The DBuffer layout and most code is the same as for the desktop renderer.
Added new CVar r.Mobile.DBuffer so that is disabled by default. DBuffer for the SM5 renderer on mobile is enabled by default.
#rb Dmitriy.Dyomin, Sebastien.Hillaire
[CL 28598190 by florian penzkofer in ue5-main branch]
* Add neural profile to config which neural network to use. Texture index is used by default where [-1,3,H,W] is supported. Use Buffer index to support [-1,C,H,W]. Need to config the batch size override or the tile size correspondingly.
* Config in the post process material which neural profile to use and enable `used with neural network`.
* Use neural input/output to control the input and output of the neural network. For neural input, [B,C, U, V] is the index, input0 takes 3 effective channels, mask determines if we should use it as input to the network. Only one node can be used per material; Multiple neural outputs can be used. For texture indexing, automatic bilinear downscaling and upscaling is applied. For Buffer indexing, the user has full control to access the value ( user needs to perform their own filtering).
#jira UE-197119
#rb aleksander.netzel,nico.ranieri,jason.nadro,massimo.tristano
[CL 28569169 by tiantian xie in ue5-main branch]
[FYI] Tiantian.Xie
Original CL Desc
-----------------------------------------------------------------
Add neural post processing (experimental).
* Add neural profile to config which neural network to use. Texture index is used by default where [-1,3,H,W] is supported. Use Buffer index to support [-1,C,H,W]. Need to config the batch size override or the tile size correspondingly.
* Config in the post process material which neural profile to use and enable `used with neural network`.
* Use neural input/output to control the input and output of the neural network. For neural input, [B,C, U, V] is the index, input0 takes 3 effective channels, mask determines if we should use it as input to the network. Only one node can be used per material; Multiple neural outputs can be used. For texture indexing, automatic bilinear downscaling and upscaling is applied. For Buffer indexing, the user has full control to access the value ( user needs to perform their own filtering).
#jira UE-197119
#rb aleksander.netzel,nico.ranieri,jason.nadro,massimo.tristano
[CL 28506280 by aleksander netzel in ue5-main branch]
* Add neural profile to config which neural network to use. Texture index is used by default where [-1,3,H,W] is supported. Use Buffer index to support [-1,C,H,W]. Need to config the batch size override or the tile size correspondingly.
* Config in the post process material which neural profile to use and enable `used with neural network`.
* Use neural input/output to control the input and output of the neural network. For neural input, [B,C, U, V] is the index, input0 takes 3 effective channels, mask determines if we should use it as input to the network. Only one node can be used per material; Multiple neural outputs can be used. For texture indexing, automatic bilinear downscaling and upscaling is applied. For Buffer indexing, the user has full control to access the value ( user needs to perform their own filtering).
#jira UE-197119
#rb aleksander.netzel,nico.ranieri,jason.nadro,massimo.tristano
[CL 28494917 by tiantian xie in ue5-main branch]
* Missing ThinApp.sh
* Missing some quotes around a variable in GenerateUniversalDSYM.sh
* Correctly filter out the stub .app's and such when copying the Staged data directory into the .app
* Disabled the Mac's Sign To Run Locally option by default, as that causes Xcode's Validation/Upload to not have the team name embedded in it, causing a hassle while pushing up to AppStore/TestFlight
* Fixed the PRODUCT_NAME for BP projects
* Made CrashReportClient be a sandboxed app that inherits from parent
* Fix Hybrid apps to check all platforms before in the project generator, so that if a project needs a temp target for IOS (via a plugin)
* For hybrid projects, don't append Game to the end of the Game target names, as that is the default, and it makes for better named products
* Disabled the LoginFlow module from OnlineFramework plugin - this was causing issues with having CEF and EpicWebHelper embedded into a sandboxed .app, and LoginFlow isn't seemingly actually used by anything
#jira UE-196297,UE-196299,UE-196296,UE-196300,UE-196295
#rb adam.kinge
[CL 28463729 by josh adams in ue5-main branch]
Added override cook platform for use by content beacon to the datadrivenplatforminfo.
[REVIEW] [at]Josh.Adams
[CL 28223391 by daniel lamb in ue5-main branch]
* Added an option to be able to consider the paint layer's hardness parameter when using the erosion tool. Set to false by default because this used to be misleading for users
* Set hardness to 0 by default for landscape layer info assets (for the same reason as above)
* Added several missing tooltips
* Added support for double-clicking on a paint layer thumbnail to open the corresponding landscape layer info asset
#rb chris.tchou
[CL 28185367 by jonathan bard in ue5-main branch]
Cooker: When launched from the editor, use a different setting for cookprocesscount, so licensees can reduce amount of cookprocesses (and therefore the amount of memory) used when the editor is running.
#rnx
#rb Zousar.Shaker
#jira UE-191615
[CL 27900751 by matt peters in ue5-main branch]
