I am using RT to process raw files from Canon 5D mark IV. I am using settings to get a linear TIFF file as the result.
The file is part of an HDR sequence and the area that turns pink is very overexposed - but that should push it to white, not pink.
I feel like this is probably a setting somewhere that I have set wrong.
Here is a small crop of the TIF file
_04A5720 crop.tif (3.1 MB)
Here is the original CR2
_04A5720.CR2 (28.1 MB)
Nathan
Can you also share the corresponding pp3 file please? I have a good idea what is going on, but I want to make sure.
Also, which version of RawTherpaee are you using?
The pp3 sidecar would help.
I opened your CR2 file in RT 5.8, and if I turn exposure compensation down, I get pink/magenta in the sky. But If I turn on Highlight reconstruction and choose Blend mode, then the pink/magenta color goes away.
Oooh, a good example of the implications of blown highlights…
When there’s just too much light at a pixel for it to handle, the recorded measurement is just the max number the sensor can produce. Your image has a lot of those, and when the image was captured, all those pixels just maxed out at (more or less) the same number, which could be considered as “white”.
However, that data needs to be whitebalanced, each channel’s data multiplied by a different number to actually make white White. What that does to all of those saturated channels is to shift them away from each other, making the pixel “not-white”. Here’s your image, demosaiced and white balanced, but still linear:
Look at the histogram, and note the separated saturated channels. That’s what makes those pixels “look pink”…
There are a couple of ways to handle that. One is to just set the white point at the green spike, and everything previously pink wouild then go white. I call that “Highlights Abandonment”. The better way to handle it is to use a Highlight Reconstruction operator to use the data between the green and red spikes to make some semblance of definition in those pixels.
Maybe more than you wanted to know… 
Here is the pp3 file. The version is 5.8
_04A5720.CR2.pp3 (11.9 KB)
I would like an automatic way to deal with this. So is Highlight Reconstruction a setting for RT?
Highlight reconstruction is just above Exposure compensation:
See how it easily makes the blown highlights white.
By the way, I would also check “Auto-matched camera profile” in Color Management instead of “Camera standard”.
@nathanm Please enable this checkbox again, because you should not have unticked it in the first place:
Wow, is that it? Thanks so much for the help!
Blend mode is very good for too bright sky. But it is not always the best choice: “Color Probpgation” does a much better job on overexposed areas of faces: if the areas are too light and eroded, the natural color tones of the skin are reconstructed much better. That’s the beauty of RT, that you have a good choice for every purpose.
micha
Yes but be careful, sometimes color propagation leaves some pink highlights.
But in your original post, you said that the image belongs to an HDR sequence, so why do you need to deal with highlights recovery? Don’t you have some images in the sequence that are properly exposed for the sky?
I don’t know exactly what you mean by my “original post”. Maybe you are mistaking me for someone else. I just want to say that “Color Probpgation” does a much better job on faces and skin tones.
micha
Sorry I thought I was answering to the OP! 
I’ve been flummoxed by that, too. The forum software apparently expects the next post to be an answer to immediately prior post, so it doesn’t show it as a reply to a specific person. Like we’re having conversations here… 
It doesn’t.
There is a reply button in the bottom right of each reply on a post. Clicking that should let you reply to a person. If that doesn’t work, then quote what you’re replying to, or @ mention the person, @ggbutcher so they get a notification 
I have been reading all the posts and I assumed that the reply was for me.
To reply - yes I do have parts of an HDR sequence that have the sky properly exposed, but it is unclear that HDR processing algorithms would treat the properly. I figured I had a setting wrong in RT - and it turns out that I did.
