My world has a layered camera setup and requires for the reference (main) camera of the world to be set to render with
Clear Flags
set to
Depth only
and only render from players eyes up to 1000 meters in order to maximize performance.
Secondary camera of the world is set to render only a low quality copy of the terrain and surrounding mountains in the horizon from 1000 meters all the way up to 4000 meters before the main camera of the world renders the scene.
Despite the player's POV looking perfectly normal on desktop and in VR, the in-game camera, steadycam and personal mirror all break because they inherit all applicable properties of the reference camera and don't properly clear the render, resulting in unacceptable ghosting in the sky.
I'm asking for the VRCSDK to include an option to set the camera properties of VRChat's in-game user cameras independently from the reference camera, e.g. fields for Clear Flags, Near Clip, Far Clip on the VRCSceneDescriptor so they can still keep functioning without the world author having to provide a completely custom camera solution from the ground up in order to mitigate this issue.
If exposing the properties in the SDK is too big of an ask, at least ensure that the in-game cameras do not have their Clear Flags set to Depth Only or Don't Clear.
Thank you.
P.S. This issue could also be fixed by allowing the world to set the Clear Flags of the player's main camera during runtime with Udon, related post: https://feedback.vrchat.com/udon/p/give-us-an-interface-to-the-main-camera-via-the-vrcgraphics-class