Applying a process container to an image doesn't seem to have the same effect

[Nocturnal]

Hello,

I'm not sure this is a bug or my misunderstanding so feel free to move the topic if needed. This is my sequence of steps with the latest PI 1.2:

- turn of STF
- load image
- perform DBE
- subtract background and create new image
- do extra processing (curves, SCNR, curves)
- create processing icon for the history of this image
- go back to original image
- subtract background but now modify the target
- drag and drop the saved processing icon on the target
- images are not the same

Before I go through the trouble of uploading screen shots and psm files, should what I did have worked?

The reason I did this is that I wanted to maintain a single history for my image. The history was already broken when I had to save the image separately from the PSM but I didn't want to fragment it any more than I had to.

Alright, I uploaded a snapshot anyway. Since this board can not deal with non-image extensions I'll post a link to my gallery page:

http://gallery.tungstentech.com/main.php?g2_itemId=783

[Juan Conejero]

From the screenshots you've uploaded, I'd say that the only (apparent) difference between both images is in overall brightness (i.e., a simple displacement of the main histogram peak). Otherwise the images seem essentially identical.

A common source of these differences in brightness is in the "rescale" option of PixelMath when subtracting a background model. This is because the rescaling operation is relative to the resulting pixel values after subtraction. If one of the images had different pixel values before applying PixelMath, then the rescaling operation may easily lead to different overall values.

In the sequence of operations that you've described, namely:

- turn of STF
- load image
- perform DBE
- subtract background and create new image
- do extra processing (curves, SCNR, curves)
- create processing icon for the history of this image

I assume the saved psm has a ProcessContainer icon, which includes the DBE and PixelMath (background subtraction) instances before the "extra processing" sequence. Then you:

(1) go back to original image
(2) subtract background but now modify the target
(3) drag and drop the saved processing icon on the target

As I understand this, you're subtracting a background model twice: the first time you do it manually (2), and a second time when you drop the ProcessContainer icon (3). This may be also a cause for the observed differences.

Does what I've said make any sense?