Dealing with clipped highlights - an example

[Morgan_Hardwood]

In reference to Building Filmulator - #12 by Jud_Johnson
Using this raw file from @Jud_Johnson: 009.CR2 (21.0 MB)

@Jud_Johnson

I’ve tried RawTherapee, Photivo, Darktable, Photoflow etc. They all work, but it’s a constant fight over light and dark global adjustments. With Lightroom, it’s pull down the Highlight slider - that’s it. Then I use your software. It’s pull down the Highlight slider - that’s it.

In RawTherapee it’s enable highlight reconstruction using color propagation, then push up the highlights slider.

That isn’t something you need to remember, it’s something you see by looking at the preview and histogram and respond to accordingly. Let’s go step by step:

1. Open the photo, apply the neutral profile (which you normally wouldn’t do because you’d have your own default profile which makes photos from your camera look good straight away the way you want them, but I use neutral in explanations), and notice that the corner is blown out:

2. Drag up the highlight compression slider:

http://rawpedia.rawtherapee.com/Exposure#Highlight_Compression
3. The histogram and the preview both tell you that at least one of the channels is clipped. It is clipped in the raw file, so it doesn’t matter which raw program you use - Lightroom, UFRaw, Filmulator, darktable - they all have to deal with the missing data somehow, and RawTherapee lets you see this and decide how to deal with it. So you enable highlight reconstruction and the histogram and preview show that now data has been made up for the clipped channel:

http://rawpedia.rawtherapee.com/Exposure#Highlight_Reconstruction
4. There are four methods of reconstructing highlights, because no method in any program is perfect. You change the highlight reconstruction method to “color propagation” because it looks best:

That’s the sort of thing you could include in your default processing profile for that camera. The other things you would include are a default tone curve and some way of dealing with saturation, my preferred method is using the Lab* CC curve:
http://rawpedia.rawtherapee.com/Lab_Adjustments#CC_Curve
http://rawpedia.rawtherapee.com/Creating_processing_profiles_for_general_use

Here are three versions of the photo which deal with the dynamic range in various ways:

1. Using Shadows & Highlights
   009 rt morgan sh.jpg.out.pp3 (9.4 KB)
   

2. Using Tonemapping
   009 rt morgan tm.jpg.out.pp3 (9.4 KB)
   

3. Using Tonemapping in CIECAM02 mode
   (slight difference, basically the same thing - CIECAM02 curves behave differently to other curves, but I haven’t used them in this example)
   009 rt morgan tm ciecam02.jpg.out.pp3 (9.4 KB)
   

Each raw processing program will handle things differently and assume different defaults and be more or less automated giving you more or less control over what happens. Choose one that suits your character.

[LZ1]

Sometimes it could be useful to pull down raw white point (in addition).

[Andrius]

This actually did the trick for me! (In darktable I had to pull it right (high) not down)
To be honestly I did not notice any changes on the picture but the over exposure indicator has gone away.

[Morgan_Hardwood]

@LZ1 could you show (in a new topic, not this one) when pulling down the raw white point is better than using the correct approach (assuming that the raw format you’re using has correct white levels)?

[CarVac]

When highlight reconstruction gives false colors, lowering the white point can help.

It’s used in addition to, not instead of, highlight reconstruction and the other techniques you listed.

[LZ1]

Here - Dealing with clipped highlights — using raw white point

[LZ1]

Exactly