Page MenuHome

Shadow catcher refactor to support indirect light
Open, NormalPublic

Tags
None
Tokens
"Burninate" token, awarded by 3Rton."Like" token, awarded by Schamph."Love" token, awarded by Chrischan."Like" token, awarded by Fracture128."Like" token, awarded by Piquituerto."Love" token, awarded by aliasguru."Love" token, awarded by xdanic."Love" token, awarded by dpdp."Love" token, awarded by sam_vh."Like" token, awarded by dark999."Love" token, awarded by juang3d."Like" token, awarded by EAW.
Assigned To
None
Authored By

Description

More details here:
https://devtalk.blender.org/t/lets-finally-fix-the-shadowcatcher/691

If any developer from the Render & Cycles team is interested on this, please bring the content from devtalk to here, in a tangible way. If not we can just remove this task, honestly.

Small note: "An efficient GPU implementation is not straightforward, though it can be solved probably" -- Brecht

Details

Type
To Do

Event Timeline

I've attempted summarising the DevTalk post. Of course the original article and its reactions contain much more information - and a bit of sidetracking - but I've tried to condense the gist of it as best I could. With just a hint of personal opinion here and there.

Primary and secondary rays on the shadow catcher

  • The shadow catcher obviously appears to capture occlusion and shadows, but ignores any colour contributed to by scene lights and the World.
  • Emission shaders and secondary bounces from other objects are ignored as well. This includes diffuse colour bleeding and glossy reflections.

Shadow catcher as seen in secondary rays on other objects

  • In secondary rays on other objects, the shadow catcher is seen with its original material, instead of a camera mapped backplate or environment map.
  • This camera mapping is necessary on a per frame basis and therefore can't be expected to be done by the user.

Suggestion: Shadow Catcher BSDF

  • Basically a Principled BSDF whose primary rays are invisible to the camera.
  • Automatic camera mapping of the backplate or environment texture, and usage of this when seen in secondary rays on other objects.

Storage
The render result of a shadow catcher layer is quite complex as it needs to contain all of the above in a single image. This means that a pixel's colour and its opacity are not at all guaranteed to be aligned. A popular workaround is to split shadows/occlusion and emission/GI, and respectively multiply and add those layers onto the backplate, but I believe it's easily possible to store the complete result with premultiplied alpha, and composite without a hassle.