It’s not just a DCP profile that makes the difference, it’s what’s in it. I haven’t torn apart the particular profile in play here, but I’ll bet it uses a lookup table (LUT) transform instead of a matrix.
With a matrix transform, all the algorithm knows to do is to deposit out-of-gamut colors just inside the destination gamut, without any concern for giving them some gradation that approximates what they had “out there” in the original space. A well-constructed LUT will provide the information to do just that, approximate a gradation in the destination colorspace.
Making a LUT camera profile takes a bit more information than what you get from a ColorChecker target shot. I’ve tried an IT8 target and the LUT profiles from it work better, but not as well as using spectral data measured from the camera. dcamprof will make LUT-based DCP or ICC profiles from either target shots or spectral data.
The above is just a rehash/synopsis of all the crap I wrote about it here:
- https://discuss.pixls.us/t/the-quest-for-good-color-1-spectral-sensitivity-functions-ssfs-and-camera-profiles/18002
- https://discuss.pixls.us/t/the-quest-for-good-color-2-spectral-profiles-on-the-cheap/18286
- https://discuss.pixls.us/t/the-quest-for-good-color-3-how-close-can-it8-come-to-ssf/18689
- https://discuss.pixls.us/t/the-quest-for-good-color-4-the-diffraction-grating-shootout/19984/75
#3 talks about using large-patchnumber targets.
8 Likes
i was wondering when someone would say “spectral input profile”. here’s a vkdt render with a lut constructed from ssf:
(and just to match the other renditions better, here one with crazily pushed saturation, but constrained to rec709):
2 Likes
Ooh, you have ssf data for the Fuji X-t30? If you can share it, I’d like a copy for my collection… 
1 Like
Photoflow + a look with contrast and colourfulness set to ‘too much’ + OpenDRT.
I’m converting everything to ZCAM JzMzhz space for the look - you can do quite extreme things!
Camera profile was whatever matrix Photoflow has built in for the X-t30.
3 Likes
Interesting rendition. It also seems to suffer from the blue problem.
This one looks fine to me. Maybe because it uses the same DCP profiles RT uses?
Maybe someone can help me to extract the proper learning from this. As of now I feel the DCP profile makes the difference, hence RT will give inherently better results than DT. But I’m not quite sure if that explanation is maybe too simple since the vkdt rendering also looked quite good in the blue tones.
DCP are generally dual illuminant between tungsten and D65 and so you can interpolate from these two points …ICC are generally set to D50 WB and so I think a bit less useful outside the range of light from which they are produced than a DCP would be …but some of it is also software…RT lets you apply or disable components of the DCP file ie the base table the look table and the tone curve and also combine this with the auto match curve …so there are lots of options during processing and if you grab the ones from Adobe you can do so from a standardized starting point…RT also has abstract profiles that can be thrown in to the mix…however having said all that Aurelian’s approach to leverage the channel mixer to further tweak an existing profile is a nice tool on the DT side of things…
1 Like
Glen back when I was trying to sort out DCP files I found this reference.
http://dcptool.sourceforge.net/DCP%20FIles.html
They talk about the tables having a third dimension which is often set to 1 ie not used…
Is this the difference in Dcamprof when you go from the 2.5 to 3D Lut versions ie the 3D LUT uses the third dimension and having said that then are these Hue and Color Saturations tables the Luts or are the Dcam Lut profiles a different beast than the standard ones described by the specs… just wondering…
Anders describes it well in the dcamprof doc:
http://rawtherapee.com/mirror/dcamprof/dcamprof.html#camera_model
Edit: Oh, are they different from the DCP spec? Dunno, haven’t looked…
I only tried in Art not RT but ART is using a different DCP file which will be used by the automatch profile if that is what you are using. The standard adobe profile also has a look table…you should try snap shots with the default RT or ART profile and compare that with a custom profile and then select the Adobe XT-30 profile…here you will be able to turn the look on and off. I think the look might be something that would impact the overall look and the blues perhaps…I will try later if I get time…if you want to see the difference in the adobe and art DCP files you can look here…json files…had to zip them…just text so easily viewable…you can get the Adobe DCP easily if you wish to test…
XT30art.zip (210.6 KB)
EDIT:
In RT and in general with these blues…change the working profile and look at the affect …rec202 vs Prophoto vs Widegamut vs ACESP0…you might also find some magic here to tame your blues…
Tried to load your edit from the jpg…did you use a lut…looks like you did…
Thanks I’ll take a look at one later and compare…
here I used Adobe DCP
Film simulation seemed to be active
The wonderful manual mentions gamut compression in the color calibration module.
Gamut compression uses a perceptual, non-destructive, method to attempt to compress the chroma while preserving the luminance as-is and the hue as close as possible, in order to fit the whole image into the gamut of the pipeline working color space. One example where this feature is very useful is for scenes containing blue LED lights, which are often quite problematic and can result in ugly gamut clipping in the final image.
Did a quick test with the provided RAW file. It nicely removes the blue clipping and gives a more realistic tone to the overall illumination.
1 Like
I think it was part of a follow up video Aurelien made with some blue Christmas lights…
Ah, yes. I made some trial to see how it compresses the Gamut, the LUT being an identity. I Should have removed the activation.
Using ART or Adobe DCP, there is no problem in that raw with the blue instead of some images that were submitted in playraw or the blue Christmas light
1 Like