| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Was caused by a missing texture resource.
|
|
As Usual...
Note that we have to edit each C-defined of those callbacks that set
Node->id pointer, instead of just moving the generic call to
`id_us_plus()` in `node_init()` below the call to
`ntype->initfunc_api()`, since that one manipulates RNA pointers, which
are supposed to already handle usercounts themselves on assignment.
The C callbacks are moving back to manipulating directly DNA, so it's
their responsability to handle user count then (Py code should do that
automatically through RNA assignement).
|
|
Previously displacement relinking was trying to be smart and seems to be
broken in some cases. This fixes all cases by brute force.
We copy the whole branch linked to the displacement socket and tag it
accordingly. Then we only relink the nodes that are not tagged.
Moreover, we bypass bump nodes inside the dispacement trees so that the
resulting bump is the same as cycles.
Fix T66000 EEVEE: Unexpected results when displacement interact with nodegroups
|
|
|
|
Now texture storage of images is defined by the alpha mode of the image. The
downside of this is that there can be artifacts near alpha edges where pixels
with zero alpha bleed in. It also adds more code complexity since image textures
are no longer all stored the same way.
This changes allows us to keep using sRGB texture formats, which have edge
darkening when stored with premultiplied alpha. Game engines seems to generally
do the same thing, and we want to be compatible with them.
|
|
The root of the problem goes to the fact that node tree copying
uses source tree and nodes for a temporary storage.
This makes it so multiple dependency graphs can not be reliably
evaluated from different threads if they are using same original
node tree.
Solved by doing the following:
- Commonly used tree copying function (which is used by library
manager) keeps source tree, nodes and sockets untouched.
- All the related areas (like node tree's callback) now have
const qualifier on the input.
- Areas which needs to have those temporary pointers assigned are
now using explicit function.
Would be really cool to get rid of those temporary pointers
completely, but this is a bit tricky due to hairy nature of the
code. Can happen any time now though: is easy enough to generalize
the new pointers mapping.
Note that this change is only intended to solve the crash.
The fact that icons shouldn't be updated on playback will be fixed
as a separate change.
Reviewers: brecht, fclem
Reviewed By: brecht, fclem
Subscribers: brecht, fclem
Differential Revision: https://developer.blender.org/D5002
|
|
Automated using clang-tidy.
|
|
|
|
|
|
This also replaces the Use Alpha setting. We now have these alpha modes:
* Straight: store RGB and alpha channels separately with alpha acting as a
mask, also known as unassociated alpha.
* Premultiplied: transparent RGB pixels are multiplied by the alpha channel.
The natural format for renders.
* Channel Packed: different images are packed in the RGB and alpha channels,
and they should not influence each other. Channel packing is commonly used
by game engines to save memory.
* None: ignore alpha channel from the file and make image fully opaque.
Cycles OSL does not correctly support Channel Packed and None yet, we are
missing fine control over the OpenImageIO texture cache to do that.
Fixes T53672
|
|
The previous behavior here was wrong for some specific combinations of
settings, non-color RGB channels should never be affected by the alpha
channel.
|
|
|
|
Hair normals were not behaving correctly. This corrects their looks and
fix the node shader geometry that was showing the flat normal.
|
|
|
|
This removes the need to pass the Material* all over the place in the draw
manager. Cleanup comming right after.
|
|
This makes it easier to set up materials with emission and transparency.
Importers/exporters and add-ons are recommended to now use these rather than
creating separate transparent BSDF and emission nodes.
|
|
Cycles now uses the color space on the image datablock, and uses OpenColorIO
to convert to scene linear as needed. Byte images do not take extra memory,
they are compressed in scene linear + sRGB transfer function which in common
cases is a no-op.
Eevee and workbench were changed to work similar. Float images are stored as
scene linear. Byte images are compressed as scene linear + sRGB and stored in
a GL_SRGB8_ALPHA8 texture. From the GLSL shader side this means they are read
as scene linear, simplifying the code and taking advantage of hardware support.
Further, OpenGL image textures are now all stored with premultiplied alpha.
Eevee texture sampling looks a little different now because interpolation
happens premultiplied and in scene linear space.
Overlays and grease pencil work in sRGB space so those now have an extra
conversion to sRGB after reading from image textures. This is not particularly
elegant but as long as engines use different conventions, one or the other
needs to do conversion.
This change breaks compatibility for cases where multiple image texture nodes
were using the same image with different color space node settings. However it
gives more predictable behavior for baking and texture painting if save, load
and image editing operations have a single color space to handle.
Differential Revision: https://developer.blender.org/D4807
|
|
|
|
Sorry for that :(
|
|
Making them world space by default remove a lot of legacy conversion from
viewspace.
|
|
The end goal for this is to lower the number of needed matrices.
This also cleanup some uneeded transformation.
|
|
Prepare for enabling ReflowComments.
|
|
|
|
|
|
On file read we need to update group nodes in case the group they refer to
has changed its inputs and outputs. This had O(n^2) time complexity and was
updating all datablocks even if they did not change.
|
|
|
|
|
|
Comments after code can cause awkward line breaks.
|
|
Was commented out for literally 10 years.
|
|
Similar to previous commit.
Compositing with animation on socket values and image sequences
still works fine.
|
|
The function was calling update() on the time curve, but there
is no update callback on that node. So, effectively the function
was doing nothing.
|
|
|
|
|
|
|
|
|
|
|
|
Apply clang format as proposed in T53211.
For details on usage and instructions for migrating branches
without conflicts, see:
https://wiki.blender.org/wiki/Tools/ClangFormat
|
|
|
|
|
|
No functional change, this adds LIB definition and args to cmake files.
Without this it's difficult to migrate away from 'BLENDER_SORTED_LIBS'
since there are many platforms/configurations that could break when
changing linking order.
Manually add and enable WITHOUT_SORTED_LIBS to try building
without sorted libs (currently fails since all variables are empty).
This check will eventually be removed.
See T46725.
|
|
Causes extra-semicolon warning,
we'll need to disable clang-format for this file since it
won't format properly.
Reverts b389bb5ef8a
|
|
|
|
We were already getting the designated output node in
'ntreeGPUMaterialNodes()' but this wasnt used in 'ntreeExecGPUNodes()',
instead whatever node was tagged NODE_DO_OUTPUT was executed.
note: this is just the bare minimum to fix the bug, other improvements
previously done in D4482 might follow as a separate commit.
Reviewers: brecht, fclem
Maniphest Tasks: T62434
Differential Revision: https://developer.blender.org/D4630
|
|
|
|
|
|
First try to implement T57489. But unfortunately, there is a missing
dependency in the depsgraph that does not trigger the shader update.
|
|
The wrong transformation was used. Add a new matrix specially for this case.
This also fix the Node texture coordinate that was suffering the same issue.
|
|
Pass binormal sign via object info.
|
|
This was caused by the material not tagged to use Diffuse lighting data.
|
