This PR adds basic support for texture update, read back and clearing
for Vulkan. In Vulkan we need to convert each data type ourselves as
vulkan buffers are untyped. Therefore this change mostly is about data
conversions.
Considerations:
- Use a compute shader to do the conversions:
- Leads to performance regression as compute pipeline can stall
graphics pipeline
- Lead to additional memory usage as two staging buffers are needed
one to hold the CPU data, and one to hold the converted data.
- Do inline conversion when sending the data to Vulkan using `eGPUDataFormat`
- Additional CPU cycles required and not easy to optimize as it the
implementation requires many branches.
- Do inline conversion when sending the data to Vulkan (optimized for CPU)
For this solution it was chosen to implement the 3rd option as it is fast
and doesn't require additional memory what the other options do.
**Use Imath/half.h**
This patch uses `Imath/half.h` (dependency of OpenEXR) similar to
alembic. But this makes vulkan dependent of the availability of
OpenEXR. For now this isn't checked, but when we are closer to
a working Vulkan backend we have to make a decision how to cope with
this dependency.
**Missing Features**
*Framebuffer textures*
This doesn't include all possible data transformations. Some of those
transformation can only be tested after the VKFramebuffer has been
implemented. Some texture types are only available when created for a
framebuffer. These include the depth and stencil variations.
*Component format*
Is more relevant when implementing VKVertexBuffer.
*SRGB textures*
SRGB encoded textures aren't natively supported on all platforms, in
all usages and might require workarounds. This should be done in a
separate PR in a later stage when we are required to use SRGB textures.
**Test cases**
The added test cases gives an overview of the missing bits and pieces of
the patch. When the implementation/direction is accepted more test cases
can be enabled/implemented.
Some of these test cases will skip depending on the actual support of
platform the tests are running on. For example OpenGL/NVidia will skip
the next test as it doesn't support the texture format on OpenGL, although
it does support it on Vulkan.
```
[ RUN ] GPUOpenGLTest.texture_roundtrip__GPU_DATA_2_10_10_10_REV__GPU_RGB10_A2UI
[ SKIPPED ] GPUOpenGLTest.texture_roundtrip__GPU_DATA_2_10_10_10_REV__GPU_RGB10_A2UI [ RUN ] GPUVulkanTest.texture_roundtrip__GPU_DATA_2_10_10_10_REV__GPU_RGB10_A2UI
[ OK ] GPUVulkanTest.texture_roundtrip__GPU_DATA_2_10_10_10_REV__GPU_RGB10_A2UI
```
Pull Request: https://projects.blender.org/blender/blender/pulls/105762
In Vulkan multiple commands can be in flight simultaneously.
These commands can share resources like descriptor sets or push
constants. When between commands these resources are updated
a new version of the resources should be created.
When a resource is updated it should check the submission id of the
command buffer. If this is different than last known by the resources,
the previous resources should be freed.
If the submission id is the same than previously it has to create a
new version of the resource to not intervene with other commands that
uses the resource before the update.
When the resource wasn't updated between multiple usages in the same
submission id it could reuse the previous resource.
This PR introduces a `ResourceTracker` and a `SubmissionTracker`.
A submission tracker can check if the command buffer is submitted.
In this case all resources of the resource tracker should be freed.
Unmodified resources in the same submission can be shared.
A resource tracker will keep track of all resources that are in
flight. After the resources are used (submission + execution) have
finished the resources can be cleared.
Pull Request: https://projects.blender.org/blender/blender/pulls/105183
Viewport assumes that when the number of planes of the image
buffer is less or equal to 8 it is a gray scale image. In that
case it will optimize the texture to be stored as a grayscale
image on the GPU.
When using a targa file with a palette, the bitplanes were not
extracted from the actual colors, but from the number of colors
that were present in the palette.
Image buffers don't support palettes so that doesn't make sense.
This PR uses the bitdepth of the actual colors inside the palette
to identify the number of planes to use in the image buffer.
Pull Request: https://projects.blender.org/blender/blender/pulls/106047
Workaround gnome-shell including uninitialized memory when pasting
from the clipboard. Where `read` woud
Reading from the pipe into a power-of-two buffer
works around the problem.
It's not clear why this only impacts gnome-shell - as there is no
significant down-side to changing the buffer size, apply a workaround.
Only the text editor supported the primary clipboard & only for modal
selection. Now selecting text in the console & 3D text editing also
sets the primary clipboard under X11 & Wayland.
Notes:
- Pasting from the primary clipboard isn't yet exposed in the key-map
so in practice it's only useful for pasting text outside of Blender.
- Use skip-save option when pasting from the primary selection
so this is never used by the regular paste shortcut.
- This commit adds a primary-clipboard flag to WM_capabilities_flag() so
creating the the copy-buffer is only performed when necessary.
Node names should be properly separated for easy searching and reading
and for consistency with other (color) nodes.
This commit should only touch UI names which do not need versioning and does not break node setups or py scripts.
AABB aligned quads would defeat the "witness" accelerator when
using the xatlas packing strategy.
This change reorders the search to allow use of the "witness" technique.
This PR uses renderdoc for frame capturing when enabled.
It enabled an easier workflow for frame capturing.
- Capture GPU API calls from test cases
- Capture GPU API calls from background threads
- Capture GPU API calls from background rendering.
Renderdoc is an important GPU debugger used by the Eevee/
Viewport module. Previously we needed to change code in
order to record background rendering, that could on its own
lead to other side-effects.
The integration with renderdoc can be enabled using
`WITH_RENDERDOC=On` compiler option. `GPU_debug_capture_begin`
and `GPU_debug_capture_end` can be added to the section
of the code you want to debug. When running Blender inside
renderdoc this part will automatically be captured.
All GPU test cases are now guarded by these calls. In order
to capture the test cases you need to start the test cases
from renderdoc and the captured GPU API calls will appear
where each capture is a single test case.
Pull Request: https://projects.blender.org/blender/blender/pulls/105921
Happens if a scene has a PointerProperty of type collection which
is set to a collection containing rigid bodies.
The error is printed by the builder of the render pipeline graph,
which contains very minimal subset of the view layer: it includes
custom properties (which gets recursed into), but not the rigid
body simulation.
This fix is mainly suppressing the error print, without changing
the apparent behavior of the graph.
Pull Request: https://projects.blender.org/blender/blender/pulls/106045
Remove `enum eAnimFilterFlags` from `ED_keyframing.h`, the function
parameters that pass them around, and the code that uses them to filter
things.
Only one of the enum values was actually used, `ANIMFILTER_KEYS_LOCAL`.
It indicates whether to only consider keys on the datablock itself (when
set), or in the case of objects, also consider keys on its materials or
shapekeys (when cleared). However, this flag was *always* set, making it
possible to remove the code that handled it.
Finally there was only one function that received a `filters` parameter
that could be different across calls: it would either have value
`ANIMFILTER_KEYS_LOCAL` or `0`. However, the only flag it actually tested
for was `ANIMFILTER_KEYS_MUTED`, which was never actually set. So all of
that could be removed as well.
Just for grepping-through-history sake, these are the flags that were
removed:
- `ANIMFILTER_KEYS_LOCAL`
- `ANIMFILTER_KEYS_MUTED`
- `ANIMFILTER_KEYS_ACTIVE`
- `ANIMFILTER_KEYS_NOMAT`
- `ANIMFILTER_KEYS_NOSKEY`
Note: this is NOT about the flags defined in `ED_anim_api.h`, in `enum
eAnimFilter_Flags`. Note the different names, the one that's removed
doesn't have an `_`.
No functional changes.
Pull Request: https://projects.blender.org/blender/blender/pulls/106028
was assuming such cases are already handled before this function is
called, but seems to be broken on windows and linux. Do
`safe_normalize()` so the tests pass, will check the problem afterwards.
No functional changes.
A set of functions to get the bounds of a channel was created
for the `ANIM_OT_channels_view_selected` operator.
Theses functions will be used by other functions in this file,
so they need to move to the top.
This patch adds an option to the Status Bar: `Scene Duration`
This shows the duration of the current scene in frames and timecode
`Duration: <timecode> (Frame <current frame>/<total frames>)`
The timecode follows the formatting defined in the user preferences.
Pull Request: https://projects.blender.org/blender/blender/pulls/104882
Setting a negative active layer index was possible, causing a crash
when adding new layers (for example).
Clamp the active index when assigning from RNA.
Memory chunks that were in the same hashed bucket but had different keys
were comparing memory unnecessary.
In practice this didn't happen all that often in my tests so the
performance improvement isn't significant.
Follow up to 4f10800094.
These `ME_POLY_LOOP_PREV` are redundant, since they're similar to
the `poly_corner_prev` inline functions. They were also confusing,
since they took an index into the poly and returned an index into
the entire corner range. Instead structure code to use the function
version, and simplify some loops in the process.
Standardize naming, use spans and references for input parameters,
and improve documentation. Now the functions expect the lookups to
succeed as well, they will fail and assert otherwise.
The functions are also simple enough that it likely makes sense to keep
them all inlined
This allows adding spans, arrays, etc. directly to SculptSession, which
simplifies accessing mesh data, especially in #105938. A few files
aren't moved to C++ yes, so I had to add three C accessor functions.
Previously UI view items would support custom drop controllers (so they
could react to data being dragged over them and dropped). This is now
more generalized so the views themselves can do this as well.
Main changes:
- Support calculating a bounding box for the view, so this can be used
for recognizing mouse hovering.
- Rename "drop controller" to "drop target", this is more clear, less
abstract naming.
- Generalize drop controllers/targets. There is a new
`ui::DropTargetInterface` now.
- Add support for drop targets in the `ui::AbstractView` base class, so
custom views can use this.
Pull Request: https://projects.blender.org/blender/blender/pulls/105963
