My silly digitization mistake with C-41 in RawTherapee

[loccdor]

Using a Pentax K-1 with 50mm f/2.8 macro lens and pixel shift to digitize film, a raw program is required in the process. I use RawTherapee.

I found that C-41 is a bit of a different animal compared to E-6 or B&W and I struggled to get results where the colors looked right. Specificially, the blue channel would be blown out and oversaturated, even though nothing was showing as clipped in the RGB histogram. No amount of R, G, B curves adjustment in Gimp could fix that.

What fixed it was turning the Tone Mapping curve off in the Exposure settings of RawTherapee. It defaults to a gentle S-shaped contrast curve, if you set it to just be a line, the colors on C-41 come out much better. I also turn off sharpening and reduced the exposure until the highest values are about 10% from the edge of the histogram, or clipping happens on the red channel.

I thought the above problems were happening because I wasn't masking the light source well enough but it turned out to be a software thing.

Anyone have any other tips for settings while converting C-41 raws to tifs in RawTherapee? Even with the above, I found reds can go a little too magenta and need to be adjusted using Hue/Saturation in Gimp.

Thanks.

 

[Steven Lee]

@loccdor The key ingredient to camera scanning is having the right color profile for your camera. In the Adobe ecosystem it means having the right DCP profile. By "right" I mean it needs to be linear and neutral. Essentially you need the colors to be absolutely unmolested prior to inverting them.

Finding a linear profile for a given camera is not that hard, and you already discovered this in Raw Therapee. But finding neutral profiles is harder. Profiles are usually built either by camera manufacturers or RAW converter vendors like Adobe or CaptureOne. All of them aim at "improving" colors to be the most pleasing to a viewer, which means they are not neutral. When inverted, they look like crap.

I have looked at open source RAW converters like Raw Therapee and Darktable, and I did not find neutral profiles for them. Next, I found some folks online who built their own profiles by using IT8 color targets and profile building tools, and I even went that route myself by ordering a set of targets from Wolf Faust. But then I met Andre, the author of Negmaster, who offers neutral profiles for the cameras he supports. Unfortunately, they only work for the Adobe suite. Perhaps there are tools out there that allow Raw Therapee to use a DCP profie?

 

[ant!]

While I am using scanners and scan raw with Vuescan, I use for negative inversion Darktable with negadoctor, see https://www.photrio.com/forum/threads/darktable-and-new-negadoctor-plugin.178807/

Works fine for me, color picker on the film base, white balance for highlights and dark, plus a bit slider wiggling. Some (expired) films need a bit more manual work, others come great just with base settings. Not sure if with camera scanning is more needed?
For b/w I use the same module, but the 3rd tab (was it called print?) to control contrast etc, this one I leave basically untouched for color and work just with the second tab (color settings). The first tab is needed for both (film base, exposure).

 

[loccdor]

Thank you @Steven Lee , this is all new information to me. I did find https://rawpedia.rawtherapee.com/How_to_create_DCP_color_profiles and it does seem pretty complex, I will have to look into it more.

Thanks @ant! , I'll look into this program.

 

[iliks]

@loccdor Perhaps you will find my article useful https://ipalopezhentsev.github.io/p...ware/2020/06/14/inverting-in-rawtherapee.html

 

[loccdor]

@iliks Thank you very much, I'm learning many things I didn't know from your article, and your photography gallery is also beautiful.

 

[loccdor]

@iliks However, I tried installs of RawTherapee 5.9 and 5.10, and neither of them have the Film Base selection under the Film Negative tool in your article:

vs

 

[ant!]

This actually looks quite similar to the darktable interface, so wondering if the underlying algorithms are actually very similar, especially that they are both open source and might use similar libraries...