I didn’t like the colors produced by either “Camera Standard” or custom profile “Adobe Standard” for my Panasonic LX100 so I did some tweaking using the Color Toning / Color correction regions tool.
In order to free up the Color Toning tool for other purposes, I created a HaldCLUT image in RT, following the instructions in Rawpedia. However, when applying this HaldCLUT image to a picture using the Film Simulation tool, the resulting colors are not identical to the ones I have when using the Color Toning tool directly. Differences tend to get more pronounced as I apply other corrections, e.g. Exposure Compensation or White Balance, sometimes even resulting in pronounced color shifts.
Is there a reason for this? If so, is there a way to avoid this? Or am I doing something wrong?
Can’t help you with your problem (although at a wild guess, it might be something to do with the fact that film simulation and colour toning appear at different places in the pipeline, so are influenced differently by the other settings that you have).
I also have the Panasonic lx 100 so I was curious what changes you decided to apply.
I tend to use ART or darktable but both can use HALDCLUTS
I know nothing about RT, but I suppose the Color Toning tool uses formulae to calculate the transformation. A Hald CLUT is a lookup-table with a finite number of samples, and interpolation when the colour to be transformed isn’t one of those samples.
Yes, I did. From my understanding of how HaldCLUT works, I would assume that the masks “find” the right colors in the Identity.tif, and transformations are done correctly.
But then, I don’t know more about this than what I read in Rawpedia.
Color perception is highly individual, so what works for me doesn’t necessarily work for you.
The following is based on the use of “Adobe Standard” profile.
A general slight shift of the entire image from green to blue.
A shift of blue tones in the direction of purple, together with a de-saturation.
I mentioned above why Hald cluts will generally give different results to formulae. There is another possible issue.
The thread HaldCLUT link is not working - #5 by Mira_Studio has a large zip of Hald cluts. These have 8 bits/channel/pixel. So precision has been lost, due to rounding to 8 bits. However good the interpolation is, we can’t get that precision back.
8 bits is okay for a final output. We won’t notice that loss of precision. But if we are further processing the result, the loss can cause problems. The heavier the processing, the greater the problems. This may explain:
Differences tend to get more pronounced as I apply other corrections, …
I doubt that it is related to the 8bit precision – the differences are way to big for that.
I found that I get better results when I set Mask Blur in the Color correction regions to the leftmost value of -10. I thought that 0 already meant “no blur”, but apparently this is not the case.
Still, the result is far from identical. With manual tweaking of the Color correction regions I managed to get an acceptable (though still not nearly identical) result with which, I think, I can live for the time being.
Nonetheless, it would be interesting to know what really is the reason for the very obvious differences. Is it just the way it is, or am I doing something wrong?
Remember that the Film Simulation tool can only reproduce global tonal changes, so make no local changes - no local contrast, no tone-mapping, etc.; make no changes which move pixels - no distortion correction; use no sharpening or noise reduction; make only global tonal adjustments - color and saturation changes, curves, levels, Lab* adjustments. Save the sidecar file or write down the changes you made so that you can reproduce the changes in the next step.
So, I guess how you apply Color correction regions is somehow out of the scope of pixel-to-pixel LUT map – which ever way that doesn’t allow to map each color of the spectrum to another color by looking up the LUT map…
Bit late to this one but I would use the adobe dcp file for your camera. Okay now you need to tweak color. Download the Adobe Profile editor. You can tweak the tone curve. Add a second one and also pick colors in your image with a picker and then tweak them for adjustment as you need to correct…now save that as a new adjusted dcp and use that for color…unless you have a colorchecker card and then you can use that with the profile editor or use it to create a custom icc…These would be better options than your lut…
I tried the Adobe Profile Editor many years ago and couldn’t get convincing results. Might have been been because changes there are very local, or it might have been just me.
Anyway – I did manage to get the colors I wanted using RT’s Color Toning tool. My question was – how can I get these exact changes into a HaldCLUT? Because when I tried this, the results were not identical. As Vitalii above indicates, maybe it is just not possible.
My question was – how can I get these exact changes into a HaldCLUT?
A Hald clut is for global colour changes. It can’t be used by itself for local colour changes.
If you have a set of known operations in any software that apply global colour operations, then you can repeat those operations on an identity Hald clut to make a new Hald clut that will replicate those operations on any image.
An “identity Hald clut” is a Hald clut that makes no transformations. It can be made with ImageMagick:
magick hald:16 idhaldclut.png
So you then edit idhaldclut.png with global colour operations and save it as another file, eg “myNewHald.png”, and then use that new Hald on any images to replicate those operations.
Thank you. This is my understanding and I think is what I did, using the Hald_CLUT_Identity_12.tif from Rawpedia.
As already mentioned in the initial post, I used the Color Toning / Color correction regions tool for color changes, including the masks available in this tool. Since these masks are color-based and not region-based, I would assume that their effect on the Hald_CLUT_Identity is the same as on the original image. Is there a reason to assume that this is not so?
I don’t know RawTherapee. For those that do, I suggest you post whatever is needed to reproduce the problem. This would include an input image and identity Hald clut, whatever RT processing you did to those images, and the outputs.