The issue was caused by the shader always applying exposure even if
it is 1. When the input image had negative values this lead to nan
values to the input of the shader which converts to the display
space.
Solved by only doing to-scene-linear transform in the shader which
does this.
Caused by 18110744a2
Pull Request: https://projects.blender.org/blender/blender/pulls/138666
Going from fullscreen to temporary fullscreen (usually file browser),
and back, is AREA_FLAG_STACKED_FULLSCREEN. In this case we need to
ensure that we restore the last-used area subtype and not just the zero
base type.
Pull Request: https://projects.blender.org/blender/blender/pulls/138682
In some cases, multiple viewers can be active and the active viewer is
not the desired one. This is only a problem for the compositor
The fix has two parts:
1. Tag the node tree as changed after the output changes. This solves
the issue of two viewers being active at the same time
2. Deactivate all other viewers before activating the desired viewer
node (similar to how `NODE_OT_activate_viewer()` works). This ensures
that the only active viewer is the one that the user just set.
Pull Request: https://projects.blender.org/blender/blender/pulls/138671
As described in #122398, implement read and write support for a new
attribute storage system. Currently this is only implemented to support
forward compatibility; the format used at runtime isn't changed at all.
That can be done one step at a time during the 4.5 and 5.0 development
cycles. A new experimental option for testing tells Blender to always
save with the new format.
The main benefit of the new structure is that it matches the attribute
system design, it allows for future attribute storage optimization, and
each attribute is an allocated struct, which will give pointer stability
for the Python API.
The next step is to connect the attribute API and the RNA API to
AttributeStorage for the simplest geometry type, point clouds.
Pull Request: https://projects.blender.org/blender/blender/pulls/133874
This file is used to point to Blender's funding manifest hosted on
blender.org. This establishes provenance and verifies that that the
publisher of the manifest is authorised to solicit funding on behalf
of the URLs (entity, projects) described in the manifest.
This file is needed to participate in https://floss.fund/
(and potentially other funding initiatives).
Most of the fundraising info is hosted on blender.org/, but the format
requires a reference to the actual codebase. I think this is the least
intrusive way to achieve that.
In particular, the file is referenced by https://www.blender.org/funding.json
More info on the json file itself is available https://fundingjson.org/
Pull Request: https://projects.blender.org/blender/blender/pulls/138676
This commit improves the grammar when reporting which versions of
Blender a commit was backported to in the "bug fixes per major release"
script if it was backported to 3 or more verisons of Blender.
Before:
- Backported to 3.6.22 & 4.2.9 & 4.4.1
After:
- Backported to 3.6.22, 4.2.9, and 4.4.1
Along with this change, I now use the word "and" rather than the
symbol "&".
Pull Request: https://projects.blender.org/blender/blender/pulls/138500
The script for collecting bug fixes per release have a caching feature
to speed up frequent re-running of the script by triagers to sort
commits.
This commit changes the script so that fix commits that don't have a
known module are no longer cached.
This is done because part of the sorting of commits is making sure
they're assigned to the right module. This typically means checking
on commits with "Unknown" modules and giving them a module label.
However due to caching, triagers running the script won't see those
changes until they clear their cache. To fix this, we no longer cache
commits with unknown modules.
Pull Request: https://projects.blender.org/blender/blender/pulls/138503
This patch removes the translation Offset from the Scale node. The
reasoning is that it is easy to insert a Translate node afterwards to
perform any necessary translation. And since we are moving options to
inputs, it doesn't seem worth it to provide those offsets as inputs in
the process.
Pull Request: https://projects.blender.org/blender/blender/pulls/138668
There are two parts for this PR. One is to change some of our build pipeline to make certain libs reproducible. For this part I want to clarify two things:
1. Why change python to use `--disable-optimizations`?
This is because `--enable-optimizations` turns on PGO (Profile Guided Optimization). PGO is sadly not deterministic and will create different binaries on every recompile. So to create reproducible build this needs to be turned off. This also seems to only have been turned on for Linux specifically(?) on our side. So on Windows and Mac our python build already doesn't have PGO.
2. Why split out cython and zstandard from site-packages?
Sadly pip does not seem to respect `SOURCE_DATE_EPOCH`. It also creates temporary folders with random hashes in them that is then recorded into the Cython libraries (I'll touch on this again later). I've looked at the discussions about this upstream and sadly the pip maintainers do not really want people to use pip as a reproducible build system pipeline and instead directs users to other solutions if they want reproducible builds.
The other part is about setting up our pipeline to not introduce any random hashes or build timestamps into our libraries. Here I do two things:
1. We need to set the `SOURCE_DATE_EPOCH` environmental variable to a specific date that will not change.
This is needed as the compile time date is recorded in certain libraries and files. (So hard coding it with this env var will make the end result reproducible)
2. We need to strip the created static and shared libraries. This is because the static libraries are not created in a deterministic way. For shared libraries some of our libraries includes debug symbols which contains paths to temporary files with random hashes. To solve this without stripping in post, we would need to either patch the linker on Rocky8 or patch a lot of our libraries. I think it is better to just do this as a post build step. (This seems to be what most linux distributions do as well).
With all this, we can make our Linux library builds is almost 100% reproducible. (At least on my machine where I tested)
By almost, I mean that there is sadly a catch in that certain libraries like Cython saves the source code path in their libraries for error messages. However now the builds are reproducible if the folder path is the same.
IE if the libraries are always built in `/home/builder/build_linux/deps_x64`, then they should now be reproducible.
Pull Request: https://projects.blender.org/blender/blender/pulls/134221
Briefly about this change:
- OpenColorIO C-API is removed.
- The information about color spaces in ImBuf module is removed.
It was stored in global ListBase in colormanagement.cc.
- Both OpenColorIO and fallback implementation supports GPU drawing.
- Fallback implementation supports white point, RGB curves, etc.
- Removed check for support of GPU drawing in IMB.
Historically it was implemented in a separate library with C-API, this
is because way back C++ code needed to stay in intern. This causes all
sort of overheads, and even calls that are strictly considered bad
level.
This change moves OpenColorIO integration into a module within imbuf,
next to movie, and next to IMB_colormanagement which is the main user
of it. This allows to avoid copy of color spaces, displays, views etc
in the ImBuf: they were used to help quickly querying information to
be shown on the interface. With this change it can be stored in the
same data structures as what is used by the OpenColorIO integration.
While it might not be fully avoiding duplication it is now less, and
there is no need in the user code to maintain the copies.
In a lot of cases this change also avoids allocations done per access
to the OpenColorIO. For example, it is not needed anymore to allocate
image descriptor in a heap.
The bigger user-visible change is that the fallback implementation now
supports GLSL drawing, with the whole list of supported features, such
as curve mapping and white point. This should help simplifying code
which relies on color space conversion on GPU: there is no need to
figure out fallback solution in such cases. The only case when drawing
will not work is when there is some actual bug, or driver issue, and
shader has failed to compile.
The change avoids having an opaque type for color space, and instead
uses forward declaration. It is a bit verbose on declaration, but helps
avoiding unsafe type-casts. There are ways to solve this in the future,
like having a header for forward declaration, or to flatten the name
space a bit.
There should be no user-level changes under normal operation.
When building without OpenColorIO or the configuration has a typo or
is missing a fuller set of color management tools is applies (such as the
white point correction).
Pull Request: https://projects.blender.org/blender/blender/pulls/138433
With two or more monitors of different scales, Blender would fail to
start in some cases when it opened on a monitor at a size that wasn't
a multiple of the monitors scale.
This seemed only to cause problems with LIBDECOR however the buffer size
should always be a multiple of the buffer scale, so ensure its always
valid.
See #135764 for details (a closely related bug).
This patch removes the Max Blur option from the Bokeh Blur node. The
reasoning is as follows.
- The option was unused in case of non-variable sized blur, which was
the default, so the option did nothing for the default case of the
node.
- The option was originally introduced to define the search window of
variable size blur. This is no longer the case, since we compute the
search window dynamically from the input using parallel reduction,
which is now very fast. So currently, it only works as an upper limit.
- The node options will be exposed as inputs, so the user will see two
inputs that control the radius, which can be confusing for users that
are not experienced.
- The plan is to make the node take absolute pixel sizes in the future,
instead of the arbitrary relative size now in place, which would make
it very easy for the user to impose such limits manually.
It is difficult to version this change, since the size is relative to
the image size, while max blur is in pixels. But we assume the user
chose a sufficiently large max blur for the node to be useful, so the
composite should not be expected to be drastically different.
Pull Request: https://projects.blender.org/blender/blender/pulls/138659
The use of `sprintf()` was causing compiler warnings on Apple's version
of Clang. This replaces those uses with `fmt::format_to_n()` from
libfmt, which is safer and silences those warnings.
Additionally, the format string given to `printf()` in debugging
functions for the path template unit tests was also causing warnings due
to type mismatch (long long vs long). This replaces those with
`fmt::print()`, which infers the correct type on its own, silencing
those warnings.
Pull Request: https://projects.blender.org/blender/blender/pulls/138660
Starting with C++20, `fmt::format` will process the format string at
compile time by default. We need to opt out in the cases where this is
not possible by using `fmt::runtime(...)`, like, for example, when using
our various translation utilities.
This mirrors prior commit e62aa986b2 and
fixes 2 cases that have slipped back in.
Pull Request: https://projects.blender.org/blender/blender/pulls/138640
Previously, the outline of frames was sometimes unexpectedly invisible.
The solution is to draw the frame outlines together with the zone outlines
after all the zone and frame backgrounds have been drawn already.
Pull Request: https://projects.blender.org/blender/blender/pulls/138645
By convention str_len excludes the nil terminator.
Also assert the increment is a single unit as larger numbers aren't
supported (or needed currently).
The idea here is to use `UI_but_func_tooltip_label_set` which has a shorter
duration until the tooltip shows (besides also being easier to use). It does
seem like it makes sense to use it here, because the only purpose of the error
icon is to hover over it. Therefore, one shouldn't have to wait for long until
the tooltip shows.
Pull Request: https://projects.blender.org/blender/blender/pulls/138583
Previously, we had shown link errors on the target node as part of the node
warning system. While better than not showing any information about invalid
links (as was the state before that), it's still not ideal because it's easy to
miss when just looking at the link.
This patch adds an error icon in the middle of the invalid link. When hovering
over it, it shows the error text. When the middle of the link is not in view but
part of the link is, then the error icon will also stay visible.
Pull Request: https://projects.blender.org/blender/blender/pulls/138529
Moving the cyclic mode check into a separate function simplifies the
code. Avoiding an additional local variable to track the result, and
better encapsulates the behavior.
Pull Request: https://projects.blender.org/blender/blender/pulls/138375
The old versioning code was not guarded in a `MAIN_VERSION_FILE_ATLEAST`
check and the RNA hard minimum was not updated to prevent setting this
to an invalid value.
Pull Request: https://projects.blender.org/blender/blender/pulls/138572
Recent increases in region edge hit sizes, to work better with tablet
pens, made the hitspace for Properties NavBar encroach into the tab
space. This PR just narrows it and moves it away from the tabs.
Pull Request: https://projects.blender.org/blender/blender/pulls/138622
This converts the public `uiItemR` function to an object oriented
API (`uiLayout::prop`), matching the python API.
This reduces the difference between the C++ API with the python version,
its also helps while converting code from python to C++ code (or vice-versa),
making it almost seamless.
Part of: #117604
Pull Request: https://projects.blender.org/blender/blender/pulls/138617
This is loosely related to #136285, which allows changing numerical
values while in text entry mode. Outside of that, just hovering over
numerical inputs we are currently able to increment and decrement by
Ctrl + Mouse Wheel. This works for integers, floats, sliders. Current
code also allows cycling through expanded enums. This PR extends that
so Ctrl + Mouse Wheel also changes toggles and checkboxes and also
changes values inside of strings.
Pull Request: https://projects.blender.org/blender/blender/pulls/138344
Add a new shader node to control volume coefficients (scattering,
absorption and emission) directly, making it easier to model existing
volumes with measured data.
Pull Request: https://projects.blender.org/blender/blender/pulls/136287
... if only a single view is enabled
Logic for touching placeholder files in `RE_RenderAnim` uses
`BKE_scene_multiview_filepath_get` (no matter if only a single view is
enabled), but logic in `BKE_image_render_write` falls back to using the
"regular" filepath (once gotten from `BKE_image_path_from_imformat`) if
there are less than two views.
That would also break the behavior of "Skip Overwrite".
Digging in git history is a bit hard, so not exactly sure why this was
added, but this PR changes behavior so that
`BKE_scene_multiview_filepath_get` is always used (even for a single
(non-empty) view. Done by implementing `RE_ResultIsMultiView`
Pull Request: https://projects.blender.org/blender/blender/pulls/138507
It would not use topological distances.
This was the case for Grease Pencil, Curves were right though (even
though both use the same code `curve_populate_trans_data_structs` /
`calculate_curve_point_distances_for_proportional_editing`)
So that calculation is actually right, the thing that made it fail for
Grease Pencil was that `init_proportional_edit` was calling
`set_prop_dist` with the `with_dist` argument as `true` [this would
overwrite the already calculated dist with the "plain", "non-
topological" distance again... leading to wrong values being used in
`calculatePropRatio`]
So to resolve, skip the `set_prop_dist` alltogether if T_PROP_CONNECTED
is used. Done for both Curves and Grease Pencil and move into own block.
Legacy Curve get their own codeblock (with a comment where their
topological distances are calculated).
Pull Request: https://projects.blender.org/blender/blender/pulls/138588
Part of #136993.
Share as much of the ShaderCompiler implementations as possible.
Remove the ShaderCompiler/ShaderCompilerGeneric split and make most of
its functions non virtual.
Move the `get_compiler` function from `Context` to `GPUBackend` and
creation/deletion to `GPUBackend::init/delete_resources`.
Add a `batch_cancel` function to `ShaderCompiler` (needed for the
GPUPass refactor).
As a nice extra, the multithreaded OpenGL compilation has become faster
too.
The barbershop materials + EEVEE static shaders have gone from 27s to
22s.
I have not observed any performance difference on Vulkan or Metal.
Pull Request: https://projects.blender.org/blender/blender/pulls/136676
Expand the already-existing safety margin to correct for the implosion
scenario, by a factor of 100x. That way it doesn't zoom in too much,
and makes it easy for an animator to zoom out again if necessary.
Pull Request: https://projects.blender.org/blender/blender/pulls/138235
