This new implementation does all downsampling in a single compute shader
dispatch, removing a lot of complexity from the previous recursive
downsampling.
This is heavilly inspired by the Single-Pass-Downsampler from GPUOpen:
https://github.com/GPUOpen-Effects/FidelityFX-SPD
However I do not implement all the optimization bits as they require
vulkan (GL_KHR_shader_subgroup) and is not as versatile (it is only
for HiZ).
Timers inside renderdoc report ~0.4ms of saving on a 2048*1024 render for
the whole downsampling. Note that the previous implementation only
processed 6 mips where the new one processes 8 mips.
```
EEVEE ~1.0ms
EEVEE-Next ~0.6ms
```
Padding has been bumped to be of 128px for processing 8 mips.
A new debug option has been added (debug value 2) to validate the HiZ.
Compared to the previous implementation this has a limit of 65536 lights
per scene. Lights exceeding this limit will be ignored.
This also introduce fine grained GPU light culling, making rendering
many lights in a scene more efficient as long they don't overlap much.
Compatible light panels have been unhidden.
Note: This commit does not include surface evaluation, only light culling.
This replace the previous square rings approach by sampling a disk the
footprint of the search area. This avoids sampling in areas in corners
where there isn't any weight.
This results in much less samples needed to acheive a good enough result.
The max number of samples for an area of 11x11 px is hard coded to 16 and
still gives good results with the final clamp.
The number of samples is adaptative and is scaled by the search area (max
CoC).
The High Quality Slight Defocus is not required anymore. If there is a
quality parameter to add, it would be sample count option. But I consider
the temporal stability enough for viewport work and render can still
render many full scene samples. So I don't see a need for that yet.
This is a port of the previous implementation but using compute
shaders instead of using the raster pipeline for every steps.
Only the scatter passes is kept as a raster pass for obvious performance
reasons.
Many steps have been rewritten to take advantage of LDS which allows faster
and simpler downsampling and filtering for some passes.
A new stabilize phase has been separated from another setup pass in order
to improve it in the future with better stabilization.
The scatter pass shaders and pipeline also changed. We now use indirect
drawcall to draw quads using triangle strips primitives. This reduces
fragment shader invocation count & overdraw compared to a bounding
triangle. This also reduces the amount of vertex shader invocation
drastically to the bare minimum instead of having always 3 verts per
4 pixels (for each ground).
The new implementation leverage compute shaders to reduce the
number of passes and complexity.
The max blur amount is now detected automatically, replacing the property
in the render panel by a simple checkbox.
The dilation algorithm has also been rewritten from scratch into a 1 pass
algorithm that does the dilation more efficiently and more precisely.
Some differences with the old implementation can be observed in areas with
complex motion.
The improvements over the old implementation are:
- Improved history reprojection filter (catmull-rom)
- Use proper velocity for history reprojection.
- History clipping is now done in YCoCg color space using better algorithm.
- Velocity is dilated to keep correct edge anti-aliasing on moving objects.
As a result, the 3x3 blocks that made the image smoother in the previous
implementation are no longer visible is replaced by correct antialiasing.
This removes the velocity resolve pass in order to reduce the bandwidth
usage. The velocities are just resolved as they are loadded in the film
pass.
This modules handles renderpasses allocation and filling. Also handles
blitting to viewport framebuffer and render result reading.
Changes against the old implementation:
- the filling of the renderpasses happens all at once requiring
only 1 geometry pass.
- The filtering is optimized with weights precomputed on CPU and
reuse of neighboor pixels.
- Only one accumulation buffer for renderpasses (no ping-pong).
- Accumulation happens in one pass for every passes using a single
dispatch or fullscreen triangle pass.
TAA and history reprojection is not yet implemented.
AOVs support is present but with a 16 AOV limit for now.
Cryptomatte is not yet implemented.
This adds support to render Curves attributes in EEVEE.
Each attribute is stored in a texture derived from a VBO. As the
shading group needs the textures to be valid upon creation, the
attributes are created and setup during its very creation, instead
of doing it lazily via create_requested which we cannot rely on
anyway as contrary to the mesh batch, we do cannot really tell if
attributes need to be updated or else via some `DRW_batch_requested`.
Since point attributes need refinement, and since attributes are all
cast to vec4/float4 to account for differences in type conversions
between Blender and OpenGL, the refinement shader for points is
used as is. The point attributes are stored for each subdivision level
in CurvesEvalFinalCache. Each subdivision level also keeps track of the
attributes already in use so they are properly updated when needed.
Some basic garbage collection was added similar to what is done
for meshes: if the attributes used over time have been different
from the currently used attributes for too long, then the buffers
are freed, ensuring that stale attributesare removed.
This adds `CurvesInfos` to the shader creation info, which stores
the scope in which the attributes are defined. Scopes are stored
as booleans, in an array indexed by attribute loading order which
is also the order in which the attributes were added to the material.
A mapping is necessary between the indices used for the scoping, and
the ones used in the Curves cache, as this may contain stale
attributes which have not been garbage collected yet.
Common utilities with the mesh code for handling requested
attributes were moved to a separate file.
Differential Revision: https://developer.blender.org/D14916
This module allow tracking of object and geometry data accross time.
This commit adds no user visible changes.
It work in both viewport (*) and render mode, gives correct motion
for any camera projection type and is compatible with displacement (**).
It is a huge improvement upon the old EEVEE velocity which was only used
for motion blur and only available in render.
It is also an improvement for speed as the animated objects do not need to
be rendered a 3rd time. The code is also much cleaner: no GPUVertBuf
duplication, no GPUBatch amendment, no special cases for different geometry
types, no DRWShadingGroup per object, no double buffering of velocity.
The module is still work in progress as the final output may still be
flawed.
(*): Viewport support is already working but there might be some cases where
mapping will fail. For instance if topology changes but not vertex count.
(**): Displacement does not contribute to motion vectors. Surfaces using
displacement will have the same motion vectors as if they were not displaced.
This commit introduce back support for all geometry types and all nodetree support.
Only the forward shading pipeline is implemented for now.
Vertex Displacement is automatically enabled for now.
Lighting & Shading is placeholder.
Related Task: T93220
# Conflicts:
# source/blender/draw/engines/eevee_next/eevee_engine.cc
# source/blender/gpu/CMakeLists.txt
