linear workflow 'reloaded' now online

Well the title says it all,

I hope you find it useful and enjoy reading it :wink:

www.gijsdezwart.nl

Maybe you should mention that the article will be found in the “Tutorials” section.

Not everybody understands dutch. :wink:

Mirko

Great article! I was a little late in reading the original Linear Space thread and didn’t fully understand it until now - my colors and bitmaps were always washed out. Now I know to adjust the input gamma and all is well. Thanks!

Ryan

very intresting and well writen…top stuff, and very usefull,

thanks

Great read Gijs, learned alot!!

Now just gotta find time to calibrate monintor..etc.. and adopt a new workflow :wink:

Thanks again

Great article!

I have one question and that is with the Bitmap loader gamma choices.

How do I know when a bitmap has to be gamma corrected?

Eg. I make a bitmap in photoshop from a photo I took, colour correcting it and assigning a color space to it. This map will have to have a gamma set at 1 in the bitmap loader, correct?

What about textures that came with Max or in a texture collection like Marlin Studios; what gamma should they have assigned in the Bitmap loader; 2.2?

Thanks Gijs

If it is created in linear space (when you have assigned it a linear profile, or when photoshops working space is set to linear and the image has no profile) then you assign it a gamma of 1 in the bitmap loader. All textures that have no color profile and look good on your calibrated monitor, should be assigned the gamma that your monitor is calibrated to.

Thank very much Gijs, I’ve spent awhile reviewing the material, as well as Rob’s thread. I think I understand it now.

Let me see if I got this right: Work and tweak with Max + Photoshop workspace in linear space, then at the very end, apply a correction curve to “bake” in the correction so that the final output (when viewed in IE for example) matches the image my linear workspace in max/ps. Is my understanding correct?

Also thanks for the pointers on the texture input. I think I’ll spend some time converting my textures to linear space in photoshop. :slight_smile:

I am not advising you to do so. My assumption is that if you have a 8bit texture it is converted to floating point during rendering. However if you convert the 8bit texture in Photoshop to a 8bit linear, a lot of information is lost. If you want to convert to linear, you should first convert the 8bit texture to 16bit, then do the conversion.
I still have to find out though if the max gamma adjust the gamma after converting to floating point or before. If it is the latter, it will be better to do the conversion in photoshop, but it means much more work to you existing textures…

Ok, thanks for the warning. Then I guess the best thing to do is to follow a case by case basis and control each map locally rather than a globally.

Thanks again Gijs for your work :wink:

nice work Gijs :slight_smile:
What do you recommend using for color mapping if you are using this linear workflow?

question…I work totaly in print and my monitor is calibrated for this so I really dont want to change anything there.

What I have done though is change the gamma in 3DSmax (prefs/gamma 2.2), now when i render the image is much brighter but also washed out. So iv applyed colour correction plug to a few mats (made them 2.2) and now there realy over the top colour wize.

Also, when I save the image and open it in photoshop its very dark.

I realize im missing somthing but what ?(it also doesnt help that I just read the other thread about it and am really confused now)

feeling a bit thick :smile:

dont feel think mdi - im having the same issue.
I have found that in photochop cs2 by adjusting the gamma to .5 you get the same result as the max/vray vfb

nope. this linear workflow would be pretty seriously negated by color mapping. it will transform the data in a way that’s not good for linear float image processing.
not good = not accurate.
however…it may look good…if it looks good then it’s done :wink:

If your colors look over the top, I want to know what color mapping you used before. If it is exponential, it could well be that you have been correcting your colors to look good in that color mapping mode.

as for the image looking too dark in photoshop, that’s because you need to assign the right color profile first. From the AIM-dtp website you can download AIM-RGB G1.0 profile. Assign this to your image. This way your image looks good in Photoshop and you can correct your image in linear space. When you are done, you can either convert a copy to your working space or to a CMYK space for printing etc.

Don’t do this! Assign it a linear profile instead!

I agree with Rob that in principle you should not use color mapping other than gamma correct to preview. On the other hand you can use color mapping, but only after you make sure that your linear data looks correct.

To give an example (posted in earlier thread as well)
If this is your linear data:

intensity color mapping can help to capture a larger dynamic range (though a bit at the cost of specularity of highlights:

In order not to confuse, I want to add that in the latter case, you have to see color mapping as a post processing thing. I imagine that if Photoshop would have some sort of color mapping to map a HDRI to a LDRI, it would be better to do this in post.

@Rob, how do you map HDRI linear data to lower dynamic range? Does Nuke have these kind of options?

any good compositing tool will have ways to control these types of things.
there are ways to remap your highs and lows and we have a little tool that puts a film curve on the data so that with hdr data the highlights look like they would thru a camera.