Caustics faceting

[Xntric]

Hey Guys.

I'm goofing around with the Caustics solver and i have gotten this problem.

I get a weird faceting effect on my displaced sphere. The geometry should be dense enough with Turbosmooth, and i even tried setting displacement settings to screensize 1.

It seems to be coming from my caustics.

Does anyone know a solution?

[Xntric]

I should add that this was made with the latest daily build

[TomG]

Sorry, not sure what you mean by weird faceting effect, can you give more info? A render without caustics, and a screengrab of the geometry, would maybe help us see what you think is odd with the caustics version. Bear in mind that the object will be refracting the caustics from the ground through it, and depending on the shape of the object, the caustics, etc. how the caustics look seen through it may be "faceted"? To me nothing jumps out as odd in the image.

[Xntric]

If you look closely you can see the faceting in the sphere.

However i found a solution. I needed to subdivide the mesh even more to get rid of it. I didn't think this was necessary with displacement in Corona, since i believe it makes a new mesh at rendertime, and the density of that mesh is controlled by the Screenpixel setting, right?

But going from screenpixel 3 -> 1 made no difference, turbosmooth did.

[Xntric]

I should  add that the faceting didn't occur when Caustics were disabled

[sprayer]

are you using normal map?

[pokoy]

You could disable auto bump, I think it's enabled by default, and see if it's related.

[romullus]

I should  add that the faceting didn't occur when Caustics were disabled

I think you can't see this, because everything is very dark. Put a light or a plane with self illumination behind that object, so it can be seen in refraction and inspect the render for artifacts again. I think the issue has nothing to do with caustics, but rather with displacement itself.

[maru]

How is the object made? Displacement? Modifiers? Procedural map? Bitmap?

[Xntric]

The mesh was made with procedural noise displacement. No CoronaDisplacementMod though

I tried adding light and brighten it up, but it was still only noticeable when i enabled Caustics.
Even if it isn't the Caustics, i don't understand why it would appear, and why the fix was to subdivide the mesh even further.

Could it be that the caustics solver makes some sort of approximation of the mesh?

[romullus]

Could you attach problematic scene or object, so somebody could inspect it for errors?

[maru]

It is possible that caustics don't consider shading normals, similar to SSS - https://coronarenderer.freshdesk.com/support/solutions/articles/12000012390
Could you send us this scene? (or any other where the issue can be seen)

[burnin]

Indeed, it can happen with Volumetrics.
If done as instructed here: https://www.stateofartacademy.com/en/corona-4-caustics-tutorial/

This glass is Diffuse (0,0,0) Reflection level (1.0) IOR and FRESNEL IOR (1.54) Volumetric scattering (Dispersion color HEX:b3c7ba, Distance 2cm).

... Play around a bit, see what happens.