How to get better VRay renders (aka working in linear space)

[timmatron]

Actually I do... I mention it in Chapter 6 when talking about texture baking. Applying the gamma in Max, allows you to see the material that are in gamma corrected space, as well as render the image in Max in correct space.

Ahh, sorry I haven't gotten that far yet. I had assumed you were pretty much done with that topic, but one must learn never to assume! I must watch on. One of these days I will master this linear space, gamma correction thingy.

[Gijs]

The advantage using gamma correction over the LUT-curve is that Vray adjusts its sampling to it (importance sampling). The effect is that you don't need to increase your subdiv. too much to get the shadows noise-free.

Not really convinced that this is true. If you look at the rawRGB, and rawGI, it would seem that Vray still does everything in linear space, (as it should) and then applies the color mapping at the end. Either way, I think that adjusting "Clr thresh." in the Irrdiance mapping would be a better way to push more samples in the darks.

I also think that pipelines and methodology is something that will be debated for a long time.

I am sure it is also a matter of habbits and preferences, just like you mention. I tested my statement before posting it: gamma corrected renderings take longer to render and contain less noise at the same settings, which also means that the noise level is optimized for gamma corrected viewing. (which in the very end is everyones goal)
My hard guess is that this also means that you need less overall samples to get a rendering noise free.

[flipside]

Ok I was bugging gijs with this trough email, thought I would post my question here too.

All this linear workflow is getting clear for me, except this. Reflections look totally different than what I was used to.

here is a render made with the normal workflow:



And here the same with all colors and maps corrected for linear workflow:


Using weaker reflections doesn't help, it's like the fresnell effect is completely different.

What am I doing wrong?

[Sawyer]

Seems like it's more than just the falloff as the red on the surface is gone in the second. Like the bright is getting blown out.

[flipside]

It's blown out because reflection is stronger (especially in the middle where normally reflection should be less than on the sides).

[Sawyer]

If thats the case thought shouldn't the red be brighter too on the white base?

[flipside]

hu? The red is brighter imo. I adjusted the red diffuse with color correct map to compensate for the gamma. So diffuse is correct, if I render without reflections all is ok.

Now it's just reflecting much more and I wonder why that is.

[jujubee]

Flipside - perhaps the way we've all been viewing fresnel falloff according to your sample has been incorrect all along. Just a thought.
__________________________________________________ ___
Separate note and summary from rereading this thread for the Nth time. Please correct me if I missed something:

1) We should all do is correct our monitor's colors so that they are an 'industry' average (CIE standard.) This should apply across the board for even if you don't create graphics. This is a standardization to ensure that what one person looks at image-wise is what another person sees. Unfortunately, every monitor differs to a degree and this is a necessity. This extends beyond monitors, to television and even print.

2) *Optional* Adjust Max Gamma preferences which effects viewport and mat preview only. This has no effect on final rendering but gives you a better represenatation of what a color corrected mats will look like in final renderings. To visually match a specific RGB value, the colorcorrect plugin can be used.

3) There are two methods to obtain gamma corrected renderings. Essentially, Max/Vray by default puts a linear (1.0) 'curve' on a rendered output and this often results in underexposed renderings with other unexpected results. These methods attempts to take the guesswork out of our work:

a) Throb's method of rendering to VFB and applying a .64 curve to offset the default to balance out a 'correct' view.

b) GIJ's method of using 'Vray:Color Mapping' rollout type 'Gamma Correction' with values set to 'Dark Multiplier: 1.0', and 'Bright Multiplier: 0.45454".

Throbs method is sometimes faster because it is done post-rendering. Gijs method is slower because the renderings are generally brighter (bright renders generally take longer), but results in less noise - on the other hand you could reduce samples to speed Gij's method up.

[flipside]

I think your summary is correct. There is even a point c, in new vray version, there is a button in the VFB "display colors in srgb space".

Maybe d: use max vfb which is automatically adjusted with your gamma setting in preferences.

The fresnell reflection thing still bugs me though. What you say could indeed be the case, maybe this is the correct fresnell look.

[jujubee]

Is a pretty good way to calibrate your monitor, if you don't have a probe or anything. WAY better then the adobe gamma thing that realy screws things up.

That wouldn't allow me to install - giving an error saying that it was unable to write to the LUT.

Adobe Gamma seems to work fine overall - just make sure you use the option to use the RGB sligers versus it's grayscale one. And stand far far away from the boxes as you adjus the sliders back and forth...

[timmatron]

b) GIJ's method of using 'Vray:Color Mapping' rollout type 'Gamma Correction' with values set to 'Dark Multiplier: 1.0', and 'Bright Multiplier: 0.45454".

I'm still a little confused about that one, because on Chris's interior DVD he is setting both the bright and dark multipliers at .4545. I tried rendering with that setting and it came out looking correct so I thought that was the way to do it.

[jujubee]

Earlier in this thread, Chris said he was wrong about what he did on the DVD. I was going to experiment with the dark as well because the darks/shadows are too light in my scenes with 1.0 dark and .45454 bright.

Logically, I don't see why one couldn't adjust the darks... Then again I'm not a numbers person.

[jujubee]

Well, for my particular scene setting both the dark and bright to .45454 resulted in a better render with better shadow definition...

[Gijs]

Well, for my particular scene setting both the dark and bright to .45454 resulted in a better render with better shadow definition...

then your lights were just too bright, in gamma correct colormapping, the dark acts as a (global) light multiplier.

[Gijs]

Is a pretty good way to calibrate your monitor, if you don't have a probe or anything. WAY better then the adobe gamma thing that realy screws things up.

That wouldn't allow me to install - giving an error saying that it was unable to write to the LUT.

Adobe Gamma seems to work fine overall - just make sure you use the option to use the RGB sligers versus it's grayscale one. And stand far far away from the boxes as you adjus the sliders back and forth...

Using adobe gamma alone is said to give incorrect results, you can read more about it on AIM-dtp website. The method described at the bottom of my article also uses adobe gamma, but in combination with a gamma card that you look at when adjusting the sliders. Also keep in mind that for LCD screens you should better use hardware calibration, or otherwise keep your head in locked position during calibration, because gamma changes with viewing angle on LCD's