Need explanation on parameters for freqency separation with 16 bits image.

New Here ,
Feb 23, 2021 Feb 23, 2021

Copy link to clipboard

Copied

Parameters of "Apply Image" : blending operation/scale/offest are never explained in tutorials on Frequency Separation

I understand that for image 8bits (luminance R/V/B coded from 0 to 255),  to get high frequencies (structure) we substract low pass image from original, then the result values are from -255 to + 255, so we use a scale of 2 to reduce values to -128 to +127 and then an offset of 128 to get a final 8bits value from 0 to 255.

For 16bits images (luminance R/V/B coded from 0 to 65535) all tutos use without explanation : Addition with inversion of low pass layer, scale 2 , offset 0. I don't understand how it works ?

Furthermore using substract without inversion seems to give the same result (seems to me normal since addition of the inverse is equivalent to a substract),  so what is the reason of this choice ?

Views

62

Likes

translate

Translate

Translate

Report

Report
Community Guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Adobe Employee ,
Feb 23, 2021 Feb 23, 2021

Copy link to clipboard

Copied

You've posted this in Using The Community.
Please let us know which Adobe tool you're using, and a moderator will move your post to the relevant community.

Likes

translate

Translate

Translate

Report

Report
Community Guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
New Here ,
Feb 23, 2021 Feb 23, 2021

Copy link to clipboard

Copied

I use Photoshop 22.2.0 for frequency separation method with  "Image/Applys Image..."

Likes

translate

Translate

Translate

Report

Report
Community Guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines
Adobe Community Professional ,
Feb 23, 2021 Feb 23, 2021

Copy link to clipboard

Copied

For 16 bpc, one should introduce a 128 offset with subtract, which leads to a lossy conversion (for what that may matter).

 

The "standard" 16 bpc method produces a lossless conversion, depending on why one is using frequency separation, this may or may not have a critical affect.

 

Let me dig up a link for you...

 

EDIT: There are 26 pages in the topic thread, here is page 1 –

 

https://www.modelmayhem.com/forums/post/439098

 

Here is my action:

 

https://www.dropbox.com/s/7sf1xcijtzdygyh/Frequency%20Splitter%20v4%20Conditional.zip?dl=0

 

Likes

translate

Translate

Translate

Report

Report
Community Guidelines
Be kind and respectful, give credit to the original source of content, and search for duplicates before posting. Learn more
community guidelines