Stained glass windows blown highlights clip to 100% single channel

This looks like a bug, but maybe I’m doing something wrong and I need a pointer in the right direction.

I had this problem with multiple Sony cameras and many versions of RT.
JPEG files look fine, Sony software works fine, the problem is only apparent in RT.

I see the problem mostly with blue, so I’m going to go with this example.
As soon as blue channel clips significantly, the output pixel value is 0,0,100%. However, bordering areas which are just below clipping, so darker in the original, suddenly become pure white.

Take a look at this example:
image

This is the corresponding crop from the in-camera JPEG file:
image

You could say that it’s a color thing and RT knows the intensity, but when switched to black and white, this is what it looks like:
image

I’m stumped. Of course the stained glass window has the blue channel blown out. But the JPEGs look fine and there are pure white pixels bordering the blue area with some of them reminiscent of my first steps in raster graphics in the 1990s when selecting an area of color with too little tolerance and filling it in would result in a border that has pixels just outside of the tolerance, like this:
image

Edit: These are raws from Sony A7R Mark V. I checked some from A7 Mark IV and those have an option to pick “Auto-matched camera profile” and that one results in correctly handled windows.
I previously had this problem with A7IV (in an older version of RT) and earlier with A99II. I figure I need to find a profile for my camera.

But if I don’t have a profile like that, what else can I do?

Did anyone also encounter this problem and solved it somehow?

Thanks in advance!

If you dare using a dev version, there is now a tool Gamut Compression which might help in this case.

You could try the DCPs from Adobe DNG Converter. (Which I personally would prefer instead of having to use Camera standard)

2 Likes

Thanks for the reply!

I’m going to try the suggestions and get back to you.

Would you be able to share such a raw file ?

I could, but it includes a lot of people and I’d have to ask for permission.

I’ll re-shoot a couple of pictures without people in them and post them then.

I solved it by using the bundled ILCE-7RM4 DCP profile and it looks perfect. I get that there may be subtle differences between the Mark IV and Mark V, but the resulting picture is still way better than the Camera Standard profile.

This is good advice. I’ve found that for certain cameras (and in particular small-sensor cameras) these profiles don’t exaggerate the highlights as much as the “Camera standard” or “Auto matched camera profile” in RawTherapee and give a bit more headroom for later processing.

That would be great to investigate signals at those areas in question.

I tried with the Adobe DNG converter profiles and I have to admit I’m blown away by their thoroughness, including the tone curve. I was never quite happy with how RT handled the tone curve with the default settings and I never could quite get what I expected from the picture and with those profiles it’s a much better start for further tweaks.

A huge thank you @apostel338

1 Like

You can also further tweak those profiles for RT with a free Adobe tool called the DNG profile editor… I think Robin Whalley had a nice little video on how to use that …