You can use “Using darktable-chart ” in the darktable usermanual to build a style mimicing the c1 output using the colorchecker chart in the image.
DT-Profile.zip (422.9 KB)
Use the attached profile they are for A6300 but the sensor are the same so should give you good results select Sony like base curve. One is made with CC and one with SG chart of x-rite. I have similar made for capture one too.
Please give your opinion after use.
“Display-referred workflows suck, base curves have no place before input profiles,”
Agree absolutely. So why does the channel mixer go after the base & tone curves?
It doesn’t anymore in dt 3.0. One more thing we have fixed.
Yay! Thank you.
I like C1 profiles for their sky and foliage colours but somehow I find I need to lower the saturation for skin tones – esp. the shadows get too saturated for my taste, probably because of the various RGB curves applied around the input profile. But I no longer use the C1 profiles with dt and do as @anon41087856 suggested, or use my own DCamProf profiles.
Are you sure C1 just use a base curve ? From what I see, they darken blues, desaturate greens, resaturate skin tones and shift them slightly toward red. I suspect there is an artistic LUT going on there, it does too many things at once to be a simple tone curve.
This is the closest I can get from the C1 output, using the provided .icm and only transfer functions:
Sony Alpha 6400.dtstyle (1,8 Ko)
Note that, if you truly want to achieve the same look that the C1, you will need to deal separately with colours, like desaturate greens and shift them slightly to cyan, darken cyan, lighten magenta, shift yellow slightly toward green and desaturate orange/yellow and resaturate reds. So I suggest you build a LUT: PIXLS.US - Profiling a camera with darktable-chart
Actually, the Phase One ICM profile contains 3 non-linear TRC:
So it’s easier to disregard this profile, use the Adobe DNG matrix (linear) and unroll filmic + base curve: Sony Alpha 6400 Adobe matrix.dtstyle (1,7 Ko)
But C1 still has an advantage on skin tones, and I bet they use a LUT to enhance reds and oranges, because, if you look at the colour checker, the primary patches look less saturated on C1 output, the green looks muted, yet the skin tones are more vibrant.
I am not surprised. Emulation doesn’t have to be exact. One simply needs to understand what is going on and then decide what to do with that info.
To be clear, I matched the lightness and contrast reproduction of the C1 output (± the algorithmic changes), and roughly the saturation reproduction, it’s the specific colours rendition that doesn’t match.
darktable output (filmic + colour balance with Adobe DNG convert matrix):
capture one output:
The norm and luminance are still quite different. The geometry doesn’t match either, so lots of flipping required ┻━┻︵ヽ(`Д´)ﾉ︵ ┻━┻
When I examine the orientation, there is more discontinuity next to the bottles in the dt rendition.
The grey ramp matches, so luma is ok. As I said, C1 adds red everywhere and mute the greens. For the bottle, it seems to be noise, probably linked to differences in black points. As for the distorsion, lensfun doesn’t have that lens + camera in database, so I picked the closest I got.
How do I get the Adobe DNG matrix ?
The Adobe DNG matrix is what darktable calls the standard input matrix, it’s already in the software and used automatically when you open the file.
I did some progress. Thanks everybody for the precious advice.
@Rajkhand: your profiles are pretty good, however they generate negative RGB values with saturated colors in the test image, don’t know why.
@anon41087856: you are right, C1 do use a LUT plus curves, so their look is difficult to replicate with standard tools.
All in all I found only two possible ways to replicate accurately C1 colors:
- Use their profile, which require a base curve (or RBG curve) before input color profile, which is not the default DT module order
- Use the standard color matrix + LUT. I have created a Hald clut using GMIC and it works very well. This method has also the advantage that I can replace the basecurve with filmic for HDR images. However for accurate results, the 3DLut module must be put after the basecurve / filmic, which again isn’t the default module order
At this point I have two questions:
- LUT are commonly used for artistic / color grading purposes, which is something you do after the tone adjustment. I believe the 3DLut module should be placed after basecurve / filmic, not before.
- More in general, in DT 3.0 the user can reorder modules, and I think it should be possible to save a custom module order in a .dtstyle.
What do you guys think ?
Marco
That is one application of many. It depends on what the CLUT is designed for specifically.
This and the fact that dt allows more than one instance of a module makes it fun.
Agree, but multiple instances can be put is a style, while module reorder cannot
Have you used the default Sony base curve? Which saturated colors give negative values? Also I have seen that the WB of the test image is 6700 where normally the test chart should be shot at 5000/5500. I also use Capture one 12 and the default generic profile gives me yellowish tint skin so I have done the profiles using test charts from imagine resource for A6300 you can use this link https://www.imaging-resource.com/PRODS/sony-a6400/sony-a6400A7.HTM It gives many test shots at different ISO which can give a better understanding of the profile. In my case I am satisfied with the results I get from the profile, your mileage may very as per your individual color choice.