0
\$\begingroup\$

At some angle my mirror with planar reflection shows this result. Looks like it mix with the sphere reflection capture.

How to fix it?

enter image description here

enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ This is the expected result when using screen space reflections. As the name implies, it's reflecting other stuff that's visible on your screen. It can't correctly reflect the left edge of the painting on the wall, because your screen doesn't show any pixels describing what that looks like. Do you have an option to capture a reflection map at the camera position to use as a fallback in these out-of-bounds areas? It's not a perfect fix, since the perspective will be wrong, but it's somewhat less glaring. Or you can use a second camera rendering for the reflection if you have just 1 mirror. \$\endgroup\$ – DMGregory Aug 25 '20 at 18:06
  • \$\begingroup\$ @DMGregory Thanks for the help. If it only reflect the visible part on the screen. Than why this angle works?(the second picture in the question). The bench and the cabinet is out of the camera capture but the mirror reflect them. \$\endgroup\$ – bao007fei Aug 25 '20 at 18:31
  • \$\begingroup\$ Looks like it's falling back on a separate reflection map, as I described above. I'm not sure why that's not happening in your first image. Maybe the fallback map is only considered a valid source when it's a close match to the reflected rays. Or maybe the engine switches from a screenspace reflection to an alternate strategy at a critical angle. \$\endgroup\$ – DMGregory Aug 25 '20 at 18:36
  • \$\begingroup\$ @DMGregory I set the scale of Planar Reflection back to (1,1,1). Much better than the mirror object size(0.05, 0.03, 1). Most angles and distance act normally immediately. Pretty weird, can't explain. \$\endgroup\$ – bao007fei Aug 25 '20 at 18:50
  • \$\begingroup\$ If that solved your problem, please post your solution as an answer below. \$\endgroup\$ – DMGregory Aug 25 '20 at 18:53
1
\$\begingroup\$

Sometimes the worst result arises like this Which the reflection is totally(sometimes half) white or black.

I accidentally set the planar reflection scale back to default, the artifacts went away. Couldn't explain why, but it worked:D

(Special thanks to @DMGregory)

enter image description here

enter image description here

\$\endgroup\$
1
  • 1
    \$\begingroup\$ My best guess would be that some of the reflection vector calculation occurs in object local space. If that space is non-uniformly scaled, it could mess up those calculations if they're assuming default scale. \$\endgroup\$ – DMGregory Aug 26 '20 at 11:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.