Added new shader reinitialization signaling
This was done while working on "ATOM-15728 Shader Hot Reload Fails in Debug Build", but it turned out these changes did not actually fix the issue (or any other known hot-reload issue). Still, these improvements are appropriate as they correct logical oversights.
ShaderVariant was not listening to asset reloads. It needs to know when the ShaderVariantAsset reload happens so it can reinitialize it's members as well as propagate reinitialization messages. I added a member for the ShaderAsset as the class needs this to reinitialize itself. So now the class listens for reloads of both the ShaderVariantAsset and the ShaderAsset.
Shader was not listening for ShaderAsset reinitialization events.
Updated the API for ShaderReloadNotificationBus's OnShaderVariantReinitialized to include the ShaderVariant which is the most relevant information (the other information wasn't really being used anyway).
Testing: Ran ASV full test suite in dx12 and vulkan, saw only known issues. Tested hot reload in Material Editor and main Editor.
See also aws-lumberyard/o3de-atom-sampleviewer#118
This was done while working on "ATOM-15728 Shader Hot Reload Fails in Debug Build", but it turned out these changes did not actually fix the issue (or any other known hot-reload issue). Still, these improvements are appropriate as they correct logical oversights.
ShaderVariant was not listening to asset reloads. It needs to know when the ShaderVariantAsset reload happens so it can reinitialize it's members as well as propagate reinitialization messages. I added a member for the ShaderAsset as the class needs this to reinitialize itself. So now the class listens for reloads of both the ShaderVariantAsset and the ShaderAsset.
Shader was not listening for ShaderAsset reinitialization events.
Updated the API for ShaderReloadNotificationBus's OnShaderVariantReinitialized to include the ShaderVariant which is the most relevant information (the other information wasn't really being used anyway).
* Extend MeshFeatureProcessor to allow changing the mesh bbox, which requires re-compute the culling data for that mesh
* Update actor mesh bbox when EMFX actor instance bbox changes.
Also use the actor instance global bbox to compute the local bbox for the skinned render mesh, instead of the using the static bounds based bbox, as not every actor instance is going to be using the static bounds bbox.
* Store per-instance mesh AABB in the right place.
In the MeshInstanceData, which is unique per instance, instead of in the Model, which is shared between all instances.
For greater clarity, also remove Model::m_aabb and the corresponding getter and setter, as it isn't immediately obvious whether this gets the model asset bbox or the mesh instance bbox. Callers should instead be explicit about which bbox they want.
* Bug fix: model asset is not necessarily ready in AcquireMesh
* Remove now-unused forward declaration
* Update MockMeshFeatureProcessor with SetLocalAabb/GetLocalAabb
Checked for a valid quality level in DiffuseGlobalIlluminationFeatureProcessor::SetQualityLevel.
Initialized the quality level to Low in DiffuseGlobalIlluminationComponentConfig.
* [ATOM-15472] Shader Build Pipeline: Remove Deprecated Files And Functions That
Predate The Shader Supervariants
These are the essential impactful changes as a result of deprecating the
ShaderResourceGroupAsset.
* Addressed feedback by @moudgils. Better comments in header files.
* More updates related with deprecation of ShaderResourceGroupAsset
* Deleted the temporary version 2 classes.
* Updated version of the shader asset builders.
* Updated version of all the shader related classes impacted
by the Supervariant concept and deprecation of ShaderResourceGroupAsset
* Changes to *.pass and DGI, Reflections and RayTracing.
* changes to material related assets
* changes to core lights
* Changes to auxgeom/dynamic draw.
* changes to decals, lyshine, imguipass
* changes to RPI Pass classes
* Shader for SceneSrg, ViewSrg and ForwardPass Srgs.
* changes to mesh, skinned mesh, Morphtarget.
* Fixes to RayTracingPass.cpp & now allow empty srg in shaders.
* Updated Atom_RPI.Tests
* Simplified InstanceDatabase by removing AddHandler
------------------------------------------------------------------------------------
* Updated DiffuseGI precompiled shaders.
Added RayTracingSceneSrg and RayTracingMaterialSrg shader asset.
Updated ShaderAssetCreator::Clone to handle the supervariant when processing root variants.
Co-authored-by: Doug McDiarmid <dmcdiar@amazon.com>
------------------------------------------------------------------------------------
* Changed semantics for some PassSrg to SRG_PerPass_WithFallback.
AuxGeom/FixedShapeProcessor.cpp requires SRG_PerDraw on ObjectSrg.
Removed names of SceneSrg and ViewSrg from RPISystemDescriptor.cpp
* Moved ShaderLib/Atom/Features/DummyEntryFunctions.azsli
To Gems/Atom/RPI/Assets/ShaderLib/Atom/RPI/DummyEntryFunctions.azsli
Removed redundant checking for finalization in
ShaderResourceGroupLayout.cpp
* Fixed race condition bug for Shader::FindOrCreate.
InstanceDatabase<>::CreateInstance() needs to be atomic
for instance creation and initialization.
Added optional InstanceHandler::CreateFunctionWithParams to accomodate
to the needs of Instances that need more than an asset reference
to be able to be created an initialzed.
Removed ShaderResourceGroup::FindOrCreate() only ::Create is available
now.
* Renamed scene_and_view_srgs.* as SceneAndViewSrgs.*
Changed GetAzslFileOfOrigin for GetUniqueId
* Fixed unit tests.
* Reverted the serialization name of m_uniqueId back to
"m_azslFileOfOrigin" so precompiled shaders don't fail
in layout comparison.
* Fixed AtomCore.Tests
Removed non-applicable test. InstanceDatabase.AddHandler() is not
available anymore.
* The Null rhi is re-enabled for shader compilation.
Signed-off-by: garrieta <garrieta@amazon.com>
First version of temporal antialiasing and contrast adaptive sharpening for GA. Works well in most cases but still has a few issues that will need additional time. This is only the passes and shaders with no exposure to the editor. TAA and CAS can be turned on by enabling their respective passes in the pipeline.
All of the code has been previously reviewed in smaller PRs into the taa_staging branch:
aws-lumberyard-dev#29
aws-lumberyard-dev#53
aws-lumberyard-dev#73
aws-lumberyard-dev#79
aws-lumberyard-dev#84
Main issues:
- Bloom doesn't play nice with TAA and seems to greatly amplify any flickering
- AuxGeom jitters with the camera, so TAA doesn't currently work well in editor
- Transparencies don't have correct motion vectors. History rectification keeps this from looking too bad, but could still be improved
- There is still more that could be done to inhibit flickering, usually from specular aliasing
- Motion vectors aren't correct on POM unless PDO is turned on, which can result in some blurring during motion.
- SSAO can contribute to flickering in its default half res configuration. Changing this to full res mitigates the problem.
Squashed merge of the following:
* [ATOM-13987] Initial checkin of Taa pass.
* TAA pass setup WIP. (does not work yet due to pass configuration issues).
* Taa WIP - Camera motion vectors fixed and hooked up. TAA does simple reprojection and rejection based on depth.
* Small update to use lerp and add some comments.
* Fix issue with attachments not being set up on bindings at initialization. Fixing issue with half-pixel offsets in TAA shader
* - Motion vector passes now use the same output with mesh motion vectors overwriting camera motion vectors.
- Taa pass now works with multiple pipelines.
- Cleaned up TAA shader a bit.
* Fixes from PR review.
* Adding check for multiple attachments of the same name with different resources in Pass::ImportAttachments().
* Adding camera jitter with configurable position count. Updated TAA to blend in tonemapped space.
* Fixes from PR review. Fixing camera motion vectors for background (infinite distance)
* Updates to taa shader from PR review
* Adding a rcp input color size.
* Fix comment on PassAttachment::Update()
* Updates for PR review.
* Fixing missing const on the FrameAttachment* in Pass's call to FindAttachment()
* Taa WIP - Adding filtering to both the current pixel and history. Adding rectification based on variance clipping. Adding some basic anti-flickering. Removing rejection based on depth.
* Updates from PR code review. Mostly better commenting and naming.
* Adding contrast adaptive sharpening based on AMD FidelityFX CAS to help with the softness added by TAA.
* Changing to using luminance for sharpening instead of just green. Added some comments.
* Moving Taa's NaN check to a better location. Disabling TAA and sharpening in prep for check in.
* Updates from PR feedback.
* ATOM-15658 Better option of CreateCommonBuffer requires unique buffer name
- Change the CreateCommonBuffer function to not require an unique name by default.
- Remove the code for generating unique buffer names.
- Add buffer name to BufferAsset so it can be used for device object name instead of using asset file name.
- Change RPI::Buffer to use BufferName_AssetUuid as attachment id.
Added a loop to the skin shader that will sample from wrinkle masks, multiply them by a weight, combine them, and use them instead of vertex colors for wrinkle map blending
Added an array of masks, an array of weights, and a wrinkle mask count to the DefaultObjectSrg. -Will create a follow up task to handle this a better way.
Removed motion vector (for now) from skin.materialtype since we're not using them, and removed depthtransparent since skin doesn't support transparency
Added an interface to the MeshFeatureProcessor to get the object srg
Wrapped srg->Compile in if(srg->IsQueuedForCompile()) to prevent compiling twice --This doesn't stop a race condition if both happen at the same time, but that is at least far less likely. It will need a better solution later.
Added a function to the MorphTargetExporter that will check to see if a texture that matches the blend shape name exists in a particular folder, and adds a reference to that image to the MorphTargetMetaAsset --Only supports .tif, and doesn't automatically re-process the .fbx if the folder is updated. These can be improved in later iterations
Added a null check in MaterialTypeSourceData.cpp to fix a crash I ran into
Added a for loop in two places to look for the first submesh that has a morph target, instead of just using the first to check if a lod has morph targets or not. --I have a better fix for this, but it involves more areas of the code, so I'm saving that for another change.
Modified AtomActorInstance to look for any morph targets that have a wrinkle mask reference
Then each frame, for any morph targets with non-zero weights that also have wrinkle masks, it updates the mask array, weights, and count on the object srg.
Previously, the SkinnedMeshFeatureProcessor assumed there would only be one skinning pass. However, that's not always the case. When rendering with track view, the feature processor was getting a pass that only updated once every three frames, which could lead to a condition where a skinned mesh was released, but the pass never submitted and cleared the previously added dispatch items, and one or two frames later it would go to submit after the skinned mesh and all of its resources had already been released.
-Modified the skinning and morph target compute passes to pull dispatch items from the feature processor instead of the feature processor pushing them to the passes.
-If more than one skinning (or morph target) pass is active in the frame, whichever one is first will submit all the dispatch items, and clear the feature processor's dispatch items before the next one tries to submit anything
-Moved the logic for caching shader options from the SkinnedMeshComputePass to the SkinnedMeshFeatureProcessor, since there may be more than one pass but only one feature processor per scene
* ATOM-4661 Improvement with pass templates registration from data
- Update PassLibrary so it can load pass templates from more than one files and report duplicate pass templates.
- Added load templates events to pass system so the handlers from any gems can load their own pass templates.
- Added PassSystemInterface::OnReadyLoadTemplatesEvent::Handler in FeatureCommon gem's CommonSystemeComponet to load the PassTemplates.azasset in featureCommon gem.
- Misc: moved BindlessPrototypeSrg.asli from RPI to ASV project; fixed an assert issue when exit ASV
The bug was reported that copy and paste did not work with the material component.
Copy and paste to take the worked fine.
All of the material assignments/overrides get mapped using the LOD and asset ID of materials provided with the model.
The asset IDs of materials exported by atom builders, using the scene API, are the combination of the same UUID as the model asset ID and the unique sub ID that is now hashed from the material name provided by the DCC tool.
If we map material assignments using the entire asset ID that was generated in the model builder then the mapping will only work with that specific model.
This change updates the material assignment ID equality operators and hash function to only use the sub ID portion of the asset ID.
As long as the sub IDs are generated consistently the material assignment mappings will be portable to models with the same material names.
Also moved material assignment structures to atom common features static library so this was to be moved to cpp files
* Added clone methods for the buffer, model lod and model assets.
* Cloning works via the creators.
* Mesh handle/instance now knows about the model asset it originated from as well as the cloned model asset that we need for instancing until instancing works without the dependencies to the asset ids.
* MeshComponentRequestBus returns the original asset id and the model asset. If you need access to the clone, go by the model.
* Cloth component mesh now gettings its model asset from the model as it needs to work on the clone.
* Moved the requires cloning function from the mesh loader in the mesh feature processor to the mesh component controller.
* As we need the model asset to be loaded before we can check if it requires cloning or not, we couldn't pass in a bool from the controller but had to use a callback.
* Using asset hint instead of asset id for the model lod asset creator error.
* Storing a map of source buffer asset ids and the actual cloned buffer assets rather than just their ids.
* Using the buffer assets from the new map directly and removed the search process in the mesh loop where the asset views are created.
* Fixed the vegetation mocks.
* Using the actor's mesh asset when emitting the on model ready event for the mesh component notification bus.
* Mesh components notifications connection policy changed to adapt to cloned model asset.
* Handling empty meshes in model lod asset creator.
* Removed the requires cloning callback from the mesh feature processor and made it a parameter to the acquire mesh function
* Fixing mocks and unit tests
Chris Santora
mnaumov
mrieggeramzn
Chris Santora
AMZN-mnaumov
yuriy0
Roman
dmcdiarmid-ly
Doug McDiarmid
mriegger
moudgils
galibzon
Anton Michels
antonmic
moudgils
Ken Pruiksma
Qing Tao
guthadam
hershey5045
Esteban Papp
greerdv
greerdv
jiaweig
jiaweig
AMZN-stankowi
Tommy Walton
Terry Michaels
Michael Riegger
Guthrie Adams
kberg-amzn
karlberg
Chris Burel
Benjamin Jillich