Blender 4.0 requires OpenGL 4.3 which always support SSBO's.
Platforms that don't support enough SSBO bind points will be marked
as unsupported.
Users who start Blender on those platforms will be informed via a
dialog. This PR also updates the `--debug-gpu-force-workarounds`
to match our minimum requirements. Note that some bugs are still
there that should be solved in other PRs:
* Workbench only renders the object using a unit matrix this is because
there is a bug in the workaround for shader_draw_parameters
* Navigating with middle mouse button is not working. Unsure what the
cause is, but might be a missing feature check in the OpenGL backend.
Related to #112224
Pull Request: https://projects.blender.org/blender/blender/pulls/112572
Use magenta fallback color for sockets without a `draw_color_simple`
callback. This is not ideal, but without a context or node instance
the older `draw_color` callback can't be used.
Pull Request: https://projects.blender.org/blender/blender/pulls/112658
The goal is to reduce redundancy by abstracting over the different types of node
tree zones. This makes it easier to add new zone types and makes the intend of
code more clear. For example, now it is more obvious what code deals with zones
in general and what does simulation specific things.
Pull Request: https://projects.blender.org/blender/blender/pulls/112531
It was possible to access a mutable drawing from a const GreasePencil
object. This sort of mistake is possible because DNA requires using raw
pointers.
This is a compatibility breaking change to rename all usages of the name
`asset_library_ref` with `asset_library_reference`. Brecht recently
suggested that such abbreviations should be avoided in public API names.
f6a6b27ac1 made the asset representation type available through context
wherever asset handle was previously. This moves us closer to replacing
the asset handle type.
Part of #102877 and #108806.
Makes the asset representation type available in RNA/BPY context
whenever the asset handle type is, so that it can be used instead. See
d04cd3f3e6. With this change we can now replace virtually all usages of
the asset handle type in Python with the asset representation. (Only for
the asset view template we require a collection property taking asset
handles still, for internal reasons.)
Idea is now to first get rid of all usages of asset handle in Python,
so that there's almost no need for further compatibility breaking
changes (unsure if the asset view template can be removed for 4.0
already though). Internal hacks related to it can be removed at any time
still.
Part of #102877 and #108806.
In #112326 the socket visibility functions were updated to take the
open/closed state of panels into account for visibility of the socket
icon. However, in "hidden" (collapsed) nodes the panels should be
ignored entirely, drawing all sockets on the root level. This requires
looking at the node flags to determine socket icon visibility, so a
simple method of `bNodeSocket` is not sufficient.
This patch moves the more complex visibility queries for sockets into
`bNode`, where both node and socket flags can be accessed. These should
be used for actual visibility rather than the plain flag accessors on
`bNodeSocket`.
Renamed `is_visible_or_panel_closed` back to just `is_visible`, the
other `is_visible` variant is now integrated in `bNode::is_socket_drawn`.
Pull Request: https://projects.blender.org/blender/blender/pulls/112520
This makes it clearer other "safe" functions should be used in
combination with the resulting offsets.
Also correct doc-string which wasn't updated from the "or_error()"
version of this function.
- Moving the cursor to the beginning/end of the line didn't work
with word-wrap enabled.
- Moving the cursor up/down without word-wrap enabled
didn't maintain the column.
Resolve using column conversion functions with tab support.
Strings that include Latin1 encoding or corrupt UTF8 byte sequences
could read past the buffer bounds (stepping over the null terminator).
Resolve by passing in the string length.
Other changes to support non-UTF8 byte sequences:
- BLI_str_utf8_offset_{to/from}_index were accumulating
the UTF8 offset without accounting for non-UTF8 characters
which could cause a buffer underflow or enter an eternal loop.
- BLI_str_utf8_offset_to_index would read past the buffer bounds if the
offset passed in if it was in the middle of a UTF8 byte sequence.
Since 34b4487844, attributes are always made mutable when
accessed from the RNA API. This can result in unnecessary copies, which
increases memory usage and reduces performance.
Cycles is the only user of the C++ RNA API, which we'd like to remove
in the future since it doesn't really make sense in the big picture.
Hydra is now a better alternative for external render engines.
To start that change and fix the unnecessary copies, this commit
moves to use Blender headers directly for accessing attribute and
other geometry data. This also removes the few places that still had
overhead from the RNA API after the changes ([0]) in 3.6. In a simple
test with a large grid, I observed a 1.76x performance improvement,
from 1.04 to 0.59 seconds to extract the mesh data to Cycles.
[0]: https://wiki.blender.org/wiki/Reference/Release_Notes/3.6/Cycles#Performance
Pull Request: https://projects.blender.org/blender/blender/pulls/112306
- Add IDP_EnsureProperties,
- Remove create_if_needed argument from IDP_GetProperties.
Split access & creation so intention reads more clearly without
looking up function arguments.
The `ScopedModifierTimer` would work for "real" curve modifiers, but was
skipping geometry node modifiers.
Now move it up so geometry node modifiers are included.
Pull Request: https://projects.blender.org/blender/blender/pulls/112420
#112019 included open/closed state of the parent panel in socket
visibility calculation. This prevents dragging links, but also disables
other features that should still work, such as drawing links.
A narrower condition is needed for icon visibility vs. general socket
visibility. The cases which use the new `is_icon_visible` condition:
- Drawing socket selection outlines (same as unselected sockets)
- Drawing multi-input sockets (same as unselected sockets)
- `node_find_indicated_socket`, used by a wide range of mouse click
operators, including the link-drag operator that was cause for
#112019.
Cases using the original `is_visible` (true even if parent panel is
collapsed):
- `nodeLinkIsHidden` draws links only when at least one socket is
visible.
- `node_update_basis`, sockets still added to layout even if icon isn't
rendered.
- `node_update_hidden`, panels are ignored for "hidden" nodes, all
sockets are rendered.
- `NODE_OT_link_make` operator for finding "best" sockets to connect.
- `node_link_viewer` finding sockets to connect to a viewer node.
- `get_main_socket` used for insert-on-links (find sockets to splice
into) and some shader previews
- `node_gather_link_searches`, suggestions for adding a new node at the
end of a link.
Pull Request: https://projects.blender.org/blender/blender/pulls/112326
See #101778
Remove the requirement of restarting Blender to refresh the
extended 3D view menus for node group changes. Also avoid
rebuilding the tree of relevant assets and catalogs on every
redraw, since parsing asset libraries, etc. could become more
expensive than we want. Those two goals combined mean we
have to be more rigorous in how we invalidate the cached
catalog tree.
The first main change required is to clear the tree as asset libraries
are being read, similar to other dynamic asset menus. This is done
with a 3D view header listener rather than a menu listener in this case.
However, that isn't enough, because there is an issue with the asset
system where the "all" library isn't updated when the current file library
changes. The solution is to explicitly rebuild the "all" library's catalogs
when other asset libraries are changed.
The other necessity for dynamic updates is clearing the catalog tree
to be rebuilt when the node group "asset traits" are changed. This is
done with a new notifier type (with the goal of being a bit selective
about when we re-read assets). This _also_ requires running the
"presave" callback that builds asset metadata when updating the
property. Otherwise saving the file and sending the notifier is
necessary, which is too confusing.
Pull Request: https://projects.blender.org/blender/blender/pulls/112166
Calling an API function after the node panels patch does not internally
tag the node tree with `NTREE_CHANGED_INTERFACE` any more, because the
node tree is not directly accessible from `bNodeTreeInterface`. Before
node panels the API functions for interfaces could tag the tree directly
for later update consideration, which now requires explicit tagging
calls.
The fix is to add a flag and mutex directly to `bNodeTreeInterface`, so
API methods can tag after updates. This mostly copies runtime data
concepts from `bNodeTree`. The `ensure_interface_cache` method is
equivalent to `ensure_topology_cache` and should be called before
accessing `interface_inputs` and similar cache data.
Pull Request: https://projects.blender.org/blender/blender/pulls/111741
There were two issues:
* The check for conflicting AOVs was done after the Render Layer node
was updated. This led to an unexpected state in the node.
* The check for conflicting AOVs did not work, because AOVs that already
had the conflict-flag set were ignored.
In 2788fa915b the `CurvesGeometry::blend_write` and `CurvesGeometry::blend_read` functions were added. Unfortunately, the commit also altered the writing logic and introduced a bug where loading a file with a converted `Curves` object would crash. See https://projects.blender.org/blender/blender/issues/112068.
This PR fixes the issue by making sure that `CustomData_blend_write_prepare` is called before `BKE_id_blend_write`, which is the root cause of the crash that happens on load.
The `CurvesGeometry::blend_write` function is split into `CurvesGeometry::blend_write_prepare` and `CurvesGeometry::blend_write`.
Pull Request: https://projects.blender.org/blender/blender/pulls/112280
The previous limit of 63 bytes for the panel name was an issue for
languages where characters use multiple bytes. The UI would show panel
names truncated that had reasonably long/short names.
It's easy to support dynamically sized strings here, so do that instead
of using fixed size buffers.
Fixes#111927.
Pull Request: https://projects.blender.org/blender/blender/pulls/111979
The RNA attribute API did not make sure that attributes are unshared when
the data is accessed. This is necessary to avoid accidentally changing data
on other geometries.
Unfortunately, we currently can't detect if the access is read-only or if the
attribute data is modified. Therefore, the data is always copied when it
was shared. This makes it a little bit less efficient in some cases, but that
should still be ok. The old behavior could by brought back by implementing
#112150.
Pull Request: https://projects.blender.org/blender/blender/pulls/111514
Implementation of the duplicate and move operator for grease pencil frames.
The `ACTION_OT_duplicate` operator now creates duplicates of the selected frames and stores them in the layer runtime data.
The `TRANSFORM_OT_transform` operator is updated to take care of these duplicated frames.
Pull Request: https://projects.blender.org/blender/blender/pulls/111051
Fix#100718: NLA Hold Forward Inconsistency
Action Track with 'extrapolation=Hold Forward' behaves the same as 'Hold'.
For the Action Track, we now properly treat extrapolation Hold_Forward just like the rest of the NLA system.
Co-author Wayde Moss @wbmoss_dev
Pull Request: https://projects.blender.org/blender/blender/pulls/109182
We need a separate array that we can change in during the parallel
group construction. That array tells where in each group the index
is added. Building this array is expensive, since construcing a new
`Array` fills its elements serially. There are two possible solutions:
1. Use a copy of the offsets to increment result indices directly
2. Rely on OS-optimized `calloc` instead of `malloc` and a copy/fill
Both depend on using `fetch_and_add` instead of `add_and_fetch`.
The vertex to corner and edge to corner map creation is optimized
by this commit, though the benefits will be useful elsewhere in the
future.
| | Before | 1. offsets copy | 2. calloc |
| -------- | ------- | --------------- | --------------- |
| Grid 1m | 3.1 ms | 1.9 ms (1.63x) | 1.8 ms (1.72x) |
| Grid 16m | 51.8 ms | 33.3 ms (1.55x) | 32.7 ms (1.58x) |
This commit implements the calloc solution, since it's slightly faster
and simpler. In the future, `Array` could do this optimization itself
when it detects that its fill value is just zero bytes.
Pull Request: https://projects.blender.org/blender/blender/pulls/112065
Assert that the partial is part of the particle system before
creating an index. This is a hint to developers that the cast
is safe and using a pointer offset isn't going to cause problems.
Issue raised by !111193.
Co-authored-by: Loren Osborn <linux_dr>
When using a pointer offset that may be out-of-bounds the integer
must not truncate the offset.
Issue raised by !111193.
Co-authored-by: Loren Osborn <linux_dr>
#111972 made sure copied sockets have unique identifiers, but it didn't
cover the case of copying a panel and its children.
- Added an optional `uid_gen` argument to item copy functions which
generates new unique identifiers for copied items. If not specified
the items will retain the original identifiers (e.g. when copying an
entire node tree).
- Removed the `copy_items` panel function from the API. Only used
internally and requires a uid generator now.
Pull Request: https://projects.blender.org/blender/blender/pulls/112074
Batch deletion of IDs could lead to deleting ObData ones while keeping
the related ShapeKeys.
Orphaned shape keys are not allowed anymore in Blender, they are checked
against in both file write and read code.
This would lead to assert (and crash) e.g. in the liboverride code.
This commit forcefully add shapekeys of deleted meshes, curves etc.,
when calling e.g. `BKE_id_multi_tagged_delete`.
Note that deleting the shapekey when deleting the obdata ID was already
implemented in single ID deletion (`BKE_id_delete` & co), in the
underlying private `id_free`. But this is skipped in `no main` case.
which is used by the batch deletion code for performance optimizations.
When data-block/asset previews were not stored in the thumbnail cache
yet (or were outdated), we'd read them from .blend files. This could
lead to random crashes (but quite reliable with a small number of
previews to be read).
Wasn't clearing runtime memory which could lead to the
`PRV_TAG_DEFFERED` bit being set. This meant we would try to free
deferred preview data since eefee47a8a, which was just garbage memory.
The `Library.tag` data is supposed to be runtime only, so needs to be
cleared when reading data from blendfile.
This fixes the 'need resync' status of a library staying around forever
once set, annoying visual warning in the Outliner.
