Darktable calibration module suggestion

Hmm, that’s where the marks are, but mine on the Passport are little 'L’s…

Putting them in the corners also implies the target is square-on in the image, so rotating it to line up with the edges is also good…

I’m having Max dE ~ 5 for Spydercheckr48 and Max dE ~ 3 for Colorchecker Passport under DT. But these are just numbers. Visualization gives you more freedom.

1 Like

No, it doesn’t work. I’ve adjusted perspective and added some space for white marks.

Scanin doesn’t see these marks. It tries to recognize the borders of the patches as you can see. It puts reds L but in wrong positions. Maybe it may work better with Colorchecker but I prefer spyder.

I’m going to give this approach a rest. I’d better spend more time improving my DT workflow. I think there’s still more room for this.

Thanks anyway )

1 Like

I put “-F 146,254,3814,262,3815,2410,138,2401” key to scanin command and it works now. These are coordinates of the fiducual marks (Look like cht file has wrong values). Anyway it works now, but calibration didn’t go well enough in terms of white and black points and colors looks ok.

It took me ages to manually grab coordinates )))

Just a quick question for you. When do you adjust exposure? In DT you can adjust exposure before calibration and it helps. How it can be done for dcamprof?

1 Like

In the camera, I go for a solid ETTR. So, I’ll usually shoot a 3-exposure bracket, take the highest one that doesn’t blow any highlights, even the fiducial marks. It’s better to underexpose than to blow anything…

I don’t do any adjustment in post except for black subtract and white balance. I even use a half demosaic, doesn’t change the channel values like the others will.

With dcamprof, I’ve even played with doing white balance chromatically, by not white balancing the target shot and letting the profile do it Works, but one really needs to do the target shot in the same light as the scene so it’s a bit cumbersome.

@SpaceDreamer your crop looks too loose to me. Crop to the white dots.

I’ve managed to write a small bash script for linux users and owners of spydercheckr48. If you crop target image exactly to red dots and run my script it’ll prepare coordinates for scanin program for correct patch recognition.

If someone needs it…

# Save this as make_DCP_for_SC48.sh
# Generates DCP profile for Rawtherapee (for Spydercheckr48 owners and latest Argyll)
# Usage:
# ./make_DCP_for_SC48.sh target_file_name.tif
base=`basename $1 .tif`
arrIN=`identify -format '%w %h' $1`
arrIN=(${arrIN// / })
w=${arrIN[0]} ; w="$((w-1))"
h=${arrIN[1]} ; h="$((h-1))"
./scanin -v -F $fiducual -dipn $base.tif /usr/share/color/argyll/ref/SpyderChecker.cht /usr/share/color/argyll/ref/SpyderChecker.cie $base-diag.tif
# Following depends on your configuration
# dcamprof make-profile .....
# dcamprof make-dcp .....

Thanks for the points. I’ll will try them when I have time.

In the meantime. Here’s my synthetic comparison charts of calibration same RAW file using of Darktable and Dcamprof.

Darktable (avg dE=1.47, max dE=4.35)

Dcamprof (avg dE=2.22, max dE=5.31). Note.

To get this result in Rawtherapee I had to touch the Tone curve to match 6 greyscale patches. DT didn’t require tone twicking

There’s no clean winner, but DT has the edge I believe. Probably a more experienced dcamprof user might squeeze more from it, but DT gives me this out of box. Probably one day I’ll have time to learn more about Dcamprof and get more from it…