error, it meant creating a 1:1 mapping from input
vertices to their final output wasn't reliable (when looping over
vertices first to last) as is done in
BKE_editmesh_vert_coords_when_deformed.
As merging in either direction is supported, keep the source meshes
vertices in-order since it allows the vertex coordinates to be extracted.
NOTE: Since this change introduce issues for some cases (e.g. bound
modifiers like SurfaceDeform), this change is only applied to newly
created modifiers, existing ones will still use the old incorrect merge
behavior.
Reviewed By: @brecht
Maniphest Tasks: T93321, T91444
Differential Revision: https://developer.blender.org/D13355
Revert "Fix T91444: Edge Loop Preview fails with two Mirror Modifiers"
This reverts commit 1a7757b0bc.
Caused issue reported in T93321, boiling down to the fact that other
operations or modifiers (like the SurfaceDeform one) rely on the order
of the vertices in the mesh to remain consistent.
Changing this in a modifier would mean those operations need to be
reset/re-created (e.g. rebound for the SurfaceDeform case), which is not
doable in `do_version` code.
Was happening during rendering, causing visual artifacts when doing
CPU+GPU rendering, and giving different in-progress results on different
devices.
The root of the issue comes to the fact that math used in the approximate
shadow catcher calculation might have resulted in negative alpha channel,
and negative values for display are handled differently on CPU and GPU.
Such difference in handling is caused by an approximate conversion used on
the CPU for the performance reasons.
This change makes it so no negative alpha is generated by the approximate
shadow catcher. Not sure if we need some explicit clamping somewhere to
deal with possible negative values coming from somewhere else.
The shadow catcher cornell box tests are to be updated for the new code,
but the new result seems to be more accurate.
Differential Revision: https://developer.blender.org/D13354
Noticed when was looking into T93155. Steps to reproduce:
- Open the .blend file from the report
- Hit F12 to start rendering
- After some tiles were rendered hit Esc
The issue is caused by "sticky" cancel reported via Progress. This means
that once user hit Esc all further requests for cancel state will return
truth, which was preventing OIDN denoiser from completing the denoising
task.
Now only allow stopping the denoiser when interactive rendering requests
a very fast stopping.
Aiming the fix for 3.0 branch.
Differential Revision: https://developer.blender.org/D13352
With the current code in master, scrambling distance is enabled on non-hardware accelerated ray tracing devices see a measurable performance decrease when compared scrambling distance on vs off. From testing, this performance decrease comes from the large tile sizes scheduled in `tile.cpp`.
This patch attempts to address the performance decrease by using different algorithms to calculate the tile size for devices with hardware accelerated ray traversal and devices without. Large tile sizes for hardware accelerated devices and small tile sizes for others.
Most of this code is based on proposals from @brecht and @leesonw
Reviewed By: brecht, leesonw
Differential Revision: https://developer.blender.org/D13042
A parallel loop to create the interesection meshes for each triangle
meant that with parallelism, the output order of the created meshes
could vary with each execution. Keep the parallelism for doing the CDTs
for interesection, but move the extraction of the new faces into a
serial loop afterwards, for repeatability.
This commit improves the scaling of some ui widgets when
zooming by making the radius of the rounded corners
dependent on the element's zoom level.
Needed to fix T92278 without padding issues, see D13125.
Reviewed By: Hans Goudey, Julian Eisel
Differential Revision: https://developer.blender.org/D12842
There are two functions that recalculate the boundbox of an object:
- One that considers the evaluated geometry
- Another that only considers the object's `data`.
Most of the time, the bound box is calculated on the final object
(with modifiers), so it doesn't seem right to just rely on `ob->data`
to recalculate the `ob->runtime.bb`.
Be sure to calculate the BoundBox based on the final geometry and
only use `ob->data` as a fallback
Differential Revision: https://developer.blender.org/D12282
Changes icon used to indicate blend file when overlaid over larger
document icon when in thumbnail view. Only seen when file does not
have a preview.
Followup to {rB611e4ffaab43}
For more details and examples see D13342
Differential Revision: https://developer.blender.org/D13342
Reviewed by Julian Eisel
Changes icon used to indicate blend file when overlaid over larger
document icon when in thumbnail view. Only seen when file does not
have a preview.
Followup to {rB611e4ffaab43}
For more details and examples see D13342
Differential Revision: https://developer.blender.org/D13342
Reviewed by Julian Eisel
Without this it's easy to loose track of which catalog you are dragging.
Things feel generally quite jumpy/disconnected, activating the catalog
makes things feel far less like that.
I consider this an important usability fix, therefore I'm adding it to
the release branch.
When dropping catalogs it is ensured that the name of the moved catalog
is unique within the new parent catalog. When dropping a catalog into
the parent, the catalog would not actually move to a different location,
but it would still be renamed. The unique name logic simply isn't smart
enough to ignore the catalog that is about to be moved.
Address this by disallowing dragging a catalog into its own parent. It's
already there.
This was an oversight when I added catalog drag & drop support. I forgot
to add this for dragging catalogs into the top level by dragging into to
the "All" item as well. This made the drag & drop support rather broken
because it wouldn't work for a basic case.
When a node is executed, it usually schedules other nodes.
Right now, those newly scheduled nodes are added to a
task pool so that another thread can start working on them
immediatly.
However, that leads to the situation where sometimes each
node in a simple chain is executed by another thread. That
leads to additional threading overhead and reduced cache
efficiency (for caches that are not shared between cores).
Now, when a node is executed and schedules other nodes,
the first of those newly scheduled nodes will always be
executed on the same thread once the current node is done.
If it schedules more than one other node, those will be
added to the task pool as before.
The speedup achieved by this is hard to measure. I found it
to be a couple percent faster in some extreme cases, not
much to get excited about. It's nice though that the number
of tasks added to the task pool is commonly reduced by a
factor of 4 or 5.
This part of the drawing code assumes that the bone custom object
has only one evaluated geometry component, and it also uses the
object type to check which data to draw, with the functions like
`DRW_cache_object_surface_get` that just take an object input.
Those functions usually work on evaluated objects, which use the
instancing system to access a temporary object with `object.data`
replaced for data types that don't match the original object.
That assumption used to work, but now curve, point cloud, or volume
objects can have an evaluated mesh which is not accessed with the
same object for render engine drawing.
The "correct" solution for the way this code is structured would be to
loop through all of the geometry components and try to get GPU batches
from every one of them. However, that significantly increases complexity
in an area that should probably be refactored anyway. This patch treats
the mesh as a special case, and only draws the evaluated mesh.
The **best** solution in my opinion might be refactoring this area to
use the instancing system with some sort of viewport-only flag so
the custom shape instances aren't added in the render.
The solution is "partial" because the "Wireframe" option only works
for meshes from mesh objects, even after this fix, and because other
data besides meshes is not displayed at all.
Differential Revision: https://developer.blender.org/D13038
Some enum names were changed/removed in OptiX 7.4, so some changes are necessary to
make things compile still.
In addition, OptiX 7.4 also adds built-in support for catmull-rom curves, so it is no longer
necessary to convert the catmull-rom data to cubic bsplines first, and has endcaps disabled
by default now, so can remove the special handling via any-hit programs that filtered them
out before.
Differential Revision: https://developer.blender.org/D13351
The patch slightly modifies two recently added icons "FILE_BLEND" and
"CURRENT_FILE" to better align them to the pixel grid, and change the
design of "FILE_BACKUP" to avoid alignment and readability issues, as
well as avoiding the outline version of the Blender logo which violates
the official logo guidelines.
Differential revision: https://developer.blender.org/D13346
When a manual frame range is set, allow marking an action as having
Cyclic Animation. This does not affect how the action is evaluated,
but the Cycle-Aware Keying option will automatically make any newly
added F-Curves cyclic. This allows using the option from the start
to build the cycle, rather than only for tweaking an existing loop.
The curves are made cyclic when they have only one key, either
after inserting the first key, or before adding the second one.
The latter case avoids the need to manually make the first added
curve cyclic after marking a newly added action cyclic.
Differential Revision: https://developer.blender.org/D11803
Some operations, e.g. adding a new action strip to NLA, require
knowing the active frame range of an action. However, currently it
can only be deduced by scanning the keyframes of the curves within
it. This is not ideal if e.g. curves are staggered for overlap.
As suggested by Nathan Vegdahl in comments to T54724, this patch adds
Action properties that allow manually specifying its active frame range.
The settings are exposed via a panel in the Dopesheet and Action Editor.
When enabled, the range is highlighted in the background using a striped
fill to distinguish it from the solid filled regular playback range.
When set, the frame range is used when adding or updating NLA tracks,
and by add-ons using `Action.frame_range`, e.g. FBX exporter.
Differential Revision: https://developer.blender.org/D11803
Add a new 'selected_visible_actions' property to allow querying
actions that are selected in animation related editors for use in
UI and operators. The 'selected_editable_actions' variant excludes
linked actions (the only reason an action can be read-only).
In the Action and Shape Key editors there is only one action
that is specified by the field at the top of the editor.
In Dope Sheet it scans the channel rows and returns all actions
related to the selected items. This includes summary items for
actions and groups.
In Graph Editor, it lists actions associated with selected curves.
The new property is also used for Copy To Selected and Alt-Click.
Ref D11803
This is only meant to be used for development purposes for now,
not to show warnings to the user.
Differential Revision: https://developer.blender.org/D13348
Previously, there were a couple of cases where nodes were scheduled when
that was not really necessary. This change doesn't seem to have a big impact
on performance, but simplifies the code a bit.
There are few layers of things which lead to the situation
of more noisy material preview: the do-version of the
preview.blend did not happen (at least from the Python
side as we did not investigate the C side deep). This made
Cycles to use default integrator settings for the preview
file ever since the Cycles X was merged. Those settings are
adaptive sampling with max 4K samples, noise threshold 0.01.
Opening the file in Blender 3.0 for edit did run the versioning
code which effectively lowered the number of samples used for
rendering.
This change makes it so the preview file is configured with
the exact effective settings as seen by Cycles prior to the
file was re-saved (adaptive sampling with the parameters noted
above).
This fix does not chaneg the fact that the versioning code is
not used for preview.blend, it only solves the regression in
the quality of previews.
The fix is done and reviewed with collaboration with Dalai and Sergey.
This dependency was a bit ugly and the functions from the mask modifier
did a few things that we don't need in the Delete Geometry node:
* The mask modifier uses `CustomData_copy_data` which the node doesn't need.
* The mask modifier checks for `-2` in some values, but this special value is
not used by the node.
Differential Revision: https://developer.blender.org/D13335
Unity builds are only used in the `bf_nodes_geometry` module for now.
This module has been prepared to support unity builds already.
Usually, there is a 2-4x speedup when building `bf_nodes_geometry`
compared to without unity builds (e.g. 145s to 55s).
For more information about how unity builds work and how they help
with compile times, see D13341.
Differential Revision: https://developer.blender.org/D13341
Basically, this fixes disappearing previews when editing asset metadata
or performing undo/redo actions.
The preview generation in a background job will eventually modify ID
data, but the undo push was done prior to that. So obviously, an undo
then would mean the preview is lost.
This patch makes it so undo/redo will regenerate the preview, if the preview
rendering was invoked but not finished in the undone/redone state.
The preview flag PRV_UNFINISHED wasn't entirely what we needed. So I had to
change it to a slightly different flag, with different semantics.
Basically, this fixes disappearing previews when editing asset metadata
or performing undo/redo actions.
The preview generation in a background job will eventually modify ID
data, but the undo push was done prior to that. So obviously, an undo
then would mean the preview is lost.
This patch makes it so undo/redo will regenerate the preview, if the preview
rendering was invoked but not finished in the undone/redone state.
The preview flag PRV_UNFINISHED wasn't entirely what we needed. So I had to
change it to a slightly different flag, with different semantics.
