---
Feature guarded by 'Slate.CopyBackbufferToSlatePostRenderTargets' disabled by default in initial commit.
To use simply call GetSlatePost0 material function after enabling CVar.
No overhead unless a widget is rendered using the material function, overhead is minimal (Single-copy 0.003 ms).
GetSlatePost1-4 require enabling via project settings.
#jira UE-186910
[REVIEW] yohann.dossantos jason.nadro
#virtualized
[CL 28772135 by daren cheng in ue5-main branch]
Uses GetBounds to generate a point from any spatial data.
Single data/Single Connection for now.
Exposed to Library.
[FYI] jeansebastien.guay, julien.lheureux, huw.bowles, adrien.logut
[CL 28761573 by antoine guilbaud in ue5-main branch]
- FHierarchicalBlendTarget now stores each hbias in the chain as part of an inline buffer (the same as before), but can optionally switch to a heap allocation if more than 7 levels are encountered. Using the heap here is very undesireable for performance, but at least it will work. Having more than 7 nested levels of blended hbias indicates a bigger level design problem anyway.
- The type remains 16 bytes, so multiple can still fit neatly on a cache line. There is some manual memory management necessary to support this, but that is necessary to keep the size of the type down.
- Added some automation tests to verify behavior
#rb Max.Chen
[CL 28751545 by andrew rodham in ue5-main branch]
[FYI] sara.schvartzman
Original CL Desc
-----------------------------------------------------------------
Control Rig: Try to expose StandardFunctionLibrary as a public asset
#rb none
#rnx
#tests UEFN
[CL 28714577 by jack cai in ue5-main branch]
Problems:
* We want to reach beta quality for ML inference in editor via NNERuntimeORT however currently the model optimisation, validation, shape inference and model creation are offered by the NNERuntimeORTCpu itself soon to be deprecated
* NNERuntimeORTCpu offer model validation and inference using a customized version of onnxruntime
Solution:
* Provide model optimisation, validation, shape inference and model creation in NNERuntimeORT via the NNEUtilities module itself backed by unmodified official version of ONNX & OnnxRuntime
* Onnx being compiled to the official version matching the onnxruntime we are using.
* The libs are externally compiled
* Windows only for now.
Tests:
* Win64 editor: NNE Smoke tests + NNE all tests + Style transfer demo
* win64 standalone (from editor): NNE Smoke tests + NNE all tests + Style transfer demo
* Win64 standalone packaged (both single and multithread cooking): NNE Smoke tests + NNE all tests + Style transfer demo
* Console: NNE Smoke tests + NNE all tests + Style transfer demo
* Compile non-unity
#jira UE-195403
#rb nico.ranieri
#rnx
[CL 28705882 by florent guinier in ue5-main branch]
