LUT based on print data

[Nicolas Hanson]

Is possible to build a LUT that is based on the print data only, not both print and negative?

[Lowepost]

13 minutes ago, Nicolas Hanson said:

Is possible to build a LUT that is based on the print data only, not both print and negative?

The film based LUTs that I create incorporate a negative AND a print emulation.

It is interesting but not with reality to only emulate a negative or only a print stock.

The negative imparts some color consequence and some tone scale but the main tone scale is from the print and also the dyes and other parameters limit the print color space. Also, the print softens the highlights to a graceful transition to max brightness of whites (also the color of whites).

Have a pleasant Holiday season.

Cheers Mitch Bogdanowicz

[Orash Rahnema]

3 minutes ago, Admin said:

The film based LUTs that I create incorporate a negative AND a print emulation.

It is interesting but not with reality to only emulate a negative or only a print stock.

The negative imparts some color consequence and some tone scale but the main tone scale is from the print and also the dyes and other parameters limit the print color space. Also, the print softens the highlights to a graceful transition to max brightness of whites (also the color of whites).

Have a pleasant Holiday season.

Cheers Mitch Bogdanowicz

Could you elaborate more?

How do you create luts and what's your thought process?

[Lowepost]

LUT creation is discussed here, and it might answer your questions. 

[cameronrad]

23 minutes ago, Admin said:

The film based LUTs that I create incorporate a negative AND a print emulation.

It is interesting but not with reality to only emulate a negative or only a print stock.

The negative imparts some color consequence and some tone scale but the main tone scale is from the print and also the dyes and other parameters limit the print color space. Also, the print softens the highlights to a graceful transition to max brightness of whites (also the color of whites).

Have a pleasant Holiday season.

Cheers Mitch Bogdanowicz

You rule Mr. Bogdanowicz , I constantly reread all your patents and articles on your site trying to further understand color science. It's taking a while as I'm relearning math and still don't know how to code to really to implement these techniques. Do you have any books or resources you recommend? Or any GUI friendly applications other than Lattice/3DLUTCreator that let you manipulate and apply some of these transformations.

Edited January 1, 2018 by cameronrad

[Lowepost]

On 1.1.2018 at 2:10 PM, cameronrad said:

Do you have any books or resources you recommend? Or any GUI friendly applications other than Lattice/3DLUTCreator that let you manipulate and apply some of these transformations.

Modeling of Motion picture type of 3DLuts is not well documented or taught in the literature. Many post production houses rely upon their either in-house or contracted color scientists for their "secret sauce" Luts. 

However, their is an open source 3DLut/ICC profile software package that it targeted towards displays and printers that is excellent and can be adapted to the types of 3DLuts we use with a bit of effort. It is ARGYLL CMS ( https://www.argyllcms.com/ ). On the website it has source code, tutorials etc. and explains a lot, primarily directed to ICC profiles, but the theory is sound. On the web, some users will talk about film emulation too using Argyll CMS.

Most all of the transforms I do are calculated from my proprietary code that I have developed over the years. Also, many of the specific Luts for clients are proprietary to the specific post facilities and unfortunately Ican not share them.

Another place to look for good info is the website for the Munsell Color Science Laboratory (https://www.rit.edu/cos/colorscience/rc_useful_data.php ) and(  https://www.rit.edu/cos/colorscience ). It is associated with RIT (Rochester Institute of Technology) in Rochester NY, where Eastman Kodak is located. There are a few people at Munsell that were in my group when I was in the Color Science and Engineering Lab at Kodak. One person that is very knowledgeable is David Long.

Cheers Mitch Bogdanowicz

[cameronrad]

6 hours ago, Lowepost said:

Modeling of Motion picture type of 3DLuts is not well documented or taught in the literature. Many post production houses rely upon their either in-house or contracted color scientists for their "secret sauce" Luts. 

However, their is an open source 3DLut/ICC profile software package that it targeted towards displays and printers that is excellent and can be adapted to the types of 3DLuts we use with a bit of effort. It is ARGYLL CMS ( https://www.argyllcms.com/ ). On the website it has source code, tutorials etc. and explains a lot, primarily directed to ICC profiles, but the theory is sound. On the web, some users will talk about film emulation too using Argyll CMS.

Most all of the transforms I do are calculated from my proprietary code that I have developed over the years. Also, many of the specific Luts for clients are proprietary to the specific post facilities and unfortunately Ican not share them.

Another place to look for good info is the website for the Munsell Color Science Laboratory (https://www.rit.edu/cos/colorscience/rc_useful_data.php ) and(  https://www.rit.edu/cos/colorscience ). It is associated with RIT (Rochester Institute of Technology) in Rochester NY, where Eastman Kodak is located. There are a few people at Munsell that were in my group when I was in the Color Science and Engineering Lab at Kodak. One person that is very knowledgeable is David Long.

Cheers Mitch Bogdanowicz

Thank you so much Mr. Bogdanowicz!

I will definitely look into that!

I often times wish I knew about the Munsell Science Lab when I was in high school so I could've followed that path. I've thought about taking the summer program actually. I never knew there was an entire field of color science and didn't discover my passion for it really until a few years ago. It's been a slow/tedious but addicting process learning this. Unfortunately I do wish I practiced mathematics more and didn't forget so much since H.S. I also need to improve my ability to work with code and command line utilities vs. relying on GUI based software. I'll try to find an apprenticeship/internship around me to better understand this field. I'm in Los Angeles, so i'm fortunate to be surrounded by some of the best. 

I've messed around with argyllcms before but wasn't sure if it was suited to the task. I've also thought about messing with this toolset http://colour-science.org/ however I still need to become more comfortable with command line tools. There seems to be a lot of misinformation, misunderstanding, and overall bickering between between the ICC (Stills/Computing) community and 3DLUT/Motion community. I've read about 3DLUT/Table based ICC Profiles and remember a few years ago the ICC community was examining motion picture workflows, but it seems like they've abandoned that. http://www.color.org/ICC_Chiba_07-06-19_PM_DMP_Float.pdf It seems as though motion is pushing towards the ACES approach, which from my understanding shares a lot of similarities with the ICC Color Management approach. I may be wrong but I think at the core, a lot of people misunderstand how the color management systems work. To me it seems as though ACES is a more explicit color management system than ICC but principle is the same. Transform from color space to another. The differences being how integrated the are into background processes. ICC working more in background, ACES working more through defined parameters. 

That said I hope something is agreed between the two industries. Over the past few years I've only really seen developments in the motion world and I wonder if the stills world will follow in that direction. 

Do you have any preferences on camera profiling methods for stills cameras? Adobe seemed to abandon ICC profiling for DSLR/Still Camera RAW files and now uses a DNG/DCP Profile with a "2.5D LUT" (Hue/Sat Tables) however uses ICC as working/output spaces. Adobe Camera RAW/Lightroom internally use MelissaRGB (ROMM Primaries/Linear gamma aka RIMM) as a working space but convert to an output "working space" like AdobeRGB when processed. Capture One still uses ICC, however they have a unique way of developing and implementing theirs. They are LAB based instead of XYZ and they are applied similarly to a LUT, whereas the RGB Data is just assigned with a new profile instead of converted.

One of the most desired looks/aesthetics amongst still photographers is the skin tones and overall color found in certain Kodak film stocks like Portra or Ektar as scanned through a Frontier or Imacon scanner. http://www.lettherebefilm.com/exposures/ Like the motion world, we have a plethora of "film emulation presets" for sale. But unlike cinema, with stills presets none have been properly created with proper data collection. They typically are just 1D Curves and crude adjustment presets for software like Lightroom. Although I personally don't apply "film emulation" to my still work, I am intrigued at the possibility of creating camera profiles/LUTs/ICCs/etc to help remap my digital cameras color response to a more "natural", subtractive color looking color palette. 

By the way all your patents are awesome. I'm always blown away when I read them and the descriptions for the Kodak Look Management System and the color management systems you helped develop. 

Edited January 14, 2018 by cameronrad

[Jussi Rovanperä]

I think one problem with profiling Portra or Ektar is that those don't come in 400ft rolls...

Steve Yedlin has been doing film profiling that's impressive, I've understood that he's using a remote controlled Arri Skypanel as the color source, shooting without a lens in camera, and reading the developed film with an i1 spectrometer. Also I think one reason why he likes the Alexa is that with .ARI files you can also access the raw data before demosaicing/wb/color transforms etc.