Hi, I’ve been trying to implement a post-processing effect in WebXR. At first I didn’t use the WebXRManager, so I had to implemented my own event handlers for WebXR and implement my custom rendering pipeline to render in multiple passes to the xr framebuffer. That worked fine, but i lose the functionalities of the WebXRManager or the VRButton class.
But now I want to do it simpler. I mean I want to do it using renderer.xr.enabled=true and be able to use WebXRManager methods while I am in the XR session.
But after many tests I found that every time I call renderer.render even or when I use the EffectComposer, the camera I pass as argument to the renderer is modified (its projection and model matrices are overwritten. I tried setting WebXRManager.cameraAutoUpdate=false, but even with that option, the cameras are still modified by the render method (while xr session is active). So it is not possible to use renderToTexture with an OrthographicCamera to implement for example a blur filter using a quad.
I think may be one way would be to override the renderer.render method or to derive a class from it. But I wonder if there is a cleaner way to do it.
Any suggetions?
Thanks