Ciecam and Rawtherapee

Software RawTherapee
1

Ciecam in Rawtherapee

A few weeks ago, I was reading an article in the French Photographic press about RT, here are its main conclusions.

  • performances and functionalities are equal to those of the top players on the market
  • the “advanced tab” contains 3 products - « Wavelet levels », « Retinex », « Ciecam » - we didn’t find what they could be used for.

If, for « wavelet levels », an answer has been brought recently, in particular by updating the documentation on Rawpedia, as well as by the videos provided by @Andy_Astbury1

Wavelet Levels - RawPedia, it is not the same for Ciecam.

I will develop in several points some information on « Ciecam and Rawtherapee »

  • what it’s for ?
  • what is it ?
  • where is it located in the process and what is its role in colorimetry?

What it’s for ?

I will take 6 examples - of course not exhaustive - to answer this question :

  1. We are going to process a high dynamic range image with heavily underexposed areas . We will use Ciecam to modify the Source lighting
    https://rawpedia.rawtherapee.com/CIECAM02#Example_1_.28Advanced_Tab.29
  2. Starting with an almost perfect white balance, we will use Ciecam to carry out a chromatic adaptation
    https://rawpedia.rawtherapee.com/CIECAM02#Example_2_-_Chromatic_adaptation
  3. We will show the impact of the output device and its environment (using a holiday shot to be viewed on the family TV in the afternoon). CIECAM02 - RawPedia
  4. We are going to process a high dynamic range image, using “Local adjustments” (full image), « Log encoding + Ciecam »
    https://rawpedia.rawtherapee.com/Local_controls#High_dynamic_range_image_.2B_Ciecam
  5. We are going to make a “Dodge and burn” using using “Local adjustments”, Log Encoding and ciecam https://rawpedia.rawtherapee.com/CIECAM02#Example_5_.28Local_Tab.29_-_Dodge_and_Burn
  6. we will use an abstract profile in combination with Ciecam (under development)

What is it ?

https://rawpedia.rawtherapee.com/CIECAM02#Color_Appearance_.26_Lighting_.28CIECAM02.29_and_Log_encoding_modules

Where is it located in the process and what is its role in colorimetry?

https://rawpedia.rawtherapee.com/Toolchain_Pipeline

jacques

9 Likes
2

Great to see more documentation. I just yesterday noticed the amount of local adjustments info in rawpedia. Tested a difficult back lit image with good results.

One comment. I think the output concept should result in rethinking/tweaking the Rawtherapee ui and workflow with regards to export. Currently if you adjust an image for a new output you loose your previous settings. For me thats already a limitation of Rawtherapee.

The solution would be to either allow for “virtual” copies or the ability to assign export styles like ART. Saving and loading custom pp3 works but is a bit clumsy and complicated.

3

@jdc Superbe travail et toujours aussi ciselé et pédagogique. Merci

4 Likes
4

Sounds like a case of lazy journalism.

I love ciecam, though mostly use it in creative fashion - leaving the ‘scene’ and ‘viewing’ conditions in a neutral fashion (no visible effect on image), and playing with the ‘image adjustments’.

2 Likes
5

I would like to point out an important improvement concerning Ciecam. From now on you have the choice between Ciecam02 and Ciecam16. (commit 6f9da68 - 9 december 2020)

Of course, compatability with 5.8 is preserved and by default in the main Ciecam module, the setting is set to Ciecam02.

As far as « Local Adjustments » is concerned, the current modules using Ciecam (Warm-cool, Log Encoding ) are set to Ciecam16.

You have the same functionalities, the same 3 processes, etc., mixing the 2 codes was a bit complex, nevertheless everything works correctly (I hope).

I won’t go back over the main features that have been succinctly presented in this thread, other than to recall the titles

  • process a high dynamic range image with heavily underexposed areas
  • Starting with an almost perfect white balance (I remind you that “Auto wb Temperature Correlation” is one of the best modules to realize the White Balance…it is very reliable in a majority of cases) , we will use Ciecam to carry out a chromatic adaptation (symmetric mode)
  • show the impact of source and output device and their environment
  • take into account color appearance model (CAM), which was designed to better simulate how human vision perceives colors under different lighting conditions,
  • participates in the creation and improvement of digital camera calibration processes
  • etc.

I would like you to communicate around you on Ciecam. Indeed between “The journalist who doesn’t see what it can be used for”, or users who say “I haven’t yet needed to use Ciecam in my usual photographic workflow”, show the leap that needs to be made.

It’s like having free access to an award-winning gourmet restaurant and still going to the canteen, or taking a picture in photography, having a Nikon Z7 or Sony A7III (or equivalent) in a drawer, and using an old Iphone3 to make quality photographs.

I enclose an “old” text (2005 - Mark D. Fairchild) on Ciecam, which describes the various models used in colorimetry … And don’t tell me that it is “outdated” that today’s technology allows to do better (yes a little with Ciecam16); what is described, for the most part, are the physiological aspects… which probably date back to homo-sapiens…

Ciecam-Fairchild_M._Color_appearance_models__2005 .zip (3.6 MB)

jacques

3 Likes
6

It is worth noting that simply activating the module is sometimes all that is required to give a pleasing and natural-looking lift to the image lighting,

2 Likes
7

CIECAM02 is not designed for HDR, it will most likely not hold colours as it is supposed to in SDR.

8

Hello @jdc

First off, thanks a lot for your work on RawTherapee. I follow your daily updates on github and I am really looking forward to see the 5.9 version released, as soon as it is ready, with all your improvements included (local editing and such).

I have tried to read the PDF you have zipped but it is far too diffucult to understand for me since I am far too “mathematically-challenged” (I am a poor plant pathologist…). It is “chinese” to me: I copy here this same sentence from one of your recent post on github :slight_smile: :slight_smile:

Lately, I have read that DarkTable is trying to partially move away from Lab [1] and your post here has caught me by surprise.
I have read every color mode has its pros and cons, in the end, and there is not a real “winner” in every contest.

Once again, thanks a lot for your work on RawTherapee!

[1] PIXLS.US - Darktable 3:RGB or Lab? Which Modules? Help!

9

One thing I really would like people to understand is that colour models are numeric fittings on top of empirical psychophysical measurements.

CIELab 1976 was designed by asking people to sort color patches by order of appearance for luminances between 0.2 and 100 Cd/m², if I recall correctly. So that model holds in this range.

But now, consumer monitors easily go up to 500 Cd/m² and counting. Using that kind of model outside of its validity bounds is asking for trouble. Every empirical model has bounds of validity, that’s like epistemology 101.

Arguments like “human perception should not have changed a lot since 2002” baffle me. Yes, but that’s absolutely not relevant. Do you compute supersonic drag with subsonic fluid mechanics when you design an aircraft ? I don’t think so. And yet, it’s the same old air. And yet it becomes a compressible fluid for speeds above Mach 0.9. So, the bounds matter to any model.

CIECAM02 and, as far as I know, CIECAM16 are still SDR beasts. Using them to perform adaptation from SDR image to SDR display is what they have been designed for. Using them to perform adaptation from HDR to SDR is really not what they are about.

HDR CAM are still under active research :

https://www.ingentaconnect.com/content/ist/jpi/2020/00000003/00000002/art00002;jsessionid=36htcjom239uj.x-ic-live-03

3 Likes
10

I’m not going to get into endless debates about the pros and cons of each model, whether you think of it as a dodge or not, I don’t care.

I would make only one remark concerning «Uniform perceptual Color for Lab ». since 2011/2012, after long internal debates within the team, it was set up, at my initiative, a correction of the Lab values, to make them appreciably in conformity with the values of Color Munsell. This is done by about 200 LUTs which act - as well as a gamut setting to avoid imaginary colors - as soon as one of the “avoid color shift” checkboxes is activated.

One last remark, I think there is no right or wrong way. Rather the advantages and disadvantages of RGB, Lab*, and Ciecam need to be understood and used appropriately.

jacques

3 Likes