Shadow Issue - is this the same

[jacobhutson]

had an issue with normal maps and shadow quality and went looking for why and found this
so no maps, corona materials
surely just pumping more polys at it is not the answer
its almost like shadow is not getting any subdivisions

scene is max2016 cr 1.3 release
but simple to make anyhow

any feedback
render window on left is vray 3.2

[romullus]

As you already figured out, it's shadow terminator problem: https://coronarenderer.freshdesk.com/support/solutions/articles/5000516258

As you can see it can be quite succesfully healed by subdividing mesh. While it's not always convenient or desirable solution, it's the only one we have for now. Every or almost every raytrace engine is suffering from that AFAIK, although some suffers more than others.

BTW, is there a reason why you do unfair comparison with Vray? Shadow terminator is much more prominent when there is strong directional lighting. Introduce same lighting as in Corona's example and Vray most likely will show shadow terminator artefacts too.

[snakebox]

As you already figured out, it's shadow terminator problem: https://coronarenderer.freshdesk.com/support/solutions/articles/5000516258

As you can see it can be quite succesfully healed by subdividing mesh. While it's not always convenient or desirable solution, it's the only one we have for now. Every or almost every raytrace engine is suffering from that AFAIK, although some suffers more than others.

BTW, is there a reason why you do unfair comparison with Vray? Shadow terminator is much more prominent when there is strong directional lighting. Introduce same lighting as in Corona's example and Vray most likely will show shadow terminator artefacts too.

But Vray don't... and Corona does it still even subdivided.



numbers are the number of segments in each sphere. As you can see already at 22 segments with vray the problem is completely gone in comparison.  With corona I still easily spot it with 60 segments on a sphere. So comparing it to Vray isn't right.. Vray does a WAY better job of this particular issue. whether corona does it correctly or not, I don't know.. but it doesn't look very good.

[maru]

Take a look at these:
https://forum.corona-renderer.com/index.php/topic,8240.msg54579.html#msg54579
https://forum.corona-renderer.com/index.php/topic,3279.msg24229.html#msg24229
https://forum.corona-renderer.com/index.php/topic,443.msg3268.html#msg3268

[jacobhutson]

so its not a simple fix then. Chaos group seem to have it sorted in brute force mode, will go back and try ppt.
I think VRAY is a fair comparison as that's who you are competing with now....... i think you will find you will be held up to chaosgroup alot, which is testament to the great job you have done in such a short time.
i will say its not something ive really noticed too much in production work and only noticed it from trying to fault find the real world map bump issue, but if I had to do a really really white scene with a sun and some curved objects i imagine it would be an issue.
animated that would freak me out as those triangles would jump like anything. If i get time ill see what happens.

[Ondra]

we will try to hack it away in 1.5

[Juraj]

Looking forward to this. I see this sparsly poping up but I guess most people don't outright spot it (I sometimes noticed it only on higher-res final).

What I do now is keep turbosmooth (viewport 0, rendertime X-value) modifier on every organic fabric model. It easily goes up to 2 million poly for pillow mesh....