Highlight reconstruction darktable vs. rawtherapee

I’m just guessing here, based on the fact that my renders here, one based on the libraw-supplied matrix whose provenance I do not know, and my SSF profile whose provenance I do know, are virtually identical with respect to the distant mountain color, and more-neutral than the warm render by @Jose_Figueres with darktable. And, my limited experience with Adobe DCPs from the AdobeDNGConverter. Geesh, I may have to snarf their DCP and make a primary set…

LibRaw (and dcraw) matrices are also D65 Adobe DCP ones (for the majority of cameras). No idea where the RT ones come from for the Z 6, which is different from the Z 6_2 one (which, in turn is the Adobe one and identical to Z 6 for dt and LibRaw…)

So any difference between LibRaw and dt is due to some other processing step.

I just happened to have a ICC conversion of the CameraRaw dcp, D65 white point, and it renders the mountainjust a tad warmer than the libraw primaries, but not as warm as @Jose_Figueres’ render. So yes, I’d concur with ‘other processing step’.

I will investigate whether I can answer this question. Although my knowledge is very limited

1 Like

In Darktable setting the WB as Z6 with direct sunlight (5131K)
Captura de pantalla de 2022-08-02 18-08-32

That would explain the color cast - try setting it to what the camera autodetected, or choose a cooler (yellowish/reddish) illuminant or preset manually, sunrise/sunset is more like 3400K, you can even try incandescent…

Anyhow, this is beyond the scope of this HLR subject - please search the forum for related color balancing issues in dt, or open a new one.

According to the Z6 firing parameters the temperature is 5098K. The result is very similar…

Captura de pantalla de 2022-08-02 18-15-32

Captura de pantalla de 2022-08-02 18-17-23

My version using darktable 4.0.0


DSZ_1908.NEF.xmp (14.7 KB)

I left the white balance and color calibration as it was.
Noise reduction could be better, but I still have to learn how to use the module properly.

5 Likes