Page 1 of 1
sss & alpha
Posted: Sat Mar 04, 2017 8:23 pm
by WhyGee
Hi,
I have a specular material with a scattering medium attached. Even with very high density where the object is essentially opaque, the alpha doesn't seem to take into concideration the density of the medium.
Attached is a simple scene. Am I doing something wrong?
v3.05.3.0
Thanks
Yaniv
Re: sss & alpha
Posted: Mon Mar 06, 2017 10:50 pm
by juanjgon
Hi,
I'm testing you scene with the latest 3.06 TEST build and here the density of the medium seems to affect the alpha channel, as you can see in the attached image. Can be a bug already fixed in the 3.06 build?
Thanks,
-Juanjo
Re: sss & alpha
Posted: Tue Mar 07, 2017 7:47 am
by WhyGee
The bottom one should be white not almost black in order to be opaque no?
Re: sss & alpha
Posted: Tue Mar 07, 2017 7:49 am
by WhyGee
If you try to comp it onto a background you will see that you lose the opacity that you see in the IPR with background enabled.
Re: sss & alpha
Posted: Tue Mar 07, 2017 9:17 am
by juanjgon
Hmm, yes. Probably this is an Octane limitation. I'll report it to the Octane devs. to see what can be done.
Thanks,
-Juanjo
Re: sss & alpha
Posted: Tue Mar 07, 2017 10:27 am
by WhyGee
Thanks
Yaniv
Re: sss & alpha
Posted: Tue Mar 07, 2017 11:13 pm
by abstrax
It is not possible to encode colour information into an alpha channel, because it's greyscale only. Due to this fact you can't alpha blending work like a coloured filter, which would be required to be able to incorporate the absorption and transmission colour into the render result. -> We are ignoring it and only reflections are recorded.
What the option "affect alpha" does is separate reflection paths from refraction paths and averaging all reflection paths with an alpha value of 1 and all refraction paths with an alpha value of 0. So if 30% of your paths enter the specular material (i.e. are refracted), the alpha value will be 70%, because 70% of the paths get reflected.
There should be no difference in renderings between different versions. Let us know if you see differences or incorrect results.
Re: sss & alpha
Posted: Wed Mar 08, 2017 9:04 am
by WhyGee
Hi, thanks for the response but I don't understand what the conclusion is? is there no way to properly composite a specular material with a medium?
Re: sss & alpha
Posted: Wed Mar 08, 2017 9:20 am
by abstrax
WhyGee wrote:Hi, thanks for the response but I don't understand what the conclusion is? is there no way to properly composite a specular material with a medium?
No there isn't at the moment. We would need to implement a different alpha channel system (basically an alpha channel for each of the red, green and blue channels). It's on our to-do list (like so many other things), but I don't know when we will get to work on it.
Re: sss & alpha
Posted: Wed Mar 08, 2017 9:31 am
by WhyGee
Is there a way to export the "density opacity" as an auxiliary pass as a solution until the new alpha system is implemented? If not, maybe this could be an easier feature to add as an intermediate solution for the time being?
thanks