Add a `--profile-gpu` launch argument.
When set, it generates a profile in the Trace Event Format with CPU and
GPU metrics based on GPU debug scopes.
https://profilerpedia.markhansen.co.nz/formats/trace-event-format/
The profiles are best viewed at https://ui.perfetto.dev/
Notes:
- The profiler captures everything form app start to exit.
- Being JSON based the profiles can become relatively large, but they
compress very well.
- Only OpenGL profiling is supported for now, but the report formatting
code can be shared across backends.
Pull Request: https://projects.blender.org/blender/blender/pulls/133557
Add a dedicated callback to duplicate the key for each type.
Keeping changes to the minimum necessary ones, the whole code and data
structure could use a serious update using C++ features, but this would
be totally out of scope of this fix.
Pull Request: https://projects.blender.org/blender/blender/pulls/134523
Instead of passing the `scene`, pass a pointer to the `Editing` struct.
This is the struct that owns the `StripLookup` in its `runtime`
so there isn't a good reason to use the `scene` pointer here
anyways.
Pull Request: https://projects.blender.org/blender/blender/pulls/134743
This reverts commit afec64739a.
The commit introduces a regression where opening the Asset Browser opens
the File Browser instead, same for other editor sub-types, see
blender/blender#134630.
blender/blender!134642 proposes a different solution and reverts this,
which I prefer too. Better to the revert separately from trying a
different fix for the initial bug, makes reviewing easier too.
This removes the `scene` parameter of
`SEQ_time_get_rounded_sound_offset` and
`SEQ_time_media_playback_rate_factor_get`.
Instead, we now pass the `frames_per_second` directly.
This is an attempt at getting less of the code to be
scene dependent.
Pull Request: https://projects.blender.org/blender/blender/pulls/134741
This main thread check was introduced in 4536a4c610. At the time OpenGL
only ever worked in the main thread, but that limitation was lifted in
Blender 2.80.
With this operator being converted to use the jobs system, it now needs
to work in non-main threads. Note that this code does not run in parallel
with main thread drawing anyway, because of WM_job_main_thread_lock_acquire
and draw manager locks. What changed is that we can now activate OpenGL
contexts on other threads.
Pull Request: https://projects.blender.org/blender/blender/pulls/134696
`DEGCustomDataMeshMasks` makes `IDInfo` struct non-trivial, so it should
not be allocated or freed with C-style allocation.
Instead of allocating this data, now use its only persistent user
(the `DepsgraphNodeBuilder::id_info_hash_` map) as owner, move the rest
of the code releasing COW IDs to a new destructor for `IDInfo`, and let
standard C++ destruction do the job.
NOTE: This change seems to give some performances improvements in
affected code (e.g. about 4% improvement in affected area of
`DepsgraphNodeBuilder::begin_build`). However, this does not affect global
performances in a measurable way.
Related to work on improving our memory allocation code in !134463.
Pull Request: https://projects.blender.org/blender/blender/pulls/134563
As announced in <LINK_TO_DEVTALK_POST> (*[proposed post](https://hackmd.io/@s3W_pRGTTXie8cKWRitZrA/S1SzUhqIkl)*), Big Endian support will be
fully removed from Blender 5.0.
This commit adds deprecation warning when opening such blendfiles, and a
static `#warning` when building on a Big-Endian system.
Implements 4.5-part of #125759.
Add `PointerRNA::reset()` and `PointerRNA::invalidate()` utils functions
(both simply reset the PointerRNA data to empty state).
Replace `RNA_POINTER_INVALIDATE` macro by `PointerRNA::invalidate()`.
Follow-up to !134393 and e55d478c64.
Caused by 1bde901bf2
`UI_BTYPE_CURVEPROFILE` button has undo flag set. see button-layout in:
`edbm_bevel_ui() -> uiTemplateCurveProfile()`.
This forces an undo push with the help of `ui_apply_but_undo/ui_apply_but_funcs_after`.
To handle the situation, skip undo push for buttons in hud region.
Pull Request: https://projects.blender.org/blender/blender/pulls/134552
The issue was that with the move to Grease Pencil v3,
annotations and normal GP use different IDs and support to show annotations
wasn't added back in.
This makes it so annotation keys are shown if "Only Show Selected" is disabled,
which works in the Dope Sheet and in the timeline.
In order to not show annotations of the scene twice, the scene-specific code is removed.
The annotations are now shown in a more generic way.
Pull Request: https://projects.blender.org/blender/blender/pulls/134565
Node region is not tagged to redraw when animation slot is changed.
`action_slot` RNA prop has `NC_ANIMATION`notifier set, use the same in
`node_region_listener` to tag region redraw.
Pull Request: https://projects.blender.org/blender/blender/pulls/134435
This reverts 06a2617107.
Commit has introduced few errors from user perspective (#134277) and
python side (#134380). Best to revert it now then handle the bug fixes later.
The error in #134380 was due to `context.layer_collection` and
`active_layer_collection` returning excluded collection.
Pull Request: https://projects.blender.org/blender/blender/pulls/134546
Nothing but improvements to comments in this area of code to better
explain the complexity of Emum list row and column calculation. These
things change wrapping and column count to suit available space and
needs better explanation.
Pull Request: https://projects.blender.org/blender/blender/pulls/134719
For non-categorized lists, rather than always trying to find optimal
rows and columns, revert to prior behavior and only reflow if needed,
when the calculated rows are greater than maximum rows. This is less
disruptive, has less changes at small sizes where this doesn't matter.
Yet still does the job when things get too large to fit, or the list
is very long.
Pull Request: https://projects.blender.org/blender/blender/pulls/134713
The existing behavior of the grid fill operator requires a loop of
boundary or wire edges to be selected. This algorithm tries to fit a
grid of quads into the edge loop. It also works if you select two of the
four 'rails' of an edge loop.
This new behavior allows running grid fill after selecting faces.
If the algorithm sees that a set of faces are selected, it separates
all the faces, and then attempts to grid from their exterior boundary
using the exact same logic it normally uses for interior boundaries.
This all assumes the user has selected a set of faces with an outside
border that works with grid fill. (has one clear exterior loop with an
even number of edges) If not, grid fill will halt, and notify the user
of the error, exactly the way that it already did before.
Once the grid fill is complete and successful,
the existing faces which were replaced get deleted.
UVs (including at UV island boundaries) are interpolated properly,
as are custom-data on verts, edges, faces, and face corners.
At the edges of the selection, the edges/faces/face corners that are
selected contribute, but the custom-data outside the selection does not.
This also resolves non-deterministic behavior.
The previous behavior of `edbm_grid_fill_prepare` found the vertex with
the largest angle, then walked the edge loop until it found one out of
the three best remaining corners. This established one edge of the grid
area, and the second edge was found by symmetry.
However, the direction that the algorithm walked the loop
(and therefore which second corner was found) was essentially arbitrary.
The new behavior first finds the vert with the largest angle,
then finds the vert with the second largest angle,
(after excluding the diagonally opposite corner from the first vert),
and selects the edge between those two verts.
Ref !129318.
With 5956752eb7 the "red alert" color is in the theme. But it is hard
to select a single color that works well for backgrounds and text
colors. This PR lightens the color when used for text overlay showing
the FPS. This must be quite bright as we favor using a contrasting dark
outline.
Pull Request: https://projects.blender.org/blender/blender/pulls/134700
Note: The point cloud bounding box is not being updated once the points are deleted.
This is a known issue with BKE_pointcloud_nomain_to_pointcloud and Hans is looking into it.
Code inspired/built based on the Delete Geometry node.
Pull Request: https://projects.blender.org/blender/blender/pulls/134622
Implement a basic undo system making use of implicit
sharing to avoid copies for stored attribute arrays.
Functionally this has no benefit over memfile undo,
but current undo system architecture requires each
edit mode to implement an undo type.
(Reviewed and co-authored by Hans Goudey)
Pull Request: https://projects.blender.org/blender/blender/pulls/134623
When needing a friendly title and icon for windows and screen areas we
use SpaceType callbacks. For Node editors this could crash for custom
node trees that are unregistered. node_tree_type_find returns nullptr
since this custom tree type is not in the map. In this case just return
translated "Node Editor".
Pull Request: https://projects.blender.org/blender/blender/pulls/134692
Previously we generally expected CustomData layers to have implicit
sharing info, but we didn't require it. This PR clarifies that we do
require layers with non-null data to have implicit sharing info. This
generally makes code simpler because we don't have to have a separate
code path for non-shared layers. For example, it makes the "totelem"
arguments for layer freeing functions unnecessary, since shared data
knows how to free itself. Those arguments are removed in this PR.
Pull Request: https://projects.blender.org/blender/blender/pulls/134578
