Conversation
|
Does this PR take #21682 into account? |
ralfbrown
added
bugfix
4c63482 to
abdb21a
Compare
|
I also added the correction of the Lab color values of the SpyderCheckr Photo to this PR and updated the the PR description. |
Thanks for doing this. I was trying to find the time to get this in but I have just had way too many things on my plate at home and at work.... Those are the reference values now from the actual Spyderchecker software so that should make things accurate. One other small change is as far as I could see there should only be 3 Datacolor entries... 24 patch, 48 patch and photo. I could find no evidence that there was two run's based on the dates shown. This comes from a blurry screen shot that Aurelien had and a couple of rounding issues as I think a couple of others concluded as well... So we could and should clean u the dropdown selection as well..... |
TurboGit
left a comment
TurboGit
left a comment
There was a problem hiding this comment.
A question, is this change a bug or needed because the new spydercheckr reference has changed? If the later we need another entry for the new color checker as done currently:
- Datacolor Spydercheck 24 (before 2014)
- Datacolor Spydercheck 24 (before 2018)
- Datacolor Spydercheck 24 (after 2018)
|
I don't think there is an old and new...If you look closely that is pretty obvious. So imo there is only a single version of the data for the 24 and 48 cards and the photo is newer circa 2022 and has different values. This is further confirmed in my mind by
the fact that the data I pulled are from the reference file of the data color calibration software, ie what the company uses to calibrate its cards then I think that makes it clear...
Sent from
Outlook for Android
From: Pascal Obry ***@***.***>
Sent: Sunday, 20 September 2026 05:23:07
To: darktable-org/darktable ***@***.***>
Cc: Prior,Todd ***@***.***>; Mention ***@***.***>
Subject: Re: [darktable-org/darktable] Fix spydercheckr reference patches (PR #22278)
Caution: External email.
@TurboGit commented on this pull request.
A question, is this change a bug or needed because the new spydercheckr reference has changed? If the later we need another entry for the new color checker as done currently:
Datacolor Spydercheck 24 (before 2014)Datacolor Spydercheck 24 (before 2018)Datacolor Spydercheck 24 (after 2018)
—
Reply to this email directly,
view it on GitHub, or
unsubscribe.
Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for
iOS and
Android. Download it today!
You are receiving this because you were mentioned.Message
ID: ***@***.***>
|
It is a bug fix. In my understanding Darktable never used the correct values for the Color Checkr Photo. |
Read the investigation by @todd-prior in #21682 Yes, it would appear that all previous uses have resulted in wrong results, at least for the SpyderCheckr Photo. |
|
@Donatzsky @mwenzkowski : Thanks, I missed the issue with all the explanation. So, at this point the fix is also to remove all the versioned data, so we should ends-up with a single Spyderchekr 24 and a single Spyderchekr 48. If we all agree on this, please let's do that on this PR. TIA. |
I think that the correct course of action...Thanks for taking the time to review all that...... |
|
@TurboGit @todd-prior Okay, I'll try to add that to this PR. |
The Photo target carried the SpyderCheckr 48 reference values, but the Photo card uses different pigments. Take its values from the reference file shipped with Datacolor's SpyderCheckr software, shared in darktable-org#21682. Related: darktable-org#21682
22610f8 fixed middle_grey and white being swapped on the two SpyderCheckr 48 targets, but missed SpyderCheckr Photo.
All three 48-patch targets took middle_grey from D4. Use E4 instead, matching the other color checkers, which all use a patch close to L* = 50. middle_grey is meant to be the patch closest to 20% neutral gray, and E4 reflects about 20% while D4 reflects over 25%.
c75fea5 split the SpyderCheckr 24 and 48 into pre- and post-2018 entries. As far as we know, there never was a separate 2018 version of these charts. Datacolor's own SpyderCheckr software ships a single set of reference values for them, and the pre-2018 tables differ only in a* values rounded to one decimal (six on the 48, three on the 24). Keep the post-2018 tables. They match Datacolor's software except for three more a* values rounded the same way (48 A3, 24 B1 and B5). Fixes darktable-org#21682
Three color values differed from the reference file shared in darktable-org#21682. Update them to match the file. Related: darktable-org#21682
abdb21a to
101e662
Compare
|
Done. The versioned entries are merged and the PR description is updated. I also updated three color values of the SpyderCheckr 24 and 48 to match the reference file from #21682. |
mwenzkowski
changed the title
|
Nice work |
5 participants
This fixes several issues with SpyderCheckr color checkers:
Incorrect Lab reference values for the SpyderCheckr Photo.To fix this I have taken the values provided by @todd-prior in Color Calibration - Data update for Spyder Colorchekr photo..... #21682 (comment). To double check these values I requested the Lab values from datacolor as they mention it here, near the end:
The values I received from datacolor are coarser and match the values mentioned above by rounding them to one decimal place. For example for Patch
A1datacolors reports61.5,32.3,14.4while the table from @todd-prior states61.51,32.31,14.43. I decided to use the more precise values.I don't know if it is allowed by datacolor to share their document here, so to be on the safe side, I prefer not to do this. But anyone interested can request the values from datacolor in the same way I did.
Indices for middle gray and white patches are swapped for the SpyderCheckr Photo. Similar to 22610f8.
All 48-patch SpyderCheckr's are not using the gray patch that is closest to middle gray.
The pre- and post-2018 entries of the SpyderCheckr 24 and SpyderCheckr 48 are merged into one entry each, as agreed below and in Color Calibration - Data update for Spyder Colorchekr photo..... #21682. As far as we know, there never was a separate 2018 version of these charts. The post-2018 values are kept.
Since darktablerc stores the chosen chart as a list position, a saved SpyderCheckr 24 post-2018 now opens as the SpyderCheckr 48, and either SpyderCheckr 48 as the SpyderCheckr Photo, until the chart is picked again. The saved position is clamped so the list always matches the chart in use.
Three color values of the SpyderCheckr 24 and SpyderCheckr 48 differed from the reference file shared in Color Calibration - Data update for Spyder Colorchekr photo..... #21682 and are updated to match it.
Note: I am not completely sure if the third fix is needed/desired. I only included it for consistency with the other color checker's data and because the middle gray index is documented as
If it is deemed unnecessary, I can drop this change from this PR.
Testing
I have tested the changes with the SpyderCheckr Photo (I don't own the other ones):
I applied the color calibration with the colorchecker and set the exposure compensation value as calculated by the color calibration module.
Before
After
Color Checker selection
I tested selecting each chart in the new list. I don't own the SpyderCheckr 24 or SpyderCheckr 48, so I couldn't calibrate with them.
Fixes #21682