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

[ARTECONI-CGI]

Goodmorning everyone,

Here I am to post you the tests I have been running these morning.
The following 3 images are the result of the 3 different LWF approaches.
On all the 3 images i have 7 identical spheres.
The first 3 balls are just diffuse, the following 3 are just reflection (so diffuse set to 0/0/0) and the last one is just a texture.



Those are the settings I have used on them:
Sphere 1: Diffuse RGB 1/1/1
Sphere 2: Diffuse RGB 128/128/128
Sphere 3: Diffuse RGB 254/254/254
Sphere 4: Reflection RGB 1/1/1
Sphere 5: Reflection RGB 128/128/128
Sphere 6: Reflection RGB 254/254/254
Sphere 7: Just a texture map on the diffuse

I have attached as well the settings used for all the 3 workflow in the next post. (I had problems to upload them in a single post)


As you can see the most used linear workflow (LWF_v1) produces an image different from the other 2.
The middle gray is shifted to the bright side so it gets brighter than the average.
If you color pick the fist ball in photoshop you will notice that for the workflows 2 and 3 we get an average value of RGB 1 meanwhile for the workflow 1 we get an average RGB value of 20/20/20
If you think that from the Max diffuse color picker I have chosen a diffuse value of 1/1/1 and it result in a value of 20/20/20 in photoshop you easily realise that it becomes too difficult to choose darker tones.#

So I think that overall the most correct workflow is the 2nd one but it is as well the less practical because we have to color correct every color and texture.
The third one looks good but is inconsistent on some kind of textures as explained by Chaos group and personally tested.
So again here I am to ask if we can get the LWF button fixed to work as it should so that we don't need to bother anymore about any gamma issue.

Kind Regards,

Giacomo.

[ARTECONI-CGI]

By the way I didn't use don't affect color + sRGB button form the VFB just to have the gamma burned in the image. I am aware that the correct way is to use those 2 extra settings.

[danb4026]

Giacomo, what are all of your settings when using the LWF button? Ooops, just saw your other posts with info. sorry

[danb4026]

As artists, these are not things we should have to concern ourselves with. Gamma should just work. I am very happy about the inclusion in the new service pack of the ability to adjust burn and brightness. This is the sole reason that I stray away true LWF sometimes. When can we expect the service pack to be out?

[vlado]

When can we expect the service pack to be out?

Any day now...

Best regards,
Vlado

[trixian]

Your first option is correct, assuming you have max's gamma set to 2,2 and let it save out bitmaps as gamma 1 (this option is default, but slightly obscured in 2014). Also saving to exr as float is a must for proper results down the line.
Afaik, you are not meant to use any other types of colour mapping when rendering linearly (mathematically correct/sound), so forget about Reinhard. You can "colour-map" your stuff in comp to achieve whatever look you want.

Edit: Hmm for some reason I replied to this before reading the rest of the thread. Didn't mean to repeat what others have said. I also had no idea that Chaos were including Reinhard colourmapping into this workflow.

[ARTECONI-CGI]

Hi Trixian,

Thanks for the reply.
I agree that the Lwf must be saved in exr wiithout burning the gamma in.
I see as well that you, like most of the people I know, use the 1st LWF I have posted.
It's curious to see that no one has replied directly to my doubts and concerns about the color input.
I'd honestly love to get a straight answer from Chaosgroup or anyone that could offer a "reasonable" reason.

Why a black 1,1,1 becomes a black 20,20,20 using this workflow?

Why in the following LWFs a black 1,1,1, remains a black 1,1,1? this makes much more sense to me and I'd like to understand why you all say that this is wrong.

Kind Regards,

Giacomo.

[Svetlozar Draganov]

Hello ARTECONI-CGI

First I would like to point out that there is something wrong in your test - or something is wrong with the screen-shots of the settings or screen-shots of the results.
Or the test I made is somehow different than yours even though I used the same approach.

Following screen-shots that you have posted about the Gamma settings I got the following results

LWF-1. input 1,1,1 --> result 20,20,20
LWF-2. input 1,1,1 --> result 20,20,20
LWF-3. input 1,1,1 --> result 1,1,1


In the first two scenarios the output 20,20,20 is different than input 1,1,1 because 3DSMax doesn't apply inverse gamma on the colors in the swatches. In other words you are not working with Linear colors but with sRGB colors which are always lighter than Linear ones - that's why the result is brighter. If you want to have the same result you have to add color-correction texture map - add the same 1,1,1 color in the Basic Patameters (note that they are from 0 to 1 not from 0 to 255 so 1=0.00392) and then in Lightness Rollout turn on Advanced checkbox and add RGB = 0.454 in Gamma/Contrast. This will convert sRGB color to linear color and the output should be 1,1,1

In the LWF-3 workflow - the result is 1,1,1 because with Gamma 2.2 and Linear Workflow checkbox ON in Vray Color Mapping Rollout - Vray does the same thing explained above. The input color 1,1,1 is automatically gamma-corrected by Vray and that's why the output is 1,1,1.

[ARTECONI-CGI]

HI Svetlozar,

Thanks for the reply.

Sorry for not explaining, I thought I did it.
As you can see in my 2nd test I have used a map on the diffuse and reflections.
It is actually a color correct node that loads the color with a .4545 gamma as you suggested on the post.
That's why I get a result of 1,1,1.

I think this is the 2nd most suggested workflow: leave gamma and lut panel off - correct manually all colors and textures and use 2.2 on vray color mapping.

So, as far as I understand the correct way is to use the 1st LWF and color correct manually every single color so that my value matches. This means that the suggested workflow is still unpractical because fixing every single color is a problem for me and most of the people that works in large studios.

What do you think?

LWF 1 is the workflow suggested by Vlado, so to get linear colors I have to manually color correct them. Not user friendly.
LWF 2 is the same but will have to correct the textures as well manually.
LWF 3 corrects colors and textures automatically (I love it) but then it is buggy and unreliable.

Can We have a definitive LWF that doesn't involove correcting anything and works straightforward?

Kind Regards,

Giacomo.

[ARTECONI-CGI]

I may suggest that the solution is having a floating color picker instead of an RGB value.

This might resolve the problem of not getting the colors corrected from the Gamma & Lut Panel.

Giacomo.

[Svetlozar Draganov]

Ups - I have missed that in the LWF-2 you have used image instead of color , so this is the reason for the difference between the tests.
The most proper LWF is as Vlado suggested LWF-1 - yes it is not user friendly but it is guaranteed that everything is set correctly.

Can We have a definitive LWF that doesn't involove correcting anything and works straightforward?

We are working on this , and I hope in the near future such option will be available.

[ARTECONI-CGI]

Hi Svetlozar,

Thanks a lot for confirming my doubts.

Can you confirm that this is happening because of Vray using an 8bit color picker instead of a 32bit one?
Is it something difficult to fix?

As well I would love if you can answer my questions below point by point, you'll make me an happy man and I am sure you'll make the whole vray community happy:

1) grayscale jpeg loaded as bump, reflection, refraction.... etc etc: do I have to load them in gamma 1 or by default in 2.2?

2) a grey RGB value used as a reflection or refraction: does it have to be color corrected with a .4545 node or not?

3) normal map and displacement: they have to be loaded in gamma 1? isn't it?

4) what about the vray fog maps and colors, fastss2? what has to be corrected and what doesn't?

5) what else should be corrected or not corrected with a .4545 node? vray sun & sky color? ies temperature? the colors of the temperatures looks wrong compared the the ones available in the photometric lights. Why?

6) Is there anything else that we should be aware to correct?

There's no official, complete full guide to setup the LWF in vray and the documentation is poor.

7) Can you produce something to make our life easier?

It has never been 100% clear and doesn't exist an exact documentation that expains what is right at 1 and what at 2,2.

Can you provide a detailed list that explain the detailed exact correct way to implement LWF in Vray, INCLUDING a list of all the values and maps that has to be corrected manually?

I'll be really grateful if you can go through all those points.

I am investing my career in your software and have gone through all the threads about this topic.
It is definitely confusing because a lot of people claims correct something that is not.
Having it published by the same software house will become imperative and remove any doubts to the user base.
Clarifying and publishing a detailed documentation about the LWF will only benefit Chaosgroup.

Kind Regards,

Giacomo.

[vlado]

I may suggest that the solution is having a floating color picker instead of an RGB value. This might resolve the problem of not getting the colors corrected from the Gamma & Lut Panel.

I don't think so... the color picker has nothing to do with this at all.

Best regards,
Vlado

[ARTECONI-CGI]

So what is the problem then?

I keep on asking precise question and getting generic answers.

I honestly think there's really a big problem behind the LWF implementation in VRAY that can't be solved easily.
You wouldn't be so airy in the anwers otherwise.

In the other rendering engine is a problem fixed years ago and this is still going on since 4 years...

Giacomo.

[vlado]

In the other rendering engine is a problem fixed years ago and this is still going on since 4 years...

Yes? Give me a 'for instance'. For these render engines, do you know if your reflection maps are being gamma corrected or not even if they don't need to be?

I keep on asking precise question and getting generic answers.

Because I really don't see what all the fuss is about. You are asking the wrong questions - it doesn't matter where a texture or color is used in the scene; you only need to correct textures (or colors) that you know have gamma burned into them. How do you know if a texture or a color has had gamma applied to it? It depends on where it comes from (and this is something that only you know; the renderer has no way to guess it - and don't tell me other renderers have solved this because they haven't). The rule is that most 8-bit textures coming from photographs or from PhotoShop do have gamma burned into them and should therefore be gamma corrected. There are exceptions however - sometimes bump maps and displacement maps are painted by hand in e.g. PhotoShop and in that case they should not be corrected. Normal and displacement textures coming out of sculpting programs like Mudbox typically don't need to be corrected either.

Nothing else needs to be corrected.

When you consider the above, the answers to your questions are trivial.

1) grayscale jpeg loaded as bump, reflection, refraction.... etc etc: do I have to load them in gamma 1 or by default in 2.2?

This depends entirely on how the map was produced. In some cases it will have to be gamma corrected (f.e. if the map is a JPG file that was obtained from a photo), in other cases it doesn't need to be (f.e. if an artist painted it in PhotoShop assuming that RGB 128,128,128 is medium grey). The general rule is that 8-bit color textures (JPG, PNG etc) typically need to be corrected when used as colors. For grayscale 8-bit textures, it is up to you to know how to use them.

2) a grey RGB value used as a reflection or refraction: does it have to be color corrected with a .4545 node or not?

If you picked the value from a JPG image for example, you will have to gamma correct it. If you just entered it by hand (f.e. RGB 128, 128, 128 because you wanted half of the light to be reflected), then obviously there is no need to correct it.

3) normal map and displacement: they have to be loaded in gamma 1? isn't it?

Typically these should not be gamma corrected. (The only exception would be displacement maps if they were derived from a photo somehow, but I don't think this is typical.)

4) what about the vray fog maps and colors, fastss2? what has to be corrected and what doesn't?

Most often than not, these textures come from photos or are painted in PhotoShop and as such need to be gamma-corrected. Same goes for colors - if you picked the color from an 8-bit texture in PhotoShop, it needs to be gamma corrected. Otherwise - not.

5) what else should be corrected or not corrected with a .4545 node? vray sun & sky color? ies temperature?

None of these should be corrected. They are computed in the physical space of the renderer - they are not input values from textures, so there is no need to touch them.

the colors of the temperatures looks wrong compared the the ones available in the photometric lights.

V-Ray uses a slightly different color space than the standard photometric lights. It would be just as correct to say that the standard photometric lights are wrong for V-Ray.

6) Is there anything else that we should be aware to correct?

Anything that has had a gamma applied to it before it gets to the renderer, should be corrected. It is your responsibility to know which are those textures and colors. The rule here is that 8-bit textures typically need to be corrected on input, as well as any colors that you "picked" from an 8-bit texture.

There is no need to get obsessed with this and correct everything everywhere - as I pointed above, the basic rule is very very simple.

Best regards,
Vlado