Gamma Encoding with RAW Processing

[dje]

In particular with ACR/Photoshop

I believe that I have a reasonable understanding of the basic concept of gamma correction – ie where an image has gamma encoding (usually 2.2) applied to it to offset the inverse characteristic of the display device. However my question relates to exactly what happens in ACR and then when it is opened directly from ACR in Photoshop.

My understanding is that adjustments made in ACR are non-destructive and are applied to the linear raw data (after they have been converted to either 8 or 16 bit). This includes tonal adjustments. There is no specific setting or adjustment for gamma encoding in ACR and I assume that it is merely applied to the image for viewing purposes before it is fed to the display.

So what happens when the image is opened in Photoshop ? Does the same approach apply in relation to gamma ? Is the editing still being done on the raw data as modified by the ACR tonal adjustments and the output to the display merely gamma encoded to offset the gamma of the display ? And what happens when the edited image is saved as a PSD ?

For the editing of a jpeg, I assume the situation is different as the gamma encoding would already have been done before compression of the original jpeg file.

I’d appreciate your thoughts.

Dave

[Colin Southern]

Hi Dave,

ACR uses Gamma 2.2 as a starting point, but what a lot of people don't seem to realize is when they make adjustments to things like white balance - exposure - brightness - contrast etc, ACR doesn't apply the white balance adjustment - then the exposure adjustment - then the brightness adjustment - then the contrast adjustment etc ... what it does is rolls ALL of the adjustments in to one BIG adjustment. So one isn't making - say - 20 adjustments that may lose quality as data ranges are pushed and pulled - one is only ever tailoring a single conversion.

If you look at it in that context, gamma 2.2 really only applies to the whole image when there are no other adjustments.

ACR is a parametric editor - not a pixel editor - so it doesn't make changes to pixel data ... it only compiles a "list" of what needs to be done to the file each time it's opened. Once one passes the file through to Photoshop "proper" though, the conversion is locked in place and passed through in the requested colourspace & at the requested bit depth. So by the time the file is passed through a lot of information has been discarded, which is why HDR programs like Photoshop prefer access to DNG files, and actually give you a warning if you try to use 16 bit TIFFs / PSDs instead.

Does that answer the question?

[dje]

Hi Dave,

ACR uses Gamma 2.2 as a starting point, but what a lot of people don't seem to realize is when they make adjustments to things like white balance - exposure - brightness - contrast etc, ACR doesn't apply the white balance adjustment - then the exposure adjustment - then the brightness adjustment - then the contrast adjustment etc ... what it does is rolls ALL of the adjustments in to one BIG adjustment. So one isn't making - say - 20 adjustments that may lose quality as data ranges are pushed and pulled - one is only ever tailoring a single conversion.

If you look at it in that context, gamma 2.2 really only applies to the whole image when there are no other adjustments.

The above makes a lot of sense Colin.

Once one passes the file through to Photoshop "proper" though, the conversion is locked in place and passed through in the requested colourspace & at the requested bit depth. So by the time the file is passed through a lot of information has been discarded, which is why HDR programs like Photoshop prefer access to DNG files, and actually give you a warning if you try to use 16 bit TIFFs / PSDs instead.

I guess it has to be locked in place as PS is not a parametric editor ? But I wouldn't have thought there would be much information discarded if you use a 16 bit file ?

Thanks for your response. It has cleared up some misconceptions on my part.

Dave

[Colin Southern]

But I wouldn't have thought there would be much information discarded if you use a 16 bit file ?

There is because it's no longer in linear gamma - so tonal range compression at one end and expansion at the other has occurred. I might add that one doesn't normally necessarily need that data, but it's lost none the less.

[dje]

There is because it's no longer in linear gamma - so tonal range compression at one end and expansion at the other has occurred. I might add that one doesn't normally necessarily need that data, but it's lost none the less.

Understood, thanks.

Dave