This is a more correct fix to the issue Brecht was fixing in D6600.
While the fix in that patch worked fine for linking it broke ASAN
runtime under some circumstances.
For example, `make full debug developer` would compile, but trying
to start blender will cause assert failure in ASAN (related on check
that ASAN is not running already).
Top-level idea: leave it to CMake to keep track of dependency graph.
The root of the issue comes to the fact that target like "blender" is
configured to use a lot of static libraries coming from Blender sources
and to use external static libraries. There is nothing which ensures
order between blender's and external libraries. Only order of blender
libraries is guaranteed.
It was possible that due to a cycle or other circumstances some of
blender libraries would have been passed to linker after libraries
it uses, causing linker errors.
For example, this order will likely fail:
libbf_blenfont.a libfreetype6.a libbf_blenfont.a
This change makes it so blender libraries are explicitly provided
their dependencies to an external libraries, which allows CMake to
ensure they are always linked against them.
General rule here: if bf_foo depends on an external library it is
to be provided to LIBS for bf_foo.
For example, if bf_blenkernel depends on opensubdiv then LIBS in
blenkernel's CMakeLists.txt is to include OPENSUBDIB_LIBRARIES.
The change is made based on searching for used include folders
such as OPENSUBDIV_INCLUDE_DIRS and adding corresponding libraries
to LIBS ion that CMakeLists.txt. Transitive dependencies are not
simplified by this approach, but I am not aware of any downside of
this: CMake should be smart enough to simplify them on its side.
And even if not, this shouldn't affect linking time.
Benefit of not relying on transitive dependencies is that build
system is more robust towards future changes. For example, if
bf_intern_opensubiv is no longer depends on OPENSUBDIV_LIBRARIES
and all such code is moved to bf_blenkernel this will not break
linking.
The not-so-trivial part is change to blender_add_lib (and its
version in Cycles). The complexity is caused by libraries being
provided as a single list argument which doesn't allow to use
different release and debug libraries on Windows. The idea is:
- Have every library prefixed as "optimized" or "debug" if
separation is needed (non-prefixed libraries will be considered
"generic").
- Loop through libraries passed to function and do simple parsing
which will look for "optimized" and "debug" words and specify
following library to corresponding category.
This isn't something particularly great. Alternative would be to
use target_link_libraries() directly, which sounds like more code
but which is more explicit and allows to have more flexibility
and control comparing to wrapper approach.
Tested the following configurations on Linux, macOS and Windows:
- make full debug developer
- make full release developer
- make lite debug developer
- make lite release developer
NOTE: Linux libraries needs to be compiled with D6641 applied,
otherwise, depending on configuration, it's possible to run into
duplicated zlib symbols error.
Differential Revision: https://developer.blender.org/D6642
We compile zlib as own dependency, but are not informing BLOSC
to use it. This leads to zlib symbols defined twice when linking
Blender: one set comes from libz.a and another one from libblosc.a.
Tested on Linux Debian testing and CentOS 7.5.
It is possible that this change on its own will lead to linking
errors after libraries are re-compiled, This will be fixed as
a dedicated fix to Blender's build system.
Reviewed By: brecht, mont29, LazyDodo
Differential Revision: https://developer.blender.org/D6641
The problem was related to the smooth of the strength. As the factor was very low and the value was smoothed , the result was almost nothing when the radius was very small. Now the factor is higher and the smooth is done after clamping pressure.
This adds some kind of dashing to the tube and cone limits. Although the
dashing is in object space and is not a good as old dashing. But it is
the least time consuming and least complex solution.
This commit will prevent the crash. It does produce a result that may not
be desirable (only briding one face of the cube), but at least it's better
than crashing.
The contents of the Rotation panel was already disabled when a bake
exists, but the checkbox in the header wasn't. Since rotations are part
of the baked data, it doesn't make sense to enable/disable rotations
after baking.
This commit belongs to T72894. It's related to (my) previous commits on pointer exchanges (today + yesterday). It cleans up the functions by describing their usage in the comments, adds additional nullptr checks, and fixes the reference count responsibilities of newly created PyObjects.
The OptiX SRT motion expects a motion defined by translation,
rotation, shear and scale, but the matrix decomposition code in
Cycles was not able to extract shear information and instead
produced a stretch matrix with the information baked in. This
caused conflicting transforms between traversal and shading
and lead to render artifacts.
This patch changes the matrix decomposition to produce factors
inline with what OptiX expects to fix that.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6605
Waveform is freed in `sound_load_audio()` when sound length is `queried by BKE_sound_info_get()`.
Add argument free_waveform, so `BKE_sound_info_get()` can skip waveform freeing, as it is not expected to alter sound.
Reviewed By: sybren
Differential Revision: http://developer.blender.org/D6053
Refactor code to use `eSpaceSeq_Proxy_RenderSize` or corresponding `IMB_Proxy_Size`
enum items directly.
`SEQ_PROXY_RENDER_SIZE_100` has assigned value 99 to distinguish from `SEQ_PROXY_RENDER_SIZE_FULL`.
This caused error in image size calculation and because of that image had to be scaled.
Author: EitanSomething
Reviewed By: ISS
Differential Revision: http://developer.blender.org/D6368
Sometimes on flat open vertices the thickness would suddenly start
to be jumping in powers of 2, also when bending a plane there is a
noticeable jump in the geometry.
When offset is set to -1 or 1 weird glitches happen.
See D6559 for test cases that failed.
Some modifiers (like particle system, but also some other physics
simulations) require extra steps besides creating the modifier itself,
so had to switch to higher-level `ED_object_modifier_add()` to add
inserted modifiers in an overriding object data-block.
Due to how particle modifier handles copying, some extra nasty
hacks/work-around were also needed to get it working.
This data is only used to get current time/frame value, which is never
mandatory to add a modifier.
Needed by incoming fix to support particles modifiers in liboverrides.
The `give_parvert()` function was only considering the mesh's original
vertex indices when the parent vertex index was valid for the evaluated
mesh. However, when using the Mask modifier the evaluated mesh can have
less vertices but still have the parent vertex.
Since the `if (nr < numVertex)` condition wasn't used to prevent any
out-of-bounds access, and seems just an incorrect optimisation, it could
be removed.
Even though we build USD as static, it still feels the need to mark its
symbols with declspec(dllexport) which means the blender binary now exports
these symbols.
this patch fixes that unwanted behaviour, however USD libs still need to
rebuild before this becomes visible in the blender binary
Differential Revision: https://developer.blender.org/D6563
Reviewed By: sybren
The USD landing broke building with clang on windows
due to a couple of reasons:
1) Some incompatibilities in their headers [1] only one
of them was important for us and is included in our patchset
now.
2) clangs lld wanted the full path to the libusd_b library
when using the whole archive link option, while msvc can
figure it out from just the library name.
Tested with clang/msvc and msbuild and ninja generators
[1] https://github.com/PixarAnimationStudios/USD/issues/1030
When using the color picker on a image editor it tries to read the color
from the original image. When there is no original image the code
crashes during the determination of the UDIM tile number.
There are 2 approaches to solve this.
1. Modify `BKE_image_get_tile_from_pos` to support NULL pointers.
2. Modify `ED_space_image_color_sample` with an early exit.
This patch modifies `ED_space_image_color_sample` with an early exit.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D6629
There were two strange things going on:
- Start frame was clamped to 1, even when frame 0 is always a valid number.
This also ignored the 'Allow Negative Frames' user preference.
- Start frame was only clamped when setting the end frame, so first setting
the end frame and then the start frame would result in a different result
than doing it in the opposite order.
This commit fixes both issues by:
- Clamping the lower bound of the start frame only if negative frames are
not allowed, and
- apply that clamp both when setting the start and the end frame.
FreeStyle line styles were not part of the dependency graph, and
blacklisted from the Copy-on-Write system. As a result, animated
FreeStyle properties would not be updated by the animation system,
resulting in T72213. There was an explicit call to run the animation
system on the original datablocks, but that was (for good reasons)
removed in D5394.
This commit adds the FreeStyleLineStyle datablocks to the dependency
graph and allows them to be handled by the CoW system. As a result
- the UI now updates properly when properties are animated, and
- animated property values are actually used when rendering.
This commit includes @Sergey's patch P1222, which unifies two bits of
code that did the same thing: check whether datablock type is covered by
copy-on-write.
Reviewed By: sergey, brecht
Differential Revision: https://developer.blender.org/D6609
Related to T62882
When the SupportsMapping modifier flag was added to the dynamic paint
modifier it was added commented out. It may have not worked back then,
but it seems to work fine now.
MOD_weightvgedit.c was already updated similarly.
Reviewed By: jbakker
Differential Revision: https://developer.blender.org/D6072
Similar to the Microfacet Closures, the Principled BSDF Sheen closure is
added at a high weight but typically results in fairly low values.
Therefore, the default weight is a bad indicator of importance.
The fix here is the same as it was back then for Microfacets:
Compute an average weight using the normal as the half-vector
and use it to scale down the sample weight and the albedo channel.
In addition to drastically improving denoising of materials with
sheen when using the new Denoising node, this also can reduce noise
on such materials considerably.
This would happen if a tile is found on disk, painting would actually
request that tile (because corresponding uvs were in that range), but
that tile was not added in blenders list of tiles in that Image.
Need to also check tile in `image_quick_test` (regardless of iuser
having passed).
thx @lukasstockner97 for additional input!
Maniphest Tasks: T73110
Differential Revision: https://developer.blender.org/D6578
selection
False positive when a sequences end would be the same as active
sequences start.
Also thx @sybren for the heads up to make this more readable.
Maniphest Tasks: T72546
Differential Revision: https://developer.blender.org/D6451
I doubt we can do much better. Most of the aliasing comes from the edge
detection which does not use a lot of samples. We could use more samples
but then the detection becomes way more complex and expensive.
The second issue comes from the reconstruction (AA pass) that only bleed
adjacent pixels in if their line direction is perpendicular to the offset.
This makes corner gaps on certain diagonals.
Thanks @mano-wii for providing the patch!
The original code assumed that when `animation_player_preset` was not
`CUSTOM`, the `animation_player` path would be empty. This is a bad
assumption, as it can be quite useful to be able to toggle between a custom
and a built-in player. As such, `animation_player` should not be reset to
an empty string when a non-custom player is chosen.
'Straight Cut' is actually documented as creating N-Gons. However, the
code was disallowing this. This is probably a mix-up as the "allow N-Gons"
option is documented as a "force quad/tri" option.
There is already code for this, it just wasn't working. Now add the
slash after checking for an empty string.
Differential Revision: https://developer.blender.org/D6568
The gflags library was not fully configured to be built as a static
library, only one of two files was changed.
Differential Revision: https://developer.blender.org/D6611
