Rhino WIP Feature: Caustics & Gem Dispersion

We added Caustics and Dispersion to Rhino WIP!

What are Caustics and Dispersion?

These are effects that simulate how light behaves when interacting with reflective or transparent materials.

Caustics are the bright patterns formed when light rays are focused or concentrated after reflecting or refracting through a surface.

Dispersion is when light splits into different colors because each color bends differently when passing through a transparent material.

Why are they useful?

Caustics and Dispersion are useful for creating realistic images of jewelry or consumer products that contain reflective or transparent elements. Think of gems or a wine glass. It will simulate the sparkling colors in a diamond or bright patterns under a glass.

How to use them?

  • To activate these, select the Product setting in the Rhino Render Advanced Settings on the Rendering panel, under Presets.

  • To deactivate these, select the Architecture setting. This options renders exactly as the system did before presets were introduced.

Try it out in Rhino WIP. Make sure to give us feedback in this thread!

Hi! This is good to see progress on implementation of deeper rendering options.

It would appear the following notes:

  • Renders require higher levels of samples (expected)
  • The soft caustic (integrated) appears to be good, with nice soft edges as expected
  • The dispersed caustic is poor, with a denoiser almost being essential due to high levels of speckle.
  • The RGB nature of Cycles is very evident at macro views, where the separation visualises as Huffman-like compression glitches. This is more evident at higher contrast imaging.

these settings are indeed very expensive to compute and will require higher samples and more importantly some patience as they run.

Hi Kyle.

Yes, I agree that more samples are needed for such calculation. For a modern 40-Series GPU, I am fine with 150k samples. However, the samples should decrease the speckling over time, not introduce more as the integration continues. Moreover, the speckles seem almost randomised, and do not cluster as expected for such a calculation.

I think there is some issue with convergence with the dispersive caustic that is throwing the calculation off severely. This seems to be exaggerated as Cycles is RGB only.

this is kind of a big improvement. I tried it with an old scene with 1500 samples - denoised. HDRI Light (a room with 2 Windows)