I have only some in the same place and having the same camera settings. Does it help?
Well, we can try. If there is at least a small completely overexposed region in the raw, it could work. The larger the region the better.
I lost my rock. In pilot training, the joke was that you held a rock in your right hand to remember which way… 
1 Like
Right is where the thumb is left 
2 Likes
There’s no entry for the Canon 400D in camconst.json; dcraw.c has it, and the white value is 3726. That’s what libraw reports in its metadata.
When I open this image in rawproc, all three channels saturate at 4056.
@msdobrescu There is a file camconst.json in your rt installation folder.
Open it with an editor (not with notepad.exe) and insert this lines
{ // Quality C
"make_model": [ "Canon EOS 400D" ],
"ranges": {
"white_max": 4018
}
},
after
// Canon mid-range DSLRs (Rebels)
Edit: hold please. there is something wrong.
Thanks, done, but remember, I’m on Sabayon Linux, will be overridden at update time (this is a rolling release distro).
It will be included in dev branch of rt
Ah, learning in progress: what did you do to arrive at that number?
I basically followed this instructions: RawTherapee/rtengine/camconst.json at dev · Beep6581/RawTherapee · GitHub
@msdobrescu As the entry above was wrong, here’s the fixed one
{ // Quality C
"make_model": [ "Canon EOS 400D DIGITAL" ],
"ranges": {
"white": 4018
}
},
Thank you so much, this improves things a lot!
New white level pushed to dev
Edit: I used the more progressive wl by @ggbutcher because it worked fine
2 Likes
What did we learn from this overeposed example?
- on an overexposed raw file, the correct white level is important (correct wl is up to the software, not to the photographer)
- on an overexposed raw file, there is still a lot of details to recover
7 Likes
Not knowing the innards of raw white point, I poked around to this result, for what it is worth.
P1000015.RW2.pp3 (11.5 KB)
I confirm! For me, RT is the thing to go with! Only if Hugin would be as easily capable!