search

darktable-org / darktable

darktable is an open source photography workflow application and raw developer
https://www.darktable.org
GNU General Public License v3.0
9.65k stars 1.13k forks source link

Fujifilm X-T3: Incorrect color rendition #10413

Closed wilecoyote2015 closed 1 year ago

wilecoyote2015 commented 2 years ago

Describe the bug/issue In comparison to the other Cameras I am using with darktable (X-T10, X-E3, Nikon D40), the input profile for the X-T3 seems wrong: Colors are muted and a greenish color cast exists.

To Reproduce Please provide detailed steps to reproduce the behaviour, for example:

  1. Go to the Dpreview comparison tool
  2. Download Raw images for the X-T3 and T-T10
  3. Import the Images to darktable

Expected behavior Darktable should transform both images' input values properly so that they look (approximately) identical.

Screenshots I am not sure if it is legal to share images generated from the DPreview sample images.

Platform

kmilos commented 2 years ago

Indeed, it looks like the color matrix for A light instead of one for D65 somehow got merged way back in https://github.com/darktable-org/darktable/commit/afc12d4282ac521f947e20faca2bf3f1cc6845fb (could even be a bug in that old version of ADC) @LebedevRI

The correct one should be identical to X-T30, X-T4, X-E4, and X-S10 as confirmed by ADC 14.0, as they all use the same sensor:

{ "Fujifilm X-T3", { 13426,-6334,-1177,-4244,12136,2371,-580,1303,5980 } },

kmilos commented 2 years ago

Note also that this camera was used in dt before. @Mangus78

See also e.g. https://discuss.pixls.us/t/trying-to-tame-excessive-saturation-in-darktable-x-t3/18731

Mangus78 commented 2 years ago

Note also that this camera was used in dt before. @Mangus78

See also e.g. https://discuss.pixls.us/t/trying-to-tame-excessive-saturation-in-darktable-x-t3/18731

Agree, sometimes there seems to be some strange rendition both of colours and/or fine details in x-t3 raws (I haven't tried other fuji's cameras) I tried like the op said downloading some test raw from dpreview.

X-t10 is indeed different from the x-t3 raw, BUT, the X-T30 seems almost identical to the X-T3 (but ATM I'm not using a very good monitor at all, even if it's calibrated). X-T30 and X-T3 shares the same sensor, while the X-T10 is 2 generation prior.

github-actions[bot] commented 2 years ago

This issue did not get any activity in the past 60 days and will be closed in 365 days if no update occurs. Please check if the master branch has fixed it and report again or close the issue.

jakkul commented 2 years ago

I think that this might be just the way X-T3 renders colors in RAWs. Basically you can see this difference in DPreview as well. But there's a way out! You need to get yourself a colorchecker chart and calibrate using color calibration module in Darktable. If you don't have that, you can get RAW files with the charts photographed from the net! What I did:

  1. I'm on Darktable 3.8.1, linux, hardware calibrated EA271q nec monitor.
  2. I've found a portrait I made of my friend photographer holding an x-rite color checker lit by bigass broncolors. I've made it B-W for a joke but in this case it's no joke.
  3. I've downloaded the RAW from dpreview's pages for X-T3 here https://www.dpreview.com/reviews/image-comparison?attr18=daylight&attr13_0=fujifilm_xt3&attr13_1=fujifilm_xt10&attr13_2=fujifilm_xt30&attr13_3=fujifilm_xt4&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=160&attr16_1=200&attr16_2=160&attr16_3=160&normalization=full&widget=1&x=0.8611172444248179&y=0.17340469328941951 ISO 160
  4. same from Imaging Resource https://www.imaging-resource.com/PRODS/fuji-x-t3/fuji-x-t3THMB.HTM
  5. built color profiles out of the 3 raws. I've set the WB first, then I've generated a calibration profile using color calibration module, no other changes.
  6. now I've used these 3 profiles on photos of my wife and it proves to be very consistent! All 3 give much more pleasing tangerine-hotdog like skin! This is surprising given the fact that all 3 profiles are done using different cameras and different lighting!
github-actions[bot] commented 2 years ago

This issue did not get any activity in the past 60 days and will be closed in 365 days if no update occurs. Please check if the master branch has fixed it and report again or close the issue.

kmilos commented 1 year ago

This can be closed now? @Nilvus

wilecoyote2015 commented 1 year ago

@kmilos @LebedevRI Hmm, I h ave tested with current master and still, the color rendition differs greatly from the results of the calibration with a chart and also from the color rendition achieved with other cameras (for example, see the sample raws from imaging-resource or dpreview).

I have profiled my X-T3 with a Datacolor Spyder Checkr 24 Chart according to the following article (point D): https://ninedegreesbelow.com/photography/articles.html#profile-digital-camera The resulting profile is attached: icc_t3.zip

Below is a comparison of the standard dt input matrix (top) and my custom profile, with only exposure, crop, rotate and perspective and lens correction enabled: _DSF0824_dt _DSF0824_custom

kmilos commented 1 year ago

Not much to be done TBH apart from custom profiling like this or corrections in the color calibration module - the same sensor is in X-T30, X-T30 II, X-T4, X-E4 and X-S10, w/ identical color matrix for all (from Adobe DNG converter, like we do for any camera).

Mangus78 commented 1 year ago

Not much to be done TBH apart from custom profiling like this or corrections in the color calibration module - the same sensor is in X-T30, X-T30 II, X-T4, X-E4 and X-S10, w/ identical color matrix for all (from Adobe DNG converter, like we do for any camera).

Would be useful for other X-T3 owner to use the icc profile made and shared by @wilecoyote2015 ?

The same problem affects the X-T5 in your knowledge?