PixInsight Forum (historical)

PixInsight => Bug Reports => Topic started by: georg.viehoever on 2009 April 11 09:58:46

Title: ABE/History explorer forget mask
Post by: georg.viehoever on 2009 April 11 09:58:46

Hi,

I am setting a mask before applying an ABE. The history explorer does not record that ABE is applied with mask. Actually, i think the real bug here is that ABE does not honor masks at all, even if you apply the model to the masked image.

Georg

Title: ABE/History explorer forget mask
Post by: Juan Conejero on 2009 April 13 11:51:51

Hi Georg,

ABE is not a maskable process, so this is normal behavior.

However, it's true that there's no (easy) way to know whether a particular process is maskable or not, and this is indeed a problem. Version 1.5 makes these things clearer (I hope) thanks to a redesigned Process Explorer.

Title: ABE/History explorer forget mask
Post by: georg.viehoever on 2009 April 13 12:04:49

Why not enable masking for ABE, at least for the "pixelmath" part of it? Would be more consistent.

Georg

Title: ABE/History explorer forget mask
Post by: Juan Conejero on 2009 April 13 12:18:27

Hi Georg,

Without having looked at ABE's source code for a long time (which makes any assertion risky), I agree with you. This process was first designed and implemented by Carlos Milovic back in 2005, as a command-line utility using an early version of the PCL. We implemented it as a PixInsight module in 2006, and if I can remember well, the last revision was made in 2007. That means that this tool isn't up-to-date with respect to most processing tools.

I'll try to update ABE as soon as possible, and I see no a priori problem to implement masking for the correction phase of the process. If correctly applied, ABE is a very efficient tool for gradient removal in relatively simple cases; it definitely deserves a complete revision.