Disable Max Gamma Correction and use VRay's LWF button?

[vlado]

If you adjust colors visually in 3ds Max, you don't have to do anything specific.

However if you need to use a RGB value that you picked from e.g. PhotoShop or directly from a JPG file viewed in another application, then you will need to pass it through e.g. a ColorCorrection texture to remove the gamma.

Best regards,
Vlado

[ARTECONI-CGI]

Hi Vlado,

Thanks a lot for this last clarification.

Kind Regards,

Giacomo.

[Crayox13]

You guys mentioned something interesting here. I noticed that when I use batch render (vray buffer off) sometimes my exr files are 8 bit, even though I selected 32 bit. I mean, when I try to reduce exposure my whites go to gray and not the correct colour as usual. What affects that?
Though I have everything setup as mentioned here other than don't affect colours is always off on my scenes.

[BVA]

I am really confused with gamma, to be honest. I had a workflow that worked in 2013, but now it is gone with 2014.
Is there a way with the current vray and max 2014 that will work till the new version will be released?

I used settings as follow:
Max gamma 2.2
Bitmap input 2.2
Bitmap output 1

Then in Vray I set gamma to 2.2

If I try this settings now it washes my images completely, almost white. I suppose that means I added gamma twice?
If I render max's render window it looks white but save the rendered image correct. If I render with frame buffer it looks right in the viewport but save the rendered image washed out.

[vlado]

The correct settings for 3ds Max 2014 are as follows (and the defaults in the 2014 builds of V-Ray will be very similar):

Max gamma 2.2
Bitmap input 2.2
Bitmap output 2.2

V-Ray VFB sRGB button ON
V-Ray color mapping Gamma 2.2
V-Ray color mapping "Don't affect colors" ON
V-Ray Linear workflow OFF

Best regards,
Vlado

[Crayox13]

I'm still on 2009. Should I do anything differently? Is bitmap output (which is always 1.0) the output in exr too?
If I use Don't affect colours I can't use Reinhard values either. What should I change if I have that off?

[vlado]

I'm still on 2009. Should I do anything differently?

Yes, there are a few differences.

Is bitmap output (which is always 1.0) the output in exr too?

For OpenEXR output, you will need to keep the Output gamma to 1.0. 3ds Max 2014 is smart enough to know that OpenEXR files should always be saved with gamma 1.0, but for 3ds Max 2009 you will need to specify that explicitly.

If I use Don't affect colours I can't use Reinhard values either. What should I change if I have that off?

This will be possible with the next service pack, where the "Don't affect colors" option is changed so that it only skips the gamma, but can apply the other color corrections.

Best regards,
Vlado

[BVA]

Thanks Vlado, will try with my next scene

[Crayox13]

So what are my differences? Especially if I have Don't affect colours turned off.
Everything else is like it was said here.

[rleas]

Vlado, does your suggested workflow hold for RTT baking as well?

[grantwarwick]

3ds Max 2014 is barely smart enough to know that OpenEXR files should always be saved with gamma 1.0

Corrected.
Glad they got something right though.

[vlado]

Vlado, does your suggested workflow hold for RTT baking as well?

I imagine it would be the same for RTT as well.

Best regards,
Vlado

[LudvikKoutny]

I just recently found out that i got Vray LWF all wrong. I was using gamma 1.0 and sRGB framebuffer button, but DMC sampler was freaking out because pixel value variance is a lot different at Gamma 1.0, so it was hard to get noise clean in darker areas without ridiculous rendertimes

Now, that i use correct LWF i found out in page 1 of this thread, almost everything finally works as i'd expect... almost, except for color mapping, such as reinhard. (That was pointed out in this thread too)

My question is... could we get at least some rough ETA on when the next SP with color mapping working with "don't affect colors" option is released?

Thank you in advance

[ARTECONI-CGI]

Hi All,

Thanks for digging deep into this topic that is so confusing and so important at the same time.

I accept the theory that the inputs needs to be linearized (textures and rgb values) to have a correct render in linear space.
I understand as well that the file itself (32 bit exr/hdr/ etc etc) is stored in linear and is only DISPLAYED with an sRGB curve on top whitin the vray framebuffer and later in a compositing package as nuke, fusion, ae using a lookup table.
Again, the reason for such a workflow is that we don't want to break the math rules behind the image and have all working properly in post.
I am ready as well to take for granted that we get a flat image that needs to be color corrected to give its best and stick with the rule of cgi: what looks good is good!
What I don't get is why my images looks much more photorealistic using the lwf button than the workflow suggested by danb4026 and confirmed by Vlado.

I am very well used to danb4026 workflow. It is the workflow I use since 4/5 years and always thought that there is something wrong with the math behind it.
I have largely used it and experimented with it and I always end up with an image too flat. Too much.
It is pretty much impossible to control a reflection, all the important values are concentrated at the early numbers.
Example: if you need to make a black material slightly reflective you'll give it a diffuse value of 1 or 2 out of 255.
After that it starts to become a dark grey. As well the reflection value has to be selected in the range of the first 10 values, after that you get something much more close to a mirror material than a black reflective mat.
What about the remaining 240 values?
This as well happens with the procedural maps, fallofs gradients. the fog color for example is another value that is hard to control.
And why should I choose a a white 180 to for a white wall? 180 seems grey to me, not white?

In my opinion the above workflow corrects only the bitmap input. It takes the texture and make it darker to display correctly under an sRGB LUT.
What vray doesn't do is counteracting the colour values. The colour remains the same, just get brightened up with the above workflow.
That's why most of the values are lifted on the bright side.
Try to create a gradient in photoshop, apply an SRGB curve on top and you'll see that most of the dark values are gone.
The gradient is much more versed towards the bright side. someone will probably reply that this is because psp works in srgb space so it's color picker is not linear.
The fact is that with the above workflow, if we want to reproduce a color, we have to pass it under a vraycolor node or just create a texture of it in psp and load it as a jpg.
I agree that photoshop is srgb so its color picker is already color corrected. For that reason I think that our rgb values has to be darkened to get a properly displayed in vray but this doesn't happen.

The reason why that doesn't happen in my opinion is because the Gamma & Lut panel in Max is built around the embedded rendering engine: mental ray.
there is adifference between mray and vray color values: vray select a color as an RGB value, mental ray select a color in floating point.

That makes me think that Max is clever enough to correct a floating point value but stupid enough not to correct an RGB value.
When you apply the gamma correction in max the material editor gets brighter.
Are we sure that this happens just because the light falloff is changing? to me it looks like that in vray the colors get brightened up. Is this correct?

And then what is the story with the greyscale maps? gamma 1 or 2.2?
and what happens to the procedural textures are they properly interpreted or they get broken from a wrong curve?

Again, loading textures and maps with a correction node is a little unpractical.
I usually work on huge scenes and the idea to keep the workflow consistent in a studio where we have often 10 freelancers is impossible.
Someone will break it so we need a magic button.

The magic LWF button as everyone always repeat is broken. It is wrong but it does the job. I know it doesn't correct properly the fastsss2 maps, the vraylightmaterial and something else. But overall it looks good. Much more believable lighting distribution, much faster to implement, much more photorealistic. There are few maya tutorial out there that uses it. The same Rob Nederhost, Chaosgroup in the Maya helicopter tutorial (can't remember the name of the author), Peter Guthrie on few occasions, and a great Max based company in Shoreditch I had the pleasure to freelance for in past. What I'm saying it is that a lot of top guns implements it. I agree it doesn't correct everything and it works on the entire shader not on the single maps, but the render looks so much more correct and better that makes me think that this is correct. If it's not please fix it and give us a definitive working magic button.

Regards,

Giacomo.

[voltron7]

good thread, thanks all for discussing this stuff...those settings could be welcomed on the spot3d help pages, no?