`VelloScene` doesn't respect Bevy `RenderLayers` about bevy_vello HOT 6 CLOSED

@dannymcgee Let me know if I'm offbase from your needs.

Likewise, I'm adding @musjj and @nixon-voxell because they had a similar issue on voxell-tech/bevy_motiongfx#40. Please feel free to give thoughts.

from bevy_vello.

So it looks like Bevy UI also ignores RenderLayers. 😂 Upon further inspection, there's actually a TargetCamera component in Bevy UI that serves this purpose. That doesn't fix my issue with non-UI stuff rendering to the UI camera, but I guess I'll have to come up with another solution for that. Closing this since it matches Bevy's behavior.

EDIT: Just kidding, Bevy does respect the render layers.

from bevy_vello.

This is definitely something I'd love to accept in a PR, or get to as time permits.

from bevy_vello.

Up to this point, RenderLayer was something I've never worked with. Currently bevy_vello ignores RenderLayers entirely, so at first it was unclear why adding a RenderLayer::layer(1) component to a VelloSceneBundle would make it not render.

I understand this a little better now and I'm committed to pushing this across the finish line.

Here's what I understand, for others:

• Changing the camera order will change the ordering of rendering cameras. This is a possible way to make bevy_vello render on top or on bottom of your 3d space. (camera order 1 or -1 respectively should do the trick).
  
• Render layers are used to tell a camera what it can and can't see. An entity on Render Layer 0 is never seen by a camera on Render Layer 1.

Re: The situation where a camera has a RenderLayer: I believe having multiple cameras with different RenderLayers would mean internally we would need a new framebuffer for every camera. Then we would need a check to ensure each camera is only rendering the items in its render layer.

from bevy_vello.

I'll add the reason I was looking into this as well:

I have a Camera2d where I want to render all ui from a bevy_vello powered ui crate, and a Camera3d where I want to render "everything else" (gltf scenes, gizmos, effects, etc. game stuff).

If the Camera2d doesn't have a RenderLayers specified, then it will render a bunch of "extra" stuff, such as gizmos I'm using in the 3d camera. So ideally I would be able to order +1 on Camera2d, which places it over the 3d camera, and use RenderLayers to target all vello content to the Camera2d. I believe this still fits the "one pass" model of vello as it is only one layer.

from bevy_vello.

Fixed via #68 - Please re-open a new issue if there are further render layer issues.

There is also a new example for render layers.

from bevy_vello.