Cheatsheet: How to get a Cleaner image with SP3

[^Lele^]

Could you send me or vlado a scene on the chaosgroup email (or link to)?
Many a thing could be contributing to it, hard to tell without the scene.

[eyepiz]

Why did you chose to turn Max ray intens off

I did a few test with and without and it didn't seem to make a difference. I also don't think it works on RT GPU.

[NicoC]

Could you send me or vlado a scene on the chaosgroup email (or link to)?
Many a thing could be contributing to it, hard to tell without the scene.

Hello,

Preparing an archive right now and I'll send it by PM, but please don't spread it - it should stay between you and Chaos guys, otherwise our client will kill me (the project has already opened for real, but still). Thanks in advance for any help !

Cheers.

[M.Max]

Thanks for sharing this !
For me 80% of the time the result was great and quick as a little noise doesn't harm, but some of my clients hate noise and like to see the image super cleaaaaan and here where I need to keep testing ...
I have read that using low burn value when using Reinhard color mapping could affect the sampling process badly ! could you please shed more light on that if it is correct !

Regards,

[^Lele^]

Thanks for sharing this !
For me 80% of the time the result was great and quick as a little noise doesn't harm, but some of my clients hate noise and like to see the image super cleaaaaan and here where I need to keep testing ...

Simply raise the max AA (to, say 100) and lower the noise threshold.

I have read that using low burn value when using Reinhard color mapping could affect the sampling process badly ! could you please shed more light on that if it is correct !

Only if baked into the render, and with sub-pixel mapping on.
In fact, the sampler doesn't misbehave, it just needs a lower noise threshold to be able to "see" the image well, under (much, potentially) lower variance conditions.
I encourage you (and everyone else)to experiment with the noise threshold: there is no correlation between the old and new Noise Threshold behavior, the number in the spinner means a different thing.
As such, feel free to experiment with very high, and very low values, and see how (prettily! ) V-Ray will react.

[M.Max]

Thanks for the clarification .
I think it will take me some time to accept this simpler workflow and forget about all the old tweaks that I used to do unconsciously
Regards,

[^Lele^]

Ah, man, do tell!
If ever there was an incurable tweaking syndrome, i'd have had it!
For proof of that, i was already into trying to automate render settings back in the day when i released my scripted exposimeter: it has a checkbox to "Enable Experimental 'Sample Rate Calculator'".
That's from 8 years ago.
So yes, i hear you, and everyone else, about letting go of hard-earned habits.
However, i was just hired, not brainwashed, by Chaos Group, and trust me when i say i am a true convert purely because of the numbers.
Once i let go of it, and embraced the new approach, oh boy was i not hopping with joy and amazement...

[glorybound]

Any reason to set the noise threshold to 0? The tip talks about it, however, I have never heard of anyone doing it.

[padre.ayuso]

OK, so I have set VRay on Adaptive Image Sampler, and I have set my Color Threshold as low as 0.001. I would have never done that before so forgive me for my venturing into this question. But using BF +LC I used to change the BF sampling to get cleaner images. Now, to me is a bit unreal having to change the whole of the Color Threshold just to improve the Global Illumination noise? I then got very red on the Sample Rate, so I up the Color Max Subdivisions as well.

Maybe this is not letting go of my older ways, even though I'm pretty new to VRay, but wouldn't it be worth it to have some button that helps handle the sampling of the global illumination only?

Edit: Since I have changed BF to IRM and it is faster rendering time and way cleaner, of course. GI noiseness cleaned out nicely since IR is much cleaner with lesser settings than BF by nature. However, looking at the Reflection Pass, this is noisy in some areas and clean in some other areas. Is there a way to make it all clean other than reducing the noise even lower and other than increasing the Min Subdivisions?

[Streetwise]

I noticed that the new default Image Filter is VRayLanczosFilter. Pretty sure it use to be just "Area". My renders are generally 3000x4000 interiors. Not sure what is best.

Thanks.

[glorybound]

I am having an extreme issue trying to render out things with 3.30.03. I had a large project that was taking 19+ hours, per view, and now a small project is behaving the same way. If someone can look at this, that would be great. Right now, there is no way I can make deadlines with such high render times. I had to run my current scene, though Corona, just to get it to deadline. The Corona render took less then 2 hours. It's probably something I am doing, but I can't figure it out. Also, if it matters, both scenes were started V-Ray prior to upgrading to the new 3X.

[Alan Iglesias]

...Also, if it matters, both scenes were started V-Ray prior to upgrading to the new 3X.

In my testing here it matters a great deal. When bringing in an existing project to 3.3, make absolutely sure your renderer has been reset completely. I change both Vray Advanced and RT to 3DS Max Scanline and then back to Vray again first, and then change the Sampling to Adaptive and using the defaults from there the results have been outstanding for final renderings.

You never know though, you may have something else going on there that is tripping up 3.3, but I wanted to re-iterate how important resetting the renderer is when bringing in older files to 3.3.

In fact, Vlado, why not have an option to completely reset the renderers in the dialog that shows up when bringing in a file already worked up in another Vray version?

-Alan

[Streetwise]

I am having an extreme issue trying to render out things with 3.30.03.

I noticed that when I started up 3ds max there was a dialog alerting me to the "new" methods of 3.3 and asking if I wanted to roll with that or not. I checked "yes" on it and have been reading a lot of the threads on the new usage. HOWEVER, in your case, perhaps selecting "no" and using your old settings may get you back to the quicker render times? I don't know if 3.3 works that way or not, or how you'd even get back to that original screen. Just a thought though.

[glorybound]

I have converted it awhile ago, so there is no going back. I have reset the renderer several times, so that should be good. Also, I have merged everything into a clean MAX session, but it didn't make a difference.

[Alan Iglesias]

...HOWEVER, in your case, perhaps selecting "no" and using your old settings may get you back to the quicker render times?

In my testing that would be true, so in a deadline situation that surely may help. But I'm really hoping that used correctly, the new sampling is as good for everyone here as it has been for me. But of course the only way we''ll know that is by everyone using it for different types of jobs and reporting here. Just to add to the public knowledge, I think I'll publish the results of my tests a little later...

-Alan