Overload resolution must have changed and is causing issues for one
particular code path attempting to use `isfinite(ccl::uchar)`.
Compiler output attached.
It turns out that the code in question can be simplified to just remove
the ambiguity because only the float codepath wants to check for finite
values.
----
Reduced repro: https://godbolt.org/z/YWz3Yc3x8
Pull Request: https://projects.blender.org/blender/blender/pulls/125348
`atan2(0, 0)` is undefined on many platforms. To ensure consistent
result across platforms, we return `0` in this case.
Note only the behavior of the shader node `Artan2` is changed here.
During shading, we might still produce `atan2(0, 0)` internally and
cause different results across platforms, but that usually happens with
single samples and is not obvious, plus checking this condition all the
time is costly. If later we find out it's indeed necessary to change all
the invocation of `atan2(0, 0)`, we could change the wrapper functions
in `metal/compat.h` and `mtl_shader_defines.msl`.
Pull Request: https://projects.blender.org/blender/blender/pulls/126951
This affected cursor snapping as well as the `Jump to Keyframes`
operator since both use `sum_selected_keyframes` to get the averaged
values of elements. It was already taking into account the normalization
factor, but the way a single `KeyframeEditData` was created from
multiple `bAnimListElem` was wrong:
- first, we should only add to it if we actually have some keys selected
(not a problem without normalization, but with it, we would add values
there
- second, the offset needs to be taken into account for each selected
key when adding to the absolute values
Could go into LTS probably
Pull Request: https://projects.blender.org/blender/blender/pulls/125221
When the spread of a spot light is at it's maximum (180 degrees),
then `atan(tan(theta_e))` could become quite unpredictable due to
a asymtote in the tan function when working with lights with this
spread (because theta_e is `spread * 0.5 = pi/2`).
This lead to issues like theta_e for the spotlight becoming negative,
which lead to rendering errors due to a malformed light tree.
This commit fixes this issue by adding a episilon region around the
troublesome values and sets theta_e to `pi/2` when in that region.
Candidate for backporting to 4.2 and potentially 3.6
Pull Request: https://projects.blender.org/blender/blender/pulls/125172
area light with zero spread was introduced in bf18032977. Such paths can
only be sampled with NEE, so MIS should not be used.
This fixes the discrepancy when Direct Light Sampling is set to MIS or NEE.
Pull Request: #118584
SYCL runtime currently relies on an internal driver behavior that will
break the driver version string returned by SYCL if it changes:
https://github.com/oneapi-src/unified-runtime/issues/1777
This will be fixed at SYCL runtime level but until we use a new enough
one, we need to add additional verifications to avoid blocking execution
on a driver that will change this internal behavior.
Pull Request: https://projects.blender.org/blender/blender/pulls/124084
STL/PLY (also Collada) use `BKE_mesh_assign_object` to assign a mesh
(already in main, has a usercount of 1) to a fresh object.
That function does a bunch of (unneeded) things (test modifiers/
materials which is not necessary since these are fresh objects) next to
increasing usercount. Collada steers against this by reducing usercount
again. Other importers such as alembic assign the mesh directly to
object data (which is also what this PR proposes).
Pull Request: https://projects.blender.org/blender/blender/pulls/123558
Related to #119999 and #122059.
In case the _source_ PropertyRNA was unset (i.e. its underlying
IDProperty storage did not exist), the copy operation would silently
fail.
In fact, the existing code handling IDProperties separately in
`RNA_property_copy` was pretty bad, since it would also bypass all the
RNA 'setting value' code (like custom setters, update handling).
Turns out, liboverride RNA apply code can already handle all of these
cases, so simply pass the raw 'unresolved' RNA property to it, and
remove all this special handling code from `RNA_property_copy`, solves
all the issues.
A fully unset (i.e. null IDProperty storage pointer) dynamic RNA
property would not be properly handled by RNA copy code path. This led
to having a type mismatch (because dynamic RNA booleans are stored in
integer IDProperies) betwen source and destination RNA properties.
Since the fix in 6adf5adb7f, such type mismatch will abort the copying
process.
This is fixed by updating the `PropertyRNAOrID` internal struct to also
store whether the given property is actually a dynamic RNA one with
IDProperty storage.
NOTE: Potential side-effects: That kind of change is very difficult to
fully analyse, but no side-effects from this fix are expected. The only
code which behavior is changed is `rna_ensure_property_realdata`, which
is only used by `RNA_property_copy` currently.
This change to `PropertyRNAOrID`/`rna_property_rna_or_id_get` _may_ be
useful in the future though, e.g. in liboverride codepath...
CTX_data_ensure_evaluated_depsgraph should not be used in drawing code.
This triggered a depsgraph updated which in turn called
DRW_notify_view_update. This would bind the GPU context a second time
and hang due to recursive mutex calls.
Pull Request: https://projects.blender.org/blender/blender/pulls/123024
A bit on shaky ground here, but it looks like we actually dont want to
execute the hovered buttons value/op if we chose another item in the pie
menu. This can be achieved by using the **onfree** arg to
`button_activate_exit` called on the active(hovered) button.
Not sure if this is the correct solution tbh (but maybe it demonstrates
where the bug of #122526 can be located).
I have tested pie menus with this patch and they seems to behave
normally (even nested ones).
Pull Request: https://projects.blender.org/blender/blender/pulls/122567
Pull Request: https://projects.blender.org/blender/blender/pulls/122998
Code was getting a direction (to grow in) from the last point in a curve
and another point in the middle of the curve.
On a curve with only two points, these points were the same, resulting
in a zero delta and thus doing nothing.
To resolve, take the root point instead in this scenario to get a valid
direction.
Pull Request: https://projects.blender.org/blender/blender/pulls/122644
The cloth scale Filter relies on properly set up deformation constraints
(this is done in `do_cloth_brush_build_constraints_task`).
Whenever the Brush (not the Filter) is set to `SCULPT_TOOL_CLOTH`
(having it selected prior to the filter), another code path is taken and
the setting up the proper constraints is skipped (making the Filter
disfunctional).
To distinguish between filter usage and brush usage, there were already
checks for the `StrokeCache` (indicating brush usage afaict), so to
resolve, use the same check to prevent entering the particular code path
associated with the `SCULPT_TOOL_CLOTH`.
Probably an oversight in c2f0522760
Pull Request: https://projects.blender.org/blender/blender/pulls/122440
On export, PLY creates a matrix (in `set_world_axes_transform`) -- the
inverse transpose of the regular matrix [seems like the usual way of
transforming normals]] -- by which the normals are multiplied. This can end
up in non-normalized custom normals on scaled objects though. Corrected
in this PR by just normalizing after said multiplication.
On import, `BKE_mesh_set_custom_normals_from_verts` is used with the raw
data -- which ends up in `mesh_normals_corner_custom_set` which in turn
"is expected to have normalized normals" (from the comment).
We _could_ also make sure to normalize on import, however, setting these
properly on export seems the primary choice.
Other importers also dont go the extra route of making sure to normalize
the incoming data, so this seems to be in line of what other Im-/Exports
do.
Pull Request: https://projects.blender.org/blender/blender/pulls/122432
`template_search` was introduced in c20c203b82
Seems it assumed an active item is always there.
Without an active item, we are running into a crash trying to get the
textbut width.
So solution is to not draw that text button at all until an active item
is chosen.
This also goes in line with what `template_ID` is doing (it also skips
the text button without an active item).
Pull Request: https://projects.blender.org/blender/blender/pulls/122165
`Weight` & `Threshold` sliders should **not** show in case of `Type` :
`Custom Normal` and **should** show for both `Type` : `Face Area` &
`Type` : `Corner Angle`
In code, it looks like we are gathering `loop_weight` with `val`.
- this is always 1.0 for `EDBM_CLNOR_AVERAGE_LOOP`
- this is taken from `BM_face_calc_area` for
`EDBM_CLNOR_AVERAGE_FACE_AREA`
- this is taken from `BM_loop_calc_face_angle` for
`EDBM_CLNOR_AVERAGE_ANGLE`
Code then compares not equal those values with given threshold, but for
`EDBM_CLNOR_AVERAGE_LOOP` this will never trigger (since all values are
the same), thus `count` is always zero which makes the effective
`n_weight` always 1. So all loop split normals are averaged for a vertex
with the same weight (seems to make sense to me -- at is just plain
average)
Long story short: the condition to show `Weight` & `Threshold` sliders
is just flipped (these only apply for the methods that take neighbor
faces into account).
Pull Request: https://projects.blender.org/blender/blender/pulls/121864
The issue was that when adding a constraint through
python to an object that is already animated,
the constraint did not evaluate properly at first.
The fix is to update the depsgraph when creating a
constraint through the API.
Such an update already happened when running the
operator to create a constraint.
Pull Request: https://projects.blender.org/blender/blender/pulls/122198
Remapping of internal ID pointers in 'undo swap' case (i.e. when new ID
data had to be read from undo buffer, but current ID address is kept)
needs to also ignore 'fake user' handling, otherwise it would reverse
the undoing effect.
Since [0], checking if the user preferences exists used an uninitialized
string in the unlikely situation that the users configuration directory
couldn't be found.
[0]: 0c62906a41
There are still node versioning bits running after linking, for example
`version_geometry_nodes_replace_transfer_attribute_node`. This code
relies on node types being untouched at that point.
Move the unknown node type replacement from `ntreeSetTypes` into the
`ntreeUpdateAllNew` function. This is called _after_ late versioning.
Pull Request: https://projects.blender.org/blender/blender/pulls/117441
Exposed by 1ea169d90e
Colladas `MeshImporter::mesh_add_edges` is a copy from `mesh_add_edges`
in `source/blender/editors/mesh/mesh_data.cc` (done in 2c05190374 --
there is a comment to deduplicate this into BKE, but this wont be done
in this patch).
However, as time passed there have been additions to the later, namely:
- 353fe610ed (clearing the mesh runtime caches)
- 12becbf0df (altering selection handling, guess this can be ignored
here since collada generally does not seem to select geometry)
So in order to resolve, now also clear the caches for collada as in
353fe610ed (even though just the subset of just `mesh->runtime-
>loose_edges_cache.tag_dirty()` would also be enough).
Pull Request: https://projects.blender.org/blender/blender/pulls/120737
While fixing incorrect behavior, this commit did not address any
critical issue, and by modifying the behavior of an existing feature, it
can break some workflows.
Such breackage should be avoided in LTS's corrective releases.
Fix#120788: Object > Library Override > Make doesn't create editable overrides.
This reverts commit 21a54d5b58.
The issue seems to be caused with the recent NVidia drivers, which will
crash if OptiX context is created prior to the OpenGL context on Linux.
This happens with drivers 545.29.06, 550.54.14, 550.67 and kernel 6.8
on Ubuntu 24.04, and also happens with NVidia driver 550.67 on Gentoo.
While it does seem the issue is likely on the driver side, the timeline
for it being resolved there is unknown. Until then it is possible to
apply workaround on Blender side, which will initialize GPU context
prior to rendering with an external render engine when it is known the
GPU context will be needed later on.
Note this is not a complex fix, because in general it's still possible to
render a frame without and then with grease pencil, in which case OptiX
will still be initialized first. But we don't want to always initialize
OpenGL, and we can't predict future operations.
Pull Request: https://projects.blender.org/blender/blender/pulls/120026
If a bone was specified (but now missing), the driver would fallback to
using the object as a target (which can lead to unintended behavior).
So now check if a bone is specified, if it is missing, mark the driver
invalid. If no bone is specified at all, then use the armature object as
the target.
NOTE: `DTAR_FLAG_INVALID` is not granular enough to distinguish the
object and bone targets, so both will be marked in red in the UI (there
is already comments about it in code). If necessary, we could introduce
an additional DTAR_FLAG_BONE_INVALID and use that in a couple of places.
Pull Request: https://projects.blender.org/blender/blender/pulls/119533
This could happen if objects were hidden after going to editmode and
knife was used on them.
In that case, raycasting would (rightfully) fail, but there is a
fallback in place using the back-buffer selection method (in which the
hidden object is still present). So a face would be found, but this
makes all following code confusing/wrong since we are working with
coordinates / faces under the assumption there would be a valid/
corresponding object to it in `KnifeTool_OpData` > `objects` -- which is
not the case...
So to resolve, just check if the object is visible before calling
`EDBM_face_find_nearest` in knife code.
Alternativeliy, we could also add a check for Base viewport visibility
to all the bmesh `find_nearest` functions (a bit hesitant though since i
am not sure this would be desired in all cases)
Pull Request: https://projects.blender.org/blender/blender/pulls/119383
Resolves an issue with stroke rendering in
Metal using the geometry shader fallback
path. Stroke rendering now matches OpenGL
which should enable the GPencil fill tool to
function correctly at all zoom levels.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/119660
Buttons for mesh symmetry on the toolbar, and other buttons of the same
functionality located in other areas of the window, weren't updated when
clicking on them.
Pull Request: https://projects.blender.org/blender/blender/pulls/118508
a958ae36e8 introduced support for UI lists to reference items that would
never be shown, regardless of filter settings. This was to skip assets
in the asset view template that were not of the requested type. UI list
sorting code wasn't updated to account for such items that should be
entirely ignored.
Pull Request: https://projects.blender.org/blender/blender/pulls/109157
