This was caused by the shadow map Z range loosing too much
precision on the large object.
This can be replicated by adding a very large object in the scene.
The normaly un-shadowed objects gets shadowed because the
float precision is not enough to represent their depth correctly.
To fix this, we bias the shadow depth by 2 ULP to make sure that
there is no self shadowing.
This bias has already been there, but was somehow removed because
of other fixes that simingly fixed this issue. But they did not.
Pull Request: https://projects.blender.org/blender/blender/pulls/137555
The Clay Strips brush calculates a brush plane and local matrix to use
for a cube distance test. This test is done based on the sampled plane
center, not the current cursor position.
To make the node mask match this brush area, this commit reworks the
node mask calculation similar to the Plane brush, also calculating the
plane normal and plane center to be used later.
Finally, the relative radius scale of the brush is fixed to be a
constant SQRT(3) to ensure that the cube influence area is fully
circumscribed by the spherical distance query.
Further work here can change the type of BVH distance query done to
reduce the number of nodes processed by the brush.
Resolves#123768
Related to #84169
Pull Request: https://projects.blender.org/blender/blender/pulls/137371
We have an icon type called ICON_TYPE_VECTOR that are dynamically drawn
at runtime. These are created in icon_draw_size, which takes float
positions, but their callbacks all use integers instead. Truncating so
early in the process takes away flexibility to draw these at more ideal
sizes or to center them better. The complaint is a situation where we
mix ICON_TYPE_VECTOR for the colored collections and ICON_TYPE_SVG_MONO
for the white ones. Even though they both ultimately just draw the same
SVG icon the former way truncates position and size so there are times
they won't exactly line up.
Pull Request: https://projects.blender.org/blender/blender/pulls/132338
Making local an intermediate collection in a liboverride hierarchy e.g.
will likely break the link between the objects in that collection, and
their hierarchy root ID (typically the top overridden collection).
This is fixed by this commit, by:
* Adding a new util, `lib_override_root_is_valid`, that check whether a
given liboverride ID is still 'connected' to its hierarchy root.
* Using this test in `BKE_lib_override_library_main_hierarchy_root_ensure`
to strengthen the detection of invalid root hierarchy pointers.
Part of #82877
This commit applies the same calculation used when tilting the normal to
the inactive cursor display inside sculpt mode, as a way to indicate to
users how the normal is affected by the current tilt settings and pen
orientation.
Pull Request: https://projects.blender.org/blender/blender/pulls/137250
Pen tilt values come into Blender through ghost platforms and device
drivers with inconsistent and ambiguous directions. This PR codifies
that we consider tilt-x positive values to be rightward, negative
values as leftward. Similarly tilt-y positive is away from the user,
negative is toward the user. Comments are updated to make this clear.
This fixes Windows Ink (currently always returning left and away). This
also reverses the Wayland tily-y values to match. We might still have
to consider a user preference to swap directions if we find differences
within a platform.
Pull Request: https://projects.blender.org/blender/blender/pulls/137501
In the past couple months, the basic operations needed for a new object
type were implemented, as well as basic edit mode tools. With that, the
object type is in a good enough state that it doesn't need to be hidden
behind an experimental option. That also resolves some inconsistency,
because the object type is already created by geometry nodes and used by
importers; it's weird that you can't add it manually as original data.
This also removes the "bounds" drawing flag for the object created from
the add menu. It's unnecessary now and there was even a TODO comment.
Pull Request: https://projects.blender.org/blender/blender/pulls/137546
Shortcut is actually assigned to the operator but inside wrong keymap
("Animation"). When "add shortcut" is invoked for a property within
side panel or properties tab, add the new keyitem to "User Interface"
keymap.
Pull Request: https://projects.blender.org/blender/blender/pulls/137080
Masks for scrollbars were properly shrunk to match their drawn
representation in #135021, however, click logic was not updated
alongside this change.
For one, `mouse_in_scroller_handle` incorrectly assumed that if the
slider filled up the entire scrollbar extents, we would only be able to
select the bar. This code never ran in practice since `scrollers` were
always smaller than their mask, but now that they're the right size,
it's being run. Fix by removing the case.
It is also guaranteed that the "scroller"/"bubble"/"thumb" is not out of
view of the scrollbar track itself because it is clamped at the end of
`view2d_scrollers_calc` (which always runs before
`mouse_in_scroller_handle`), so this is dead code that can be removed.
Finally, rename `mouse_in_scroller_handle` to `scrollbar_zone_get` to
better represent its function.
Pull Request: https://projects.blender.org/blender/blender/pulls/137437
This adds initial support for showing logged data in closures. This is more
tricky than the other zone types, because the zone content is evaluated
elsewhere. The main new thing here is a function that attempts to find where a
given closure is evaluated statically.
Finding this place statically is also important because we generally decide
which compute contexts we want to log before evaluation starts. That's because
we don't want to log everything (which is expensive), but just the places that
the user is currently looking at.
This also changed a bunch of CMakeLists.txt files so that these modules can
include NOD_* stuff, which is generally fine everywhere in editors code.
Pull Request: https://projects.blender.org/blender/blender/pulls/137403
This was introduced during EEVEE-next developement
cycle to not make the buildbot fail because of EEVEE
render tests.
These have stabilized now and we can remove this option.
Pull Request: https://projects.blender.org/blender/blender/pulls/137545
This patch replaces our own FindTBB module with TBB's own TBBConfig
module. On the technical side of things, this harvests the TBB CMake
config modules, and switches our TBB CMake find_package calls from
Module mode to Config mode. Integration was done using OpenPGL as a
blueprint for parsing TBB target's property back into CMake variables
(TBB_LIBRARIES and TBB_INCLUDE_DIRS).
Pull Request: https://projects.blender.org/blender/blender/pulls/137063
Not sure why it was not done before... But it should be fine to
completely ignore the loopback relationships (mainly liboverride
hierarchy root pointers themselves, and owners of embedded and
shape key IDs) when handling hierarchies.
Quick tests on complex production cases here also showed no obvious
issue (or even change at all) with this tweak, so think it's safe to try
it out in main.
This make the shadow rays always multiply the lighting,
which can make renders with large light sources darker.
This bias introduced too many light leaking issues. And
these issues are made more visible since the introduction
of 410282e156.
Its first use was to avoid self shadowing artifacts from
large light sources. Since the shading was already
integrated over the visible hemisphere, it made sense
that the shadowing was also in the visible hemisphere.
However this only works if the shading is actually
clipped to the visible hemisphere above the geometric
normal, which might not always be the case.
Given it is better to have self shadowing noise than
light leaking, it is better to just removing this trick.
Self shadow noise can be mitigated by increasing the
sample count or by increasing the terminator bias.
Light leaking, on the other hand, has no known workaround.
Fix#136594
Pull Request: https://projects.blender.org/blender/blender/pulls/137485
The "Make Panel Toggle" operator does not check if active item is an input socket,
leading to a situation where it can be called on Boolean output sockets.
Doing this creates weird behavior where the panel and associated
operators believe it has a toggle, but is not reflected in the UI drawing.
Other weird behaviors include being able to rearrange output sockets
and apply the operator again to another output socket.
This patch adds a check in the operator's poll function to prevent it
from being called on output sockets.
Pull Request: https://projects.blender.org/blender/blender/pulls/137526
This adds a simple `compute_context_for_edittree` function that returns the
"active" compute context for the given node editor. This is used in various
places, but previously one had to construct the compute context in multiple
steps (first find the root context (modifier/operator), then handle the tree
path). Since the edittree already has a specific active context, there should be
an easy way to retrieve that.
This also adds a few extra check that avoid redundant work that was done before.
Pull Request: https://projects.blender.org/blender/blender/pulls/137525
Seems that this was a fundamental design problem that was resolved
with 3b6ee8cee7.
Since the vertex group names are stored on the mesh now, we can
just store their state in the undo mesh, and restore it when decoding
an undo step.
Pull Request: https://projects.blender.org/blender/blender/pulls/137225
Creating f-curves one by one is quadratic complexity due to:
1) scanning all existing f-curves for possible name+index duplicates,
2) channelbag invariants recalculation after each f-curve addition.
3) f-curve array storage is reallocated for each f-curve addition.
In cases when many f-curves are created at once (primary case:
importing an animated character), using the newly added
fcurve_create_many is much faster, because it:
1) uses a hashtable for duplicate name+index checking, instead of
linear scan, and
2) recalculates channelbag invariants just once.
3) does f-curve array reallocation just once.
Importing USD file exported out of Blender Studio "Rain Restaurant"
scene (about 20k f-curves), on Ryzen 5950X goes 2.0 sec -> 0.9 sec.
Pull Request: https://projects.blender.org/blender/blender/pulls/137004
Every call to BM_faces_join and BM_faces_join_pair has been adjusted to
provide the new face pointer, and a BLI_assert_msg has been added so
that doubles are now consistently identified and flagged as a problem.
BM_faces_join is now also capable of automatically reusing a double
when it is found. This new behavior is currently unused at this point.
Future patch-sets will begin to use it, allowing simplification of
calling functions.
Ref: !137406
Boolean sockets were not handled in shader nodes, so they were assigned
a NONE GPU type which causes crashes later on. No such sockets exist
yet, so this is a fix for future work.
With the move to the new snapping system,
the snapping flags were no longer stored on the editor but on the scene.
While that is good for all editors that deal with animation,
it also affected the Driver Editor since that is effectively a Graph Editor.
However since the driver editor doesn't deal with frames,
it is unlikely that settings should be shared with other animation editors.
With this PR the Driver Editor gets it's own snapping flag,
which means snapping can be controlled without affecting
other animation editors.
This also removes snapping options
* Snap to markers doesn't make sense, there are no markers in drivers
* Snap to seconds also makes no sense
Only snap to frames, i.e. whole numbers is kept with the option
to snap directly to whole numbers or offset the current position
by a whole number.
Pull Request: https://projects.blender.org/blender/blender/pulls/136826
Descriptor pools were never discarded, leading to out of memory issues
when running for a long time. This PR discards used descriptor pool when
the render graph is submitted to the device.
- Detected that to descriptor sets could be uploaded multiple times
however once was always empty.
- When render graph is flushed all descriptor pools are discarded.
- Improved debugging of discard pools.
Pull Request: https://projects.blender.org/blender/blender/pulls/137521
The pose mode Select Mirrored operator now keeps bones selected when
they cannot be mirrored.
This means that you can now select the spine and the left arm, mirror
the selection, and it'll have the spine and the right arm selected.
Before this change, all non-mirrorable bones (like the spine) would
always get deselected.
I also snuck in a little bugfix, where locked-for-selection bones could
still get selected.
Pull Request: https://projects.blender.org/blender/blender/pulls/136836
A few smaller cleanups to the pose mode Select Mirror operator.
- Don't store flags in a `void *` on the pose channel, but just use a
map to store the pre-mirroring selection flags.
- Instead of checking the `extend` flag for every bone, use two
functions to either add to the selection or set the selection, and use
the 'extend' flag to choose between those once.
- Flip some conditions and use `continue` to reduce cognitive
complexity.
No functional changes.
Pull Request: https://projects.blender.org/blender/blender/pulls/136836
Do this only when applicable.
This allow better compile time checking in Shader C++ compilation.
Moreover, this allows to have `constexpr` in shared code between
C++ and GLSL.
After investigation the `const` keyword in GLSL has the same
semantic than C/C++.
Rel #137333 and #137446
Pull Request: https://projects.blender.org/blender/blender/pulls/137497
This commit adds some extra prints to terminal related to oneAPI driver
information in the situation that the driver version is considered
incompatible with the current version of Cycles.
Pull Request: https://projects.blender.org/blender/blender/pulls/137272
