Some features are incompatible with multithreading and reliable evaluation
of dependencies. We are now removing them as part of a bigger cleanup to
fix bugs in keyframing and invalid animation evaluations.
* Dupliframes have been removed. This was a hack added before there were
more powerful features like the array modifier.
* Slow parent has been removed, never worked in 2.8. It was always
unreliable for use in production due to depending on whatever frame was
previously evaluated, which was not always the previous frame.
* Particle instanced objects used to have their transform evaluated at
the particle time. Now it always gets the current time transform.
* Boids can no longer do predictive avoidance of force field objects,
but still for other particles.
Differential Revision: https://developer.blender.org/D4274
For now this is not part of copy-on-write, and needs extra animation
evaluation.
Reviewers: sergey, brecht
Maniphest Tasks: T59939
Differential Revision: https://developer.blender.org/D4140
The problem here was that when a render result is allocated, the standard render passes are added according to the
pass bitfield. Then, when the render engine sets the result, it adds the additional passes which are then merged
into the main render result.
However, when using Save Buffers, the EXR is created before the actual render starts, so it's missing all those
additional passes.
To fix that, we need to query the render engine for a list of additional passes so they can be added before the EXR
is created. Luckily, there already is a function to do that for the node editor.
The same needs to be done when the EXR is loaded back.
Due to how that is implemented though (Render API calls into engine, engine calls back for each pass), if we have
multiple places that call this function there needs to be a way to tell which one the call came from in the pass
registration callback. Therefore, the original caller now provides a callback that is called for each pass.
Basically what we address here is to make sure the active object and the cage
are not interferring with the baking result (e.g., when baking Combined).
To do so, we take advantage of the fact that we create our own depsgraph
for baking. So now we can change the cowed objects, instead of the
original ones.
Note: There is still a way to get a crash. If you try to bake from
selected to active when is_cage, but with no cage object, we get an
assert:
```
BLI_assert failed: //source/blender/blenkernel/intern/DerivedMesh.c
mesh_calc_modifiers(), at
(((Mesh *)ob->data)->id.tag & LIB_TAG_COPIED_ON_WRITE_EVAL_RESULT) == 0
```
We can bypass this by passing ob_low instead of ob_low_eval to
bake_mesh_new_from_object on object_bake_api.c:847 . But then the edge
split modifier change will take no effect.
Since we started using looptris we no longer need a triangulation
modifier in the highpoly object. In fact having was causing a bug
where baking would be utterly broken.
This fix normal baking. Combined pass still needs a fix to hide the
objects during baking.
There were at least three copies of those:
- OB_RECALC* family of flags, which are rudiment of an old
dependency graph system.
- PSYS_RECALC* which were used by old dependency graph system
as a separate set since the graph itself did not handle
particle systems.
- DEG_TAG_* which was used to tag IDs.
Now there is a single set, which defines what can be tagged
and queried for an update. It also has some aggregate flags
to make queries simpler.
Lets once and for all solve the madness of those flags, stick
to a single set, which will not overlap with anything or require
any extra conversion.
Technically, shouldn't be measurable user difference, but some
of the agregate flags for few dependency graph components did
change.
Fixes T58632: Particle don't update rotation settings
Currently this is possible after built-in images are loaded in memory.
Allows to save memory used by dependency graph and copy-on-write.
In practice this lowers peak system memory usage from 52GB to 42GB on
a production file of spring 03_035_A.lighting.
Note, that this only applies to F12 and command line renders.
Bigger note, that this optimization is currently only possible if
there are no grease pencil objects to be rendered.
That kind of implicit includes should really only be done when totally,
absolutely necessary, and ideally only with rather simple 'second-level'
headers.
Otherwise not being explicit with includes always end up biting in
unexpected ways...
This option make the internal render size larger than the output size in
order to minimize screenspace effects disapearing at the render edges.
The overscan size added around the render is the maximum dimension
multiplied by the overscan percentage.
Loop triangles are tessellated triangles create from polygons, for renderers
or exporters that need to match Blender's polygon tesselation exactly. These
are a read-only runtime cache.
Tessfaces are a legacy data structure from before Blender supported n-gons,
and were already mostly removed from the C code.
Details on porting code to loop triangles is in the release notes.
Differential Revision: https://developer.blender.org/D3539
Terms get/set don't make much sense when casting values.
Name macros so the conversion is obvious,
use common prefix for easier completion.
- GET_INT_FROM_POINTER -> POINTER_AS_INT
- SET_INT_IN_POINTER -> POINTER_FROM_INT
- GET_UINT_FROM_POINTER -> POINTER_AS_UINT
- SET_UINT_IN_POINTER -> POINTER_FROM_UINT
With the removal of blender internal, the texture preview had to be
rewritten. Now we evaluate and write the texture directly to the
preview buffer instead of going through the renderer.
Differential Revision: https://developer.blender.org/D3670
