Cheatsheet: How to get a Cleaner image with SP3

[^Lele^]

I assume you're here because you want to get rid of noise, and are prepared to render for a bit longer to achieve that.

You can safely concentrate on two parameters only:

• Max AA Subdivs
• Noise Threshold/Color Threshold*

You will also need a SampleRate render element to decide how to proceed (it adds nothing to rendertime.).

There are just two simple rules:

1. If the SampleRate shows blue in the noisy areas, lower noise threshold
2. If the SampleRate shows red in the noisy areas, raise max AA subdivs

That's it: you can now kick back and relax, and V-Ray will do the hard work.

Attached it's a simple CheatSheet which illustrates the process.



Cheatsheet at full size (I've reuploaded and re-set the permissions for this. Sorry for any issue you may have encountered.)

* there is an inconsistency in naming the same parameter for the adaptive and progressive AA modes.
I always refer to the control shown in these two screenshots, not the global DMC one, which as of sp3 should be left to its defaults:

[short cirkuit]

Thanks - im assuming to do such changes using the default settings? Assuming it also applied to IM+LC setups?

[^Lele^]

Yes, default settings, yes, it works with the IRMap.

[king_max]

What if you're working on a complex production scene that takes 30 minutes just to render a preview image? Figuring out which settings works best takes a lot of time. Sometimes a whole day. If you have the computing power, you cold have spent that time rendering the image. If there'e a quick setup option and you can afford to disregard the render times which I suppose most of the archviz users would choose, than what's the point in meddling with any combination of settings at all?

[^Lele^]

I assume you're here because you want to get rid of noise, and are prepared to render for a bit longer to achieve that.

I did not say one needs to tweak.
I said, this is how to, should you need to.

Figuring out which settings works best takes a lot of time. Sometimes a whole day.

This is what the new approach removes: the need for endless tweaking.
You likely mistook my approaching the problem iteratively in the cheatsheet as a need, whereas i did it on purpose to also show a samplerate moving towards the optimal and illustrate the logic behind the approach.
Set a noise threshold, set a max AA, render.
In fact, don't, and render: it comes ready for most of the scenes thrown at it.

If you have the computing power, you cold have spent that time rendering the image.If there'e a quick setup option and you can afford to disregard the render times which I suppose most of the archviz users would choose, than what's the point in meddling with any combination of settings at all?

I don't quite understand what it is you're asking here.
Could you be more clear?

[king_max]

After all these years I know how to setup a scene. My point is that in my case the sp3 didn't come "ready" for a scene. I couldn't even get a clean image if I let v-ray internally handle the subdivisions so I spent the whole day tweaking the parameters. If I had a quick setup mode where I could choose "production settings" no matter how oversampled some elements could be and how much would I waste in terms of render times, in the end I would get a clean image in 5 hours instead of 3 hours but wouldn't spend 2 hours adjusting the settings. Do you get my point?

p.s. I am one of those people who likes optimizing things and meddling with various options, but sp3 was somehow advertised with the "one click render" option, which in my case it didn't turn out to be. I'm not here to troll about it. I'm very satisfied with vray and it's my renderer of choice for more than 10 years now. I'm really excited with all the new options and with the expanding possibilities it offers through every new release.

[^Lele^]

After all these years I know how to setup a scene.My point is that in my case the sp3 didn't come "ready" for a scene. I couldn't even get a clean image if I let v-ray internally handle the subdivisions so I spent the whole day tweaking the parameters.

So, let me get this straight: you reset the render settings to the SP3 defaults, touched absolutely nothing, rendered and got a dirty image?
If so, i'd love to see the scene (i believe it's on its way anyways).
If not, however, then we aren't talking of the same thing anymore: i can't fathom what you changed, and what impact your approach would have had.
The whole point is to avoid tweaking, as that is the prime source of mishaps.

If I had a quick setup mode where I could choose "production settings" no matter how oversampled some elements could be and how much would I waste in terms of render times, in the end I would get a clean image in 5 hours instead of 3 hours but wouldn't spend 2 hours adjusting the settings. Do you get my point?

I hope i do.
Back to the defaults, set max AA to 100, noise threshold to 0.001, render:do you get noise?
Now set the noise threshold to 1.0 and render: isn't it quick?.

p.s. I am one of those people who likes optimizing things and meddling with various options, but sp3 was somehow advertised with the "one click render" option, which in my case it didn't turn out to be. I'm not here to troll about it. I'm very satisfied with vray and it's my renderer of choice for more than 10 years now. I'm really excited with all the new options and with the expanding possibilities it offers through every new release.

Yet, I feel there is some resistance in letting go of putting one's finger in all the possible pies.
I hear you, you know.
It's been my very job for a good few years, and i suppose the amount of scripts and tools i produced and released to go and mass tweak scenes for optimal (or as near as) perfomance ought to lend me at least some credibility, when i say that i am a great fan of the new approach, simply because i saw the numbers.
For it to behave as advertised, you'll have to give it a chance to work as it should, refrain from the tweaking (refrain from the tweaking! sorry. i know old habits die hard ), at least for the first few tests, and once you get comfortable with the sampler, you'll be in a better position to judge for yourself if you'll ever need to spend time optimising scenes in the old way.

Try the two options above on your shot, see what they do.
If you have baked in color mapping with loads of (soft) clamping (ie. Rehinard with burn @ 0), you may need to go even lower with the noise threshold, but then the cue would be that the samplerate is very blue, and the render is very quick and very noisy.
But that would be all: noise threshold, once again.

[Jvincentfx45]

One question regarding this just so when adjusting the quality we should focus on two settings the AA and the clr threshold correct.

Just wondering cause you had said noise but generally I have stopped touching anything under the dmc out side of the new sampler to have more control but I am ok leaving that alone. Old habits do die hard.

Sorry I just want to make sure I am adjusting the correct things which would be the clr threshold and the max AA and touch nothing else.


Thanks so much j

[^Lele^]

Indeed, my bad: here's a screenshot of the controls to tweak, for both progressive and adaptive (i suppose the wording could use some love):





You are correct: anything one needs to tweak is in here.
Whenever i refer to "Noise Threshold" i mean the one in the AA, not the global DMC, rollout.
So it becomes "Color Threshold" for the Adaptive sampler.

Thanks for pointing it out!

[glorybound]

When I reset to SP3 GI is not enabled and the progressive is active at 1 minute. So, when I hit render without doing anything, I don't get much. I didn't take the "1 Click" literally, so I start changing things right off the bat. I'm I doing something wrong? Also, on SP3, I noticed that the RAM is still default of 4000, shouldn't be 0?.

[king_max]

So, let me get this straight: you RESET the render settings to the SP3 defaults, touched absolutely nothing, rendered and got a dirty image?
If so, i'd love to see the scene (i believe it's on its way anyways).
If not, however, then we aren't talking of the same thing anymore: i can't fathom what you changed, and what impact your approach would have had.
The whole point is to avoid tweaking, as that is the prime source of mishaps.


I hope i do.
Back to the defaults, set max AA to 100, noise threshold to 0.001, render:do you get noise?
Now set the noise threshold to 1.0 and render: isn't it quick?.


Yet, I feel there is some resistance in letting go of putting one's finger in all the possible pies.
I hear you, you know.
It's been my very job for a good few years, and i suppose the amount of scripts and tools i produced and released to go and mass tweak scenes for optimal (or as near as) perfomance ought to lend me at least some credibility, when i say that i am a great fan of the new approach, simply because i saw the numbers.
For it to behave as advertised, you'll have to give it a chance to work as it should, refrain from the tweaking (refrain from the tweaking! sorry. i know old habits die hard ), at least for the first few tests, and once you get comfortable with the sampler, you'll be in a better position to judge for yourself if you'll ever need to spend time optimising scenes in the old way.

Try the two options above on your shot, see what they do.
If you have baked in color mapping with loads of (soft) clamping (ie. Rehinard with burn @ 0), you may need to go even lower with the noise threshold, but then the cue would be that the samplerate is very blue, and the render is very quick and very noisy.
But that would be all: noise threshold, once again.

I always bake the gamma/color mapping to start with a lighter scene that always helped the sampler in obtaining faster/cleaner images. Did something changed in SP3 regarding that?

[Jvincentfx45]

Thanks ,yeah makes sense to have some renaming since it is the noise threshold for the progressive sampler maybe at some point it would make some sense to just make it consistent.

[^Lele^]

I always bake the gamma/color mapping to start with a lighter scene that always helped the sampler in obtaining faster/cleaner images. Did something changed in SP3 regarding that?

Well, yes.
That is the job of the sampler, as you well pointed out, so everything changed for SP3, precisely where you mention.

Hence the cries for trying with default settings and strictly linear workflow: things all but impossible with SP2 (because of fireflies, poor sampling of dark areas, excessive sampling of highlights, and so on) are a piece of cake for SP3.
To stress that i sound like a broken record (the risk of sounding like one is well past me, by now.), forget the past, give the new a chance.

I'm linking you the seminal posts about it, so you can jog your memory, see samples of the behavior, and read what Vlado explained about it.

http://forums.chaosgroup.com/showthr...SP3-sneak-peek

http://forums.chaosgroup.com/showthr...e-for-download

[jacksc02]

Lele did you leave the Global DMC noise threshold as the new defaults?

[^Lele^]

Yes, as Vlado stated elsewhere (i know the threads are making a tapestry, here...), that's now kind of obsolete too.
Here's a link to his post:
http://forums.chaosgroup.com/showthr...501#post674501