prepass based SSS and Vraqy 3.4

[ben_hamburg]

Hi there,

I just upgraded to vray 3.4 because of the support of max 2017 and I immediately found two issues that I need to get fixed asap...:

We do a lot of medical animations with scenes almost completely consisting of VrayFastSSS materials. Since we have very short turnarounds for these animations most of the time, I found the following combination to work best for me:

I mostly use "prepass based illumination" in the SSS settings and "adaptive subdivision" as my image sampler.
That usually gave me good results with very fast rendertimes.

I can probably live without adaptive subdivisions (the new sampling methods seems to do a good job "out of the box" without a lot of tweaking), but the prepass based illumination now produces ugly colored blotches.
I already tried a couple of things, the only thing that brings down the samplings is lowering the prepass rate, but the lower it gets the less of an SSS effect I seem to get.
Using "raytraced" as my multiple scattering method produces good results, but rendering times are 4-5 times higher (at least with my first tests).

So...can you help me out here? Is there anything I need to do differently to get prepass based illumination to work again...? Or is there a way to speed up the raytraced method (I´m not even using raytraced reflections)?

[ben_hamburg]

Ok, I already found the maxscript snippet to use the old adaptive subdivision method, but this seems terribly noisy now, at least with the raytraced SSS material.

[vlado]

It would be best to post a snippet of your scene or send it to me at vlado@chaosgroup.com so that I can check it out.

Best regards,
Vlado

[ben_hamburg]

Hi Vlado,

sorry, production caught up with me and I didn´t find the time to prepare a scene. But as it sems, its through all my scenes, the problem with the prepass based approach doesn´t work anymore. Maybe you can reproduce it by upping the scale, the higher the scale, th more blotches I get. Looks a lot like irradiance map blotches to me...
For now I´ll have to revert to the last version, let me know if you find anything or it gets reported by somebody else, I´ll try to fix up a scene asap...

[svetlozar.draganov]

It would be very difficult to provide any support or fix without a scene so if you have a chance please prepare one along a screenshot of the result from your environment and send it to us.

[ben_hamburg]

I could now, but I´m a little worried about my now working setup.
Can I run two versions of Vray parallel? Right now I´m working with Max 2016 and Vray 3.3 on my workstation but I also have Max 2017 installed. Could I just install Vray 3.4 for Max 2017 without messing with my working Setup? Because my workstation is also setup as licence server and I reverted to Vray 3.3 on all our rendernodes...

You can´t reproduce the scene with a very simple setup? Just a teapot, vrayFastSSS2 material with lets say a scale of 100 and prepass based illumination should already give you a clear look at the issue...

[^Lele^]

You can´t reproduce the scene with a very simple setup? Just a teapot, vrayFastSSS2 material with lets say a scale of 100 and prepass based illumination should already give you a clear look at the issue...

There's object scale missing there, and SSS depth.
However, i cannot replicate the behaviour you are talking about: raytraced and prepass largely agree in look, and there is definitely no particular issue.

A scene will be needed.

[ben_hamburg]

I can work with 2017 and vray 2.4 on another workstation, so I prepared a scene for support.

[svetlozar.draganov]

I managed to reproduce the issue and already added it into our system.
The strange thing here is that the same V-ray version 3.40.02 render correctly on Max 2016 but incorrectly with Max 2017.

[^Lele^]

ah-ah, that's why it worked with my 2016 nightly.
I'd better leave this to the pros, clearly. :P

[ben_hamburg]

I see...Well, let me know as soon as this is fixed (if its not an Autodesk Issue...), because we can´t migrate to 2017 before that...

[ben_hamburg]

Hey there, any news on this issue?

[vlado]

There is a code review for the fix, but it is not complete yet.

Best regards,
Vlado

[ben_hamburg]

hey there,

just checking in if there is some progress on this issue...it still prevents us from moving to Max 2017...And since the beginning of the year is most of the times a bit less stressful, we tend to do our upgrades then...
Raytraced SSS is still a lot slower then the prepass based approach and especially with big scale multipliers, where we don´t see much shadows, the results are almost identical.
Even though I had to increase the prepass rate from -1 to 1 recently in some scenes, since I´d get flickering otherwise with very thin geometry...

[svetlozar.draganov]

The reported issue is with status Unresolved but according to my tests with the provided scenes this issue has been fixed with the latest official 3.50.01.
Would you please test that version on your side and give a feedback whether it works as expected or not?

Looking forward to hearing from you.