Sampling and General Shadow Settings - Final renders

[ivanjay]

Hi all,

 

So we recently upgraded to Max 2014. I use mental ray. For years I set sampling to 1 and 4 and got good results. Once in a blue moon I needed to go to 1 and 16.

 

Now in Max 2014 they changed the settings and I am not sure what is equivalent. For my first rendering I made the image quality 5 and I left min and max at 1 and 128. I make glossy reflections and shadow quality multipliers 2.

 

The rendeirng is taking FOREVER....

 

Any recommendations on good starter settings to tweak or at least what the equivalent would be?

[blank...]

http://forums.cgarchitect.com/75158-help-needed-mental-ray-artifacts-2.html#post384708

[ivanjay]

http://forums.cgarchitect.com/75158-help-needed-mental-ray-artifacts-2.html#post384708

 

That was a bit helpful although still a bit confusing to me.... What I am gathering is you leave the 1 / 128 alone and adjust the slider accordingly. And, if I understand that correctly they are suggesting that I turn all samples down to 1 on the materials unless it is something that is very glossy or reflective... That one I am a bit nervous doing.

[Matt McDonald]

You are doing what Nvidia recommends but most arch-vis users complain that it is slow. Part of the reason is that unified really shines when you are doing lots of sampling: glossy reflections, AO, depth of field, motion blur, etc. In arch-vis where your samples might just be glossies and AO the overall efficiency of unified doesn't seem to be realized.

 

In Max 2014 you are able to switch the anti-aliaser back to the method you are used to working with.

[ivanjay]

You are doing what Nvidia recommends but most arch-vis users complain that it is slow. Part of the reason is that unified really shines when you are doing lots of sampling: glossy reflections, AO, depth of field, motion blur, etc. In arch-vis where your samples might just be glossies and AO the overall efficiency of unified doesn't seem to be realized.

 

In Max 2014 you are able to switch the anti-aliaser back to the method you are used to working with.

 

So the thought is for my particular purposes I am probably better off going back to how I know it in the past. And just leave it at that...

[Matt McDonald]

I've been experimenting with Unified Sampling for a couple of months now but for our production we still use the classic/raytraced method. I'm saying you wouldn't be wrong to do the same.

[blank...]

What I am gathering is you leave the 1 / 128 alone and adjust the slider accordingly.

 

Yes, but you certainly can experiment with other values, either lower for test renderings or higher for final (which if you have many glossy materials will probably be the case).

 

And, if I understand that correctly they are suggesting that I turn all samples down to 1 on the materials unless it is something that is very glossy or reflective... That one I am a bit nervous doing.

Nothing to be nervous about, that's how unified works. Not just materials - all samples everywhere. Then, if there's noise even with high max value (300 for example) and high quality value (3 or 4), up samples to 2 or 4.

 

Try it, see how it works for you.

[ivanjay]

Yes, but you certainly can experiment with other values, either lower for test renderings or higher for final (which if you have many glossy materials will probably be the case).

 

Nothing to be nervous about, that's how unified works. Not just materials - all samples everywhere. Then, if there's noise even with high max value (300 for example) and high quality value (3 or 4), up samples to 2 or 4.

 

 

Try it, see how it works for you.

 

Is there a global location for change samples to 1? I have lots of materials and that is really cumbersome.

 

However, I am assuming that since my samples were 16,32,64 in cases and I had unified sampling on that would explain the ridiculously large rendering times... Plus a multiplier on the glossy samples.

 

So all samples to 1, and multiplier to 1, and render with unified on? Correct workflow?

 

I am going to render this with the old style and for next rendering post deadline give it a try so i do not tie up my machines.

[blank...]

Is there a global location for change samples to 1? I have lots of materials and that is really cumbersome.

 

Not that i know of.

 

However, I am assuming that since my samples were 16,32,64 in cases and I had unified sampling on that would explain the ridiculously large rendering times

 

Exactly

 

So all samples to 1, and multiplier to 1, and render with unified on? Correct workflow?

 

Correct start. Then you see what needs to be done. Up max for unified to 200, 300. Up quality for unified to 2 or 3. If there are still problems, then go to problematic material and set glossy samples to 2 or 4 (same for lights).

[ivanjay]

Do you find unified to be faster than the traditional in terms of rendering time? I was having a hard time with very slow results. So I just reset all of my samples upwards. However, I have no issue going back to change it back and trying to follow this. Just concerned about lengthy rendering time....

 

And you are saying I also need to reduce all of my light samples to 1? When I went to unified before I did not do that so however Max brought them in (photometric) is what they are set at...

[blank...]

Do you find unified to be faster than the traditional in terms of rendering time?

 

Much faster, especially with highly glossy or detailed stuff. Grass for example, it just flies over it

 

 

And you are saying I also need to reduce all of my light samples to 1?

Like i said, all samples everywhere! Lights, shadows, materials, glossy, AO (if you're using it in A&D). All to the lowest possible - that being one, not zero! (Some values in lights don't go lower than 2)