Page 1 of 1

colour banding on dispersive glass caustics

Posted: Sat Aug 29, 2020 2:00 pm
by Canopy
I'm getting really strong kinda colour banding on the edges of glass caustics with dispersion enabled, lowering exposure helps a bit but even with really low exposure I can't get rid of it entirely, is this intended and are there things I can do to mitigate the effect?

Here's two crops of the image I'm working on at different exposures

0 EV
Image
-2 EV
Image

EDIT

For clarity's sake its worth mentioning I do actually want the colour fringing on the caustics, just not with such hard transition between colours, idk if using larger lights or something would do the trick. Currently I'm using spotlights so I can't change the size in the same way you can with point lights

Re: colour banding on dispersive glass caustics

Posted: Tue Sep 01, 2020 7:24 am
by FarbigeWelt
Canopy wrote:
Sat Aug 29, 2020 2:00 pm
Currently I'm using spotlights so I can't change the size in the same way you can with point lights
In my opinion this looks exactly how it should according to physics (dispersion) for point lights.
At the edge of any spectral point light (could be a slit in a dark mask) these colors must happen because there is not any other color to be overlapped after spectrum fan out.

You may try with an area light in a hollow cylinder to get fuzzier color shades.

Re: colour banding on dispersive glass caustics

Posted: Tue Sep 01, 2020 10:33 am
by CodeHD

Perfect Prism Light Angles and The Quest of VanishingColors

Posted: Wed Sep 02, 2020 1:37 pm
by FarbigeWelt
CodeHD wrote:
Tue Sep 01, 2020 10:33 am
Possibly related to this: https://github.com/LuxCoreRender/LuxCore/issues/262
Because I am a rather curious person and because my Avatar shows spectral light dispersion of a prism and further because I did not notice missing parts of white light's spectrum I had to check myself what is going on here I mean at least have a look at current 2.5 alpha color fanning out capabilities.

To be on the safe side regarding spectral light composition I chose mesh material emission with a 6500 K black body spectrum.

After playing with parameter's possibilities and fizzling around with camera positions I had this useful idea of pseudo panoramic view of a huge ring as canvas. Still wondering about what is happening here I may ask readers having enough attention span and curiosity to check my observations themselves if these technical or scientific people have an idea what is behind the thievish spectrum appearing in a certain light angle range.
Prism-Optimal-Light-Angle
Prism-Optimal-Light-Angle
Prism-Optimal-Light-Angle
Prism-Optimal-Light-Angle
Prism2.zip
Prism2
(173.49 KiB) Downloaded 21 times