I had thought raw overexposure indicator is “absolute”, and it shows the pixels clipped in the sensor. Today I noticed that if I change white balance also the indicated areas change on the picture. Below are examples at 2000K and 25000K. At 2000K the white wall illuminated with blue light is shown as clipped, but with 25000K not.
A comment in the code explaining the rationale behind this:
// the clipping is detected as >1.0 after white level normalization
/*
* yes, technically, sensor clipping needs to be detected not accounting
* for white balance.
*
* but we are not after technical sensor clipping.
*
* pick some image that is overexposed, disable highlight clipping, apply
* negative exposure compensation. you'll see magenta highlight.
* if comment-out that ^ wb division, the module would not mark that
* area with magenta highlights as clipped, because technically
* the channels are not clipped, even though the colour is wrong.
*
* but we do want to see those magenta highlights marked...
*/
[edit: I don’t find this so convincing if we consider an unbounded linear scene-referred workflow though… it only makes sense if the modules that follow are unable to cope with scene-referred values greater than 1.0…]
That’s also my thinking. I’m working in scene-referred workflow, so I want to see the absolute clipping, not the computed > 1.0 clipping.
Turn off WB module. After that you can check what is clipped. Visualizing the raw (sensor) highlight clipping | darktable
Compare with for example Hraw Releases · lightful/hraw · GitHub
Drag the raw file onto revelator.
Hraw is else a tool to measure dynamic range from raw files https://youtu.be/N4BZkqEq_H8
I have created a feature request in GitHub to show the “real” raw clipping. Feature Request: Toggle in Raw Over Exposure Indicator to show "real" clipped pixels · Issue #6596 · darktable-org/darktable · GitHub
I rather agree with the feature request. Yes, I didn’t pursue this in #4587, but having looked at the above Visualizing the raw (sensor) highlight clipping | darktable
I don’t find this very convincing, indeed some of it seems wrong. And relating values to a max of 1 seems inappropriate for scene workflow as has been said.
How about a new extra histogram function which really shows you what’s happening at the top end. The x-axis buckets would cover whitepoint - 500 to whitepoint + 200, say, (whitepoint as in white/black point module). With a trace for R, G & B. The buckets would be actual raw values, no scaling, no adjustments at all. We’d get to better understand what’s coming out of our cameras (if interested!).
+1 for that suggestion. Back in my old days with Windows, I use constantly RAWdIgger, a very nice tool that shows graphically the RAW data (histo) and provides stats also.