Improved DBE, Pixelmath History

[georg.viehoever]

Juan,

two suggestions for some small improvements:

- since DBE is usually followed by a substraction from the source image, could you integrate this operation into the DBE dialog? You have a "Target Image Correction" section in ABE, but for some reason you chose not to have it in DBE.
- In the Pixelmath dialog, would it be possible to have something like a history dropdown list? I usually need to repeat the same operations with minor variations again and again. I know this is also possible by storing a process instance on the desktop, but you know: man is a lazy animal...


Kind regards,

Georg

[Nocturnal]

Good ideas!

If I may tack on a more systematic issue with DBE I'd like to have the DBE process be part of the processing history of the image. Right now the only way to save the DBE settings is to drag an instance onto the workspace. If you have to keep going back to DBE to fine tune the settings then this becomes cumbersome. It would be so much easier if you could use the history explorer to reload DBE.

This actually would work really nicely with your proposal Georg. Once DBE includes a target image option this would also be included in the history allowing you to re-do the entire DBE operation in one step.

[Juan Conejero]

Hi Georg and Sander,

- since DBE is usually followed by a substraction from the source image, could you integrate this operation into the DBE dialog? You have a "Target Image Correction" section in ABE, but for some reason you chose not to have it in DBE.

Definitely yes. I'd like to implement it in the next version, if possible. Thanks for the suggestion.

- In the Pixelmath dialog, would it be possible to have something like a history dropdown list? I usually need to repeat the same operations with minor variations again and again. I know this is also possible by storing a process instance on the desktop, but you know: man is a lazy animal...

Actually, not just in the PixelMath interface, but in almost *all* processing interfaces. My idea is to add a list of stored instances by means of two icons (on the interface's control bar): one to store the current instance, and other to browse/edit the list of stored instances. I have this in mind since a long time ago, but I still haven't had the time to think how to implement it well. Maybe now it's time to consider making this step forward.

I'd like to have the DBE process be part of the processing history of the image.

The problem here is that DBE doesn't modify the target image. The DBE instance always goes to the initial state of the generated background model. Indeed, if an integrated correction feature is added, this situation would change

[David Serrano]

- since DBE is usually followed by a substraction from the source image, could you integrate this operation into the DBE dialog?

Definitely yes. I'd like to implement it in the next version, if possible. Thanks for the suggestion.

I'd like to have the DBE process be part of the processing history of the image.

The problem here is that DBE doesn't modify the target image.

Hmm, making DBE perform the subtraction/division would imply that it modifies the image, so there would be no problem with adding it to the image's history. Except that DBE internally spawned an instance of PixelMath... which has its pros and cons of course :?.


About stored instances:

I have this in mind since a long time ago, but I still haven't had the time to think how to implement it well. Maybe now it's time to consider making this step forward.

I was going to answer that I didn't see the problem, but when dedicating some brainpower to it, I've found a couple of difficuties that I'm unable to extensively digress about now. But, in a couple of words:

- Each stored instance would be just like a desktop icon, but invisible. Stored instances would live in an array and the process dialog would link to the array. Thus, a shared array to all instances of a given process.
- Or, each instance would store an array of textual (javascript) representations of its parameters. This way, each instance could store its own set of instances.

Difficult to tell apart whether I'm talking about "process instances" and "process' stored instances". Ask if you're interested and don't get it.

[Nocturnal]

I see now. If I do a pixelmath operation that creates a new image that operation is not added to the source image but rather the new image.

I tend to think of the processing history as containing a record of all that has been done *with*, not just *to* the image. I guess my understanding is not correct but it's a subtlety not immediately obvious to users like myself who are still climbing the PI curve.