Black highlights with low refl. glossiness value behind glass

[solipsist]

Hello guys, i found out this kind of behaviour of materials when combination of darker diffuse colors and low reflection glossiness values is used and when object is placed behind glass (interior). I did some testing and the black highlight dissapear when lighter diffuse or higher values of glossiness are used (often the problems starts under 0,4 glossiness). After while i found out that simply ticking Thin (no refraction) in glass material resolve this. As i didnt find any kind of similar issue here in forum and dont know if it is bug or common behavior as in the end I solved the problem i just want to know what is happening here. Thanks for your reply.

[PROH]

Hi. Have experienced similar problems with wood and glass, but found out what I was doing wrong. Maybe you got the same problems.

1) The black parts of the wood in pic. 1, could be caused by the glass and wood intersecting or being coplannar. If that's the case, try to separate them with a little amount (in my case 0.1 - 0.2 millimeter was enough).

2) The noise (or some of it) in the other renders, could be caused (or enhanced) by to high bump values in the wood material. Corona is extremely "bump-sensitive", especially regarding reflections.

3) If you use "Thin" glass, you should use a plane - not a box.

Hope it helps

[maru]

Are you using portals in this scene? I remember something similar on the forums but I can't find it.

[solipsist]

Hello, thank you for your reply & suggestions , but non of those are my case. I did another  test scene  where it should be more obvious what i mean. Its just few balls with different colors and gloss values behind the glass plate (10mm). Diffuse values are 10 / 80 / 160 on level 1, reflection is 255/255/255 on level 0,7, and gloss values goes from 1 - 0,8 - 0,6 - 0,4 and last ball has also 0,4 but is not placed behind glass. I also did this for three different types of lighting. Corona light , corona sun + sky, and hdri (gamma 0,7). It seems that if there is a very bright spot and ray goes through material which is refractive - here it is glass but i try any combination of values and artifact still remains more or less , so its about refraction only, it produces instead of white highlight black highlight. Which is strange. Also as i write earlier this problem dissapear when Thin (no refraction) is ticked.

* No portals were used

[maru]

It looks slightly better in newer versions but not good.

I remembered there was such thread because I created it. :)
https://forum.corona-renderer.com/index.php/topic,5628.msg38823.html#msg38823

So what happens is reflection looks like "caustics" option was enabled for the glass, even if it's not. I was surprised there is no solution for this, because this can happen really often, but unfortunately it looks like we have to wait...

[solipsist]

Hi, thanks for reply, i just dont know if its bug or my mistake ... Forutnately it in this case (windows glass) it could by partialy solved with thin option enabled ...

[andrak]

Just tried this out as I couldn't really believe that this would happen but I get exactly the same result as you. When the object is behind glass, as the luminance value of the material goes down it seems that a black hole effect appears and grows in size related to this luminance value. The effect goes away once the luminance value increases. However when using 999 to disable fresnel and putting a falloff map in the reflection colour slot, this black hole never disappears at any luminance value and even at high gloss like 0.9. At 1.0 gloss it seems ok.

The strength of the light in the scene makes a big difference as well. At lower strengths the black hole fades away. Using a HDR to light rather than Corona sun or a Corona light rendered the object perfectly.

[Rimas]

I'm getting this error now and it bugs the hell out of me on a leather sofa under a frosted roof window! (skylight needs to be frosted because of the roof construction above it and the client specifies it that way and the leather needs to be low gloss because...leather...).
Any fixes for this?!

[maru]

Hi, here is an example workaround for this using rayswitch material. It will produce slightly different results (the frosted glass will not be visible in any reflections - you can fix that by enabling some kind of material in rayswitch's reflections slot (just not solid glass! :) ), it also won't have internal reflections). Img + max 14 file.

[Rimas]

It's not REALLY a workaround if it has to change the intended look :/

Here are the fugly black highlights I'm getting:


For now I can afford to render the scene with frosted glass set to 'thin' (no refraction) and then doing an area render of the frosted glass to comp back in. Luckily I (for now) don't have anything that reflects the frosted window clearly.
But this SERIOUSLY needs to get fixed!

[Ondra]

It is scheduled to be fixed in the next core rewrite. But that did not happen yet, because we are extremely swamped at the moment. I hope it will be fixed in time for version 1.5 or 2.0

[Rimas]

It is scheduled to be fixed in the next core rewrite. But that did not happen yet, because we are extremely swamped at the moment. I hope it will be fixed in time for version 1.5 or 2.0

Thanks for the heads-up!
Is there a roadmap / ETA type thing available somewhere where we could have a look at guesstimated timings or when and what's coming up next in Corona?

[Ondra]

We want to make it. Right now we are hiring, and we will make it as soon as we see what our new programmers could do

[Rimas]

Ok thanks! Glad you're expanding and I can't wait to see the bug and the roadmap sorted! :)

[atelieryork]

I just posted another thread on this with lots of examples to help with diagnosing.

https://forum.corona-renderer.com/index.php/topic,9336.0.html

For me this is a very major issue since the bug is very bad visually on common projects and the workarounds cause incorrect/inaccurate results and also serve to overcomplicate scene management. I would hope this could be pushed high up your priority list if possible!