Make sure that proxy and original images are scaled to same size before
applying offset or crop.
During testing, I discovered, that raw cache will lose information whether
this image was proxy or not. Because of this, proxy images will not create
this cache type. It would be fairly easy to implement this functionality for
cache, but I have decided to not do it now, because I did not want to pass yet
another mostly hard-coded bool flag to cache system. Since image is proxy, it
should be fast to read anyway.
In case of using offset property, code was modified to make sure we scale
image only once. I also tried to make code more readable and streamlined and
cleaned up surrounding functions a bit.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8203
This adds `blender::RandomNumberGenerator` in `BLI_rand.hh`.
Furthermore, `RNG` is now implemented in terms of this new generator.
No functional changes are expected, the generated random numbers
are not changed by this commit.
Reviewers: campbellbarton, brecht
Differential Revision: https://developer.blender.org/D8259
Time will tell whether we need to expose more RNA override flags here.
Implements/Fix T78534.
Differential Revision: https://developer.blender.org/D8250
The triplet static RNA / runtime RNA / custom properties is a real pain to
deal with...
Using the new `PropertyRNAOrID` struct helps clarifying and properly
dealing with all three cases.
Note that this makes override of py-defined RNA properties working
(support for that will be committed next).
Differential Revision: https://developer.blender.org/D8249
Introduce new PropertyRNAOrID structure, storing most useful data about
an 'opaque' PropertyRNA in relation with a given PointerRNA struct.
It deals with all the three cases (pure static RNA, runtime RNA where
data is actually stored in IDProperties, and pure IDProperties, aka
custom data.
Use binary search for querying deform weights.
Spring 02_020_A.anim.blend on Ryzen 1700X goes from 12.4 to 12.7fps.
During profiling it was detected that adding new items to the head was faster than adding to the tail.
Reviewed By: Campbell Barton
Differential Revision: https://developer.blender.org/D8127
This changes the behavior of rip when entire faces are selected.
Now face regions are isolated and moved instead of ripping the edge-loop
extracted from the selection boundary.
This is a convenient alternative to separate selection & move.
Resolves T78751.
Loop over faces and calculate their centers instead of zeroing the
face center array and accumulating all faces vertex corners.
Move subsurf face center extraction into it's own loop since it works
differently.
This disables all Sculpt Vertex Colors tools, operators, panels and rendering capabilities and puts them under the "Use Sculpt Vertex Colors" experimental option.
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D8239
As Face Sets IDs start from 0 and increase by 1 each time a new face set
is created in a mesh, when joining multiple meshes it could happen that
the same ID is used by several unrelated areas in multiple objects. This
checks the Face Sets IDs when joining meshes and ensures that they are
not repeated between different objects when joining them, so in the
resulting mesh all previous face sets will have different IDs.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8224
Note that this behavior is enforced on user level for now, but on code
side it is controlled with a flag, which should make it easy to refine
that behavior if needed.
Only exception is when we duplicate a linked ID directly (then we assume
user wants a local deep-copy of that linked data, and we always also
duplicate linked sub-data-blocks).
Note that this commit also slightly refactor the handling of actions of
animdata, by simplifying `BKE_animdata_copy_id_action()` and adding an
explicit new `BKE_animdata_duplicate_id_action()` to be used during ID
duplication (deep copy).
This also allows us to get rid of the special case for liboverrides.
As tools iterators skip not visible vertices, fully hidden nodes can
also be skipped and considered as masked.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8244
This function was using the wrong flag to update the visibility state of
the nodes, so I assume that most of the partially visible optimizations
were not working.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8243
