As darktable-chart creates a curve based on the gray ramp of a color checker I’ve just used what it produced:
https://blog.pixelbook.org/2018/10/sony-alpha-7-iii-basecurve-and-camera-profile-for-darktable/
1 Like
That is very interesting. I just begun to explore darktable and I am also having a hard time getting the colors right for my Sony Alpha 6000. I know this is a bit old but it’s it possible by any chance that you might share your extracted color profile (or the correction style that uses your settings) - it might help me learn how to process my images and I wouldn’t need to install capture one to extract the profile you mentioned. That would be very kind…
Hi @MalcolmMcConnor and welcome!
I am a bit puzzled: why cannot you follow the advice given two posts above,
i.e. what @guyru wrote on Oct 18?
Have fun!
Claes in Lund, Sweden
Hm, it’s just a matter of taste. I can’t make up my mind which one I love better. But I think Dakrtable image seems a tiny bit fuzzy. Obviously C1 images all have more contrast but I’m wondering if any deconvolution has been done. But I think it wasn’t because the difference is so small.
Yeah, C1 looks a bit better but I presume just using filmic and maxing out contrast slider will do the trick. The curve will be yellow but what can you do if that’s the way to achieve this look.
Malcom if you don’t have a color checker try to grab a test shot from imaging resources. This will be crude but I have done if for some other people and then suggested an improvement. Grab both the jpg and the raw and follow the instructions for using Darktable chart you can make two styles…one is comparing the jpg to the raw to give you a profile to match that and one can be compared to the CC24 values for color matching. If you get stuck I can try it for you. You will end up with a LUT and a tone curve in a style…See if that is better than the base curve or your current approach…if you are not shooting your own image just ignore the first part and make your pfm files and run the program. YOu will find it in your DT bin directory. PIXLS.US - Profiling a camera with darktable-chart
Default image no processing
Default scene referred
Darktable chart style…I did WB the raw bit forgot to WB the JPG
THis is the tone curve
I’ll go back and WB the jpg to see if it makes a difference.
This is the matching JPG
EDIT Re ran DT chart wb both Raw and JPG on the third gray patch
RAW with WB style…
Your tone curve should be in LAB space, no?
Ya I think it should be but that is the way chart makes the style…from lab output exported as pfm files and the style uses the tone curve module…maybe a default change I can’t say as I ever noticed what the module said before or if this is usual or a new thing?? I just applied it as created by DT chart??
Thank you for the detailed information you provided also through the links. Probably my understanding of the matter is way too crude to fully understand how to do that myself. I was under the impression that Sonys “capture one” software is provided with some kind of calibration file or calibration table which could be used to achieve the same look the jpg have out of the camera - therefore I asked if that could be made available (That being said I also think the colours are not perfectly accurate OOC but pleasing to my eye). I thought that the OP managed to extract those calibration file or a table and used it to reproduce the cameras look. With that file I could have tried to reproduce that myself.
To my understanding color reproduction from raw in darktable is one part base curve (which I adjusted to a pretty pleasing curve myself starting from the alpha like preset - I will try the a6300 curve mentioned above though) and some color calibration. My colors are off even after the base curve applied. Especially the Blues are visibly shifted to cyan and a bit Red is missing. Tried to compensate for that with several colorcorrection modules within darktable but was not able to remedy the issue.
I would even start to calibrate using the raw and corresponding jpg of the camera but it seems after reading the above posts I still have not understood how to do it within darktable. There seems to be much more to read on than I can handle before I could even try to begin…
In the meantime I too have taken on the approach with the C1 Alpha 6000 Profiles taken from Capture One Express 21 - too bad weather to shoot my own target to get profiling.
It turns out I can now get very close to the OOC JPG (which is a good thing in my opinion). The critical part is the adjustment of the base curve.
DSC00447.ARW.xmp (14.8 KB)
DSC00452.ARW.xmp (23.7 KB)
I am not quite sure why the picture with the blue jacket differ in color when uploaded - here on my Spyder 5 calibrated display they look pretty much the same (the not OOC darktable edited JPGs blue actually a bit shifted to cyan but not as much as seen here). I suspect I got something wrong with the settings but I am not adept enough to see just what.
Anyway with that generic setting and that basecurve I get an good starting point most of the time. There is also a ProStandard Profile which still emphasizes the cyan even more so I am not using this at the moment… The basecurve looks somehow odd to me - I suspect I should have fiddled with the “unbreak profile” module but could not figure out good settings for this…
Some time ago I did the same experiments and I developed styles to emulate C1 colors for my Sony cameras (A6000 and A6400).
For A6000, see attached:
Tone & Color_Sony A6000 C1 colors.dtstyle (1.8 KB)
For the style to work, you have to have the C1 color profile in the right folder.
Of course to fully match the C1 results, you have to adjust the exposure case by case.
In C1 the tone curve (or base curve) is applied in more steps and one part is inside the .icm color profile.
However, I no longer use those styles, as I fully embraced the new DT scene referred workflow.
I am beeing curious: do you have a similar style for “scene-deferred” emulation of the OOC style of the Sony Alpha 6000. I have not yet read fully into the scene-deferred workflow but I am thinking of taking the plunge eventually - even then I would still like to have a reasonable starting point for the edits… perhaps you like to share your findings?
Just to be clear, when I switched from C1 to DT, I developed some styles to emulate C1 colors, which I found to be pleasing, not OOC jpegs. They are very different!
Having said that, for the A6000 the only style I have done is the one I posted above.
For the A6400 I have done more styles, using different combinations of tools, here below listed in pipeline order:
- unbreak input profile + input color profile with C1 icc + tone curve - default module order, similar to the A6000 posted
- unbreak input profile + base curve + input color profile with C1 icc - custom module order, this is the most accurate color match
- filmic rgb + color look up table - custom module order
When coming to scene-referred vs display-referred workflow, as I said the C1 color profile contains a non linear tone curve, plus it requires the module unbreak input profile, also non linear. Therefore using it is incompatible with the scene-referred workflow, which is supposed to be linear rgb in the pipeline until reaching filmic.
So, of the three above, only the last is scene-referred compliant, where the C1 color emulation is done through a LUT after filmic, so it is treated as a pure artistic color grading. It works in normal conditions, but it is not guaranteed to be 100% equivalent.
As I said, after the switch, I learned to do my own color grading in DT scene-referred and I no longer use those styles, but if anyone is interested I can post them.
Thanks you for that explanation and the offer. If it’s not too much a hassle I would be interested in those styles. Might be a good starting point to learn… Greetings.
Here you are:
Tone & Color_Sony A6400 C1 colors (C1 profile + tone curve).dtstyle (1.9 KB)
Tone & Color_Sony A6400 C1 colors (basecurve + C1 profile).dtstyle (2.8 KB)
Tone & Color_Sony A6400 C1 colors (filmic v4 + LUT).dtstyle (2.7 KB)
The first two styles require the A6400 C1 color profile.
For all styles in order to have a perfect match you need to adjust exposure and for the last one also filmic. Of course I didn’t want’t to hardcode those in the styles.
Enjoy
EDIT: if you want a good starting point for filmic to use together with the last style, import the following preset (remove the .txt extensions first)
[filmicrgb_A6400 studio scene.dtpreset.txt|attachment]
This works well with SDR and studio images
filmicrgb_A6400 studio scene.dtpreset.txt (1.1 KB)
Based on the copyright note in the ICM files, I don’t think they should be shared here!
Copyright (c) 2019 Phase One A/S. All rights reserved.
https://discuss.pixls.us/guidelines#stealing
Ok, thanks, I will remove them
Thanks for this! The ICMs are easily obtained by legitimly downloading Capture One Express. One can find the ICMs for many Camera models in a subfolder of the program directory.
The first two presets are named the same and are the same size. Do they differ in effect somehow?
OOps, I uploaded the same file twice, fixed
@mcconnor Michael I have modified a script (colormatch) that Pascal DeBruyn created to use ArgyllCMSColormatch.zip (6.1 KB) . It creates a blended icc from a raw and jpg pair. A bit like DT Chart but it creates an icc not a style. It might be worth a try to. I find it gives and excellent result. Basically the script is set to run on a IT8 card . I just modified it to run on my Spyderchecker24. I will attach the text of my script you can compare where I modified it. The directions provided by Pascal are excellent and it creates a really nice profile GitHub - pmjdebruijn/colormatch: ColorMatch if you have a color card and even if not its not bad for color even from sample images taken from imaging-resources website. The attached zip has the necessary cht and cie files to run the script with a CC24 or Spyder checker and a modified script for each. You can compare these to the original on the site …There are only a couple of changes…I find it works nicely
Raw
JPG
Raw with ICC
Pretty nice result from a 700 k Jpg extracted from the raw and zero edit just the icc change in the imput profile
