Another Test Image - Bluing and Colour Saturation

Started 9 months ago | Discussions thread
Flat view
sharkmelley Senior Member • Posts: 1,974
Another Test Image - Bluing and Colour Saturation

Here's another instalment of the occasional series on Photoshop/Lightroom processing.

The story so far is that I have argued Roger's documented processing method (i.e. subtracting light pollution from non-linear data) leads to image bluing with scene intensity and to very noticeable colour saturation.

I documented it here:

Roger argues that he suffers no image bluing e.g. examples here:

I agree that I see no image bluing in those images. So what is going on? I've created a new test image to see if we can find out.

It's a new step wedge with colours from black body radiators at various temperatures: 2000K, 3000K, 4000K, 5500K, 8000K, 12000K, Infinite K.

Each column in the wedge is one-stop less intense than the previous one.

I would expect the colours to come out similar to what we find here: except that I've unilaterally made 5500K to be the camera white point so we at least have a row consisting of shades of grey.

There are two versions - with and without added light pollution. They can be downloaded from here:

Here are previews:

Black Body Wedge without light pollution

Black Body Wedge with light pollution

The first attempt at processing uses Process Version 3 (2012) in Photoshop CC and uses the Black Slider in Photoshop's ACR in the manner that Roger suggests in the section on Raw Conversion here:

Specifically I set Exposure to -0.5 and Blacks to -40. Once the image was open, I used a "Levels" layer to remove the residual light pollution and then applied a colour preserving stretch. Here is the result:

Light Pollution removed using Blacks Slider followed by Levels

Notice how the white row becomes blue as scene intensity falls and even some of the yellow/orange colours turn blue.

The next experiment again uses Process version 3 (2012) but with Blacks Slider set to zero. Once the image is open then all the light pollution is subtracted using a "Levels" layer. Then a colour preserving stretch is applied. There is almost no bluing:

Light Pollution removed without using Blacks slider.

Although there is very little bluing, the colours have still become very saturated.

The third attempt uses a new approach I outlined here:

It uses Process Version 2 (2010) which gives a constant gamma tone curve for Adobe RGB (I forgot to mention I'm processing in Adobe RGB) and the light pollution is subtracted in 32bit mode where it acts on linear data:

ACR Process Version 2 (2010). Light pollution subtracted in 32bit mode

Notice how there is no bluing as scene intensity drops and there are no problems with unnatural colour saturation as scene intensity drops.

So those are my 3 different experiments with Photoshop/ACR. Comments are more than welcome and I'm especially interested in your own attempts on the data.

Does this bring us any closer to understanding why Roger's images do not suffer from bluing? Maybe, maybe not. At this point in time, I don't honestly know.


P.S. If you download and play with one of these DNG files then be aware of a dangerous trap. ACR will store its process settings within the DNG file itself and not in an xmp sidecar file. So next time you open the DNG file, tick "Camera Raw Defaults" in ACR to set ACR back to its defaults, ignoring those previous settings.

-- hide signature --

Takahashi Epsilon 180ED
H-alpha modified Sony A7S

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S +1 more
Flat view
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow