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

[CCS]

Ok, I've read every thread on this topic 437 times or so, and I think it's sinking in. I still have some questions though.

#1. GIJ's tut specified calibrating monitor per the method on www.aim-dtp.net. I maxed the contrast and set the black point as described. (Then viewing the gamma evaluation swatches, the 2.40 sample looked smoothest to me.) But, per GIJ's tut, I then display the 2.2 curves and adjust sliders in Adobe Gamma until it looks correct. Am I on track so far? (Edite: #1b - Is Adobe Gamma a better choice than the gamma settings in my nVidia?)

#2. There has also been mention of working in PS in linear space, or applying a linear profile. I know a bit about profiles, but have never seen a "linear profile" or "linear space" mentioned. Do you simply mean to work with no color profile applied? Or do you mean to use the "AIM RGB Pro = WideGamut D65 G1.00" profile available on the AIM-DTP site?

#3. How should PS be setup? Should our color space be set per this?

#4. When opening unedited images (renders saved with gamma=1.0) in PS, do we then assign a profile or convert to a profile??

#5. Do we then do any necessary edits in PS (still in linear space I assume???) and only convert to a non-linear profile when we are preparing a deliverable? (either a gamma 2.2 curve or printer's profile or whatever is required)

#6. How does animation fit into this whole mess? Do I need to apply similar gamma correction or other adjustments to the animation frames?

Sheesh. The more I think this "linear workflow" makes sense, the more confused I get!

[flipside]

I could answer some of your points, but I will leave it to Gijs or throb or someone else who fully understands it otherwise I probably confuse you even more.

All your questions are good and they will take away the last confusion away if answered well

edit: I have another question:
What if you use a dual screen setup with two different monitors? Can you use adobe gamma then? I'm having major problems with it, I can't choose which screen the adobe gamma program should affect. I know it's not ideal, i'm even using a crt and an lcd on one pc...

[jujubee]

Those are good questions. I am very curious about setting up animations (#6) for this as well as I wouldn't want to render out a sequence of large HDRs.

[timmatron]

I wouldn't want to render out a sequence of large HDRs.

Actually, in my tests so far, HDR's are smaller files sizes than tifs and targas. I just don't know why, I always thought HDR's stored more information than standard image formats.

[CCS]

Need to add a #7.

When evaluating lighting, materials, etc. in test renders, do I base my decisions on what I see in the VFB?

ie - Do I have to load an image into Combustion and apply a curve just to see if I'm headed in the right direction?

I'm a little thick. GIJ's tut was a good explanation, but I need a more comprehensive step-by-step.

This is driving me nuts...

[jujubee]

When evaluating lighting, materials, etc. in test renders, do I base my decisions on what I see in the VFB?

From my experience, it depends on which method you use - Vray's 'Color Mapping' Gamma rollout or if you applied Throbs curve.

If the first method, then you can view the 'final' output directly in Max FB - or you can save out as EXR in VFB and open it in PS CS2. The image in PS will most definitely be brighter and accurate than what the VFB currently displays.

With Throbs method, the curve will show you what the final looks like in the VFB.

[Gijs]

#1. GIJ's tut specified calibrating monitor per the method on www.aim-dtp.net. I maxed the contrast and set the black point as described. (Then viewing the gamma evaluation swatches, the 2.40 sample looked smoothest to me.) But, per GIJ's tut, I then display the 2.2 curves and adjust sliders in Adobe Gamma until it looks correct. Am I on track so far? (Edite: #1b - Is Adobe Gamma a better choice than the gamma settings in my nVidia?)

Right, you followed the steps correctly. I can't judge how your nVidia is handling things, but using this procedure in combination with Adobe gamma works.

#2. There has also been mention of working in PS in linear space, or applying a linear profile. I know a bit about profiles, but have never seen a "linear profile" or "linear space" mentioned. Do you simply mean to work with no color profile applied? Or do you mean to use the "AIM RGB Pro = WideGamut D65 G1.00" profile available on the AIM-DTP site?

I would experiment with either of the three linear profiles that are provided in the zip download. I must say that I had some weird colors using the AIM RGB Pro profile, and I think that is because it stores colors that are beyond the gamut of my CRT (any CRT). If you don't use color management, it will use your systems settings to display, just the same as in your image viewing application.

#3. How should PS be setup? Should our color space be set per this?

I am no PS expert, but according to AIM DTP, it is best to work in linear space, using the linear profiles and convert only in the end to a different color space. But keep in mind to make the conversion alwyas in the highest possible bitrate, to avoid visible loss of information. I don;t know the effect of all PS settings, you I won't comment on that.

#4. When opening unedited images (renders saved with gamma=1.0) in PS, do we then assign a profile or convert to a profile??

When saving a gamma corrected image and opening it in Photoshop, a good profile to assign is the Adobe RGB gamma. When saving a linear image, open it in Photoshop and assign it a linear profile, then it will be displayed correctly as well.
If you asign no profile, PS assumes it to have the working space profile and displays it accordingly. This means that if your working space is linear and you open a 2.2 gamma image without a profile, it will look very bright.

#5. Do we then do any necessary edits in PS (still in linear space I assume???) and only convert to a non-linear profile when we are preparing a deliverable? (either a gamma 2.2 curve or printer's profile or whatever is required)

yes. I want to empasize that if no corrections are necessary, it doesn't really matter what color space you are in, but as you can see in the examples on AIM dtp, filters like unsharp masking are better done in linear space.

#6. How does animation fit into this whole mess? Do I need to apply similar gamma correction or other adjustments to the animation frames?

It doesn't matter if you are aiming for displaying on screen, print, film or video. The main thing is that editing should be done in linear space. As Rob mentioned earlier, he uses a sRGB LUT to display the linear data correctly, which is the same as assigning a linear image a linear profile in PS.

[flipside]

When evaluating lighting, materials, etc. in test renders, do I base my decisions on what I see in the VFB?

If you use the max vfb, it will be gamma corrected by your display gamma preferences setting (probably 2.2). So you what is you see is your correct render. If you save this image and open it in PS, you need to assign it a linear profile otherwise it will look too dark. In other words, the gamma correction isn't burned in here.

The vray vfb is not affected by the max display gamma preferences setting. That's why you apply throbs curve to it to see the correct image. Or you can press the srgb space button in newer versions of vray. Or you can use gamma color mapping. I use the last option because this is the easiest and the gamma is already burned into the image so you don't need to open it with profiles etc...


Since I never noticed any problems by not working in linear space in photoshop, and I find it all a hassle to do so, this is my simplified workflow to use the linear space stuff in max:
(edit: looking at the examples at aimdtp, editing in linear space seems very important, but maybe I'll start doing that after I get used to all the other workflow changes first )

-calibrate monitor (still have to do this though, you can go on without this step but note that your images will not look 100% as they should because in next steps you'll use gamma=2.2 but you're not sure that your screen is actually 2.2)
-set max display gamma to 2.2 (materials and viewports will be corrected)
-leave the general input/output gamma's set to 1.0.
-use color correct plugin to gamma correct textures (except for hdri maps because they are already in linear space!!)
-use vray vfb instead of max's (which isn't affected by your max gamma setting)
-use gamma color mapping 1/0.455 to burn in the gamma and to display it correctly in the vray vfb.
-save the image: this is your final image, no linear space profile conversion etc is needed.

You can also use max vfb and gamma color mapping, but then you need to turn the max display gamma back to 1.0 or your image will look gamma corrected twice. (if you save it it will be good though, beacause the max display gamma setting isn't burned into the image)

[jujubee]

-leave the general input/output gamma's set to 1.0.

I'm confused here - I do the same method but leave the input at 2.2 because I thought it reads the bitmaps properly.

[flipside]

Yeah you're right, but I put all my bitmaps in a color correct map instead (see next step in my short list)

but don't use input gamma=2.2 AND color correct map with gamma 2.2!!

[jujubee]

-use color correct plugin to gamma correct textures (except for hdri maps because they are already in linear space!!)

Ah. My impression was that setting 2.2 Input already corrected for textures/bitmaps. I generally leave anything with a texture in a Vray diffuse as default (no colorcorrect applied.) So basically, are we doing the same thing?

[flipside]

Yes, you have the choice when working with 'non linear space' textures:

-set input gamma to 2.2 and don't use color correct
or
-leave input gamma to 1.0 and use color correct map to do the gamma correction (or each time you load a bitmap set the correct gamma in the bitmap load dialog)

I do it locally, because when you use textures that are created in linear space already, you should load them with gamma 1.0 of course. So to avoid confusion when working with both linear and gamma corrected textures, I prefer to use the color correct map so you can clearly see what you're doing.

Most hdri maps for example are in linear space so you should load them with gamma=1.0!!

[jujubee]

when working with both linear and gamma corrected textures, I prefer to use the color correct map so you can clearly see what you're doing.

Other than HDRs/EXRs for input, when do you use use linearly-corrected textures/bitmaps? I would think 99% of all textures are non-linear. I'd think Max gamma input at 2.2 would save ALOT of steps most of the time. I may be off about this.

[flipside]

When you completely work in linear space, and create your own textures, then you'll have more linear than gamma corrected maps.

But you're right, the input gamma option will be the fastest for many people. I just say that what I wrote down is my current workflow, I never said it is the fastest or that you should adapt it. It just keeps things clear for me personally.

Just remember that when using the input gamma=2.2, when you load a hdri or some other texture created in linear space, that you should apply a color correct map to it (or change the input gamma locally in the bitmap load dialog)! So what you do here is first you apply a gamma 2.2 to it when loading, while this shouldn't be done here, then you need to correct for that again. so you're going back and forth on these textures and that's not very logical for me, that's why I prefer to do it locally instead of globally. I will decide myself which texture is in linear space and which one is not, instead of treating them all as gamma corrected by default.

I also prefer the color correct map over the local bitmap loading option because this gives me a visual indication of what I'm doing with each map. Also, even though you should set it to 2.2 if you follow all the steps, sometimes a map just looks better if you use 2.0 or 2.5 or whatever and I'm too lazy to correct for this in photoshop so the color correct map is very handy. You can also use it for saturation control etc so even without all this linear space workflow I end up using the color correct map a lot. So for me personally, adding a color correct map to my materials is not slower since I do it all the time for all kind of purposes.

I hope this didn't confuse you again, just to keep it clear, whatever method you use to correct for the gamma corrected textures in max is up to your personal taste

[CCS]

Ok.... Starting to see thru the haze a bit, but still confused on the end of the process.

I have MAX gamma set to 2.2. I have been using 'gamma correction' in the color mapping with 1.0 and 0.4545.

When I render to VFB, the image looks pretty good. (ie - not far off from what I would deliver to a client) When I click the 'Duplicate to MAX buffer' button, the image shown is WAY lighter and washed out. If I save from either the VFB or the MAX window, I get the same result. If I override output gamma at save time, gamma=1.0 gives a good looking image (just like VFB), and gamma=2.2 gives me a washed out image (just like MAX window). Additionally, if I click the 'Show in sRGB space' button in the VFB, the image gets considerably darker.

So for starters, I need to know two things (given the above workflow, color mapping, etc.):

#1 - Which saved image is "linear"?

#2 - If I assign an override material to render a neutral grey image to evaluate lighting, do I evaluate it in the VFB or in the MAX window?