FFTregistration, what happened?

[TheMadOne]

Hi,

I have used FFTRegistration in the past to align/register my finished IrRGB image files (fits) for my planetary imaging.  It had always worked very well, select reference image usually the red file & add the other channels.  Select Write registered Images: Correct for Rotation: Correct for scaling & send it on it's way.  In the past, last time I used it for this was in November of 2013 whatever was the most current release of Pixinsight then and it handled the registration flawlessly.  Perfectly aligned frames for assembling the finished IrRGB image.  Just attempted the same thing on some recent images of Jupiter and suddenly I am getting the files all rotated in different directions and completely wild arrangements.  One off in the corner, one off only half of the original image file showing (half moon Jupiter) at the top.  In fact it just seems to jumble the images about at random, no alignment at all.  What happened?!!  Am I missing something here?  Was the Script recently changed with the last 2 releases?  Does it in both the 01.08.01.1083 release and the prior .1079 release.

Any ideals?  Anyone else use FFTRegistration to do planetary color channel alignments?  HELP.....   Please, this was a very effective tool for me in the past, don't know what's happening here.

Thanks,

Mark Jordan
Greenfield, IN
www.thestardeckobservatory.com

[TheMadOne]

Sorry to double post, but here are a couple of .jps of what is going on.  Picture worth a thousand words and all!

[Juan Conejero]

Bug confirmed. This is a regression in the latest PCL version, which affects the core application and hence the JavaScript engine.

Fixing this bug requires a new release of the entire PixInsight platform, so I ask for a bit of patience. I'll try to get this fixed as soon as possible. Sorry for the inconvenience.

[Juan Conejero]

As a temporary workaround while I release a new version, you can disable parallel processing with the following command from the Process Console:

parallel -m-

Do this before running the FFTRegistration script. To re-enable parallel processing:

parallel -m+

[Juan Conejero]

I have the bug identified, fixed and its current  repercussion evaluated. It only affects the FFTRegistration script and the Statistics tool (which reports incorrect coordinates of the maximum sample value). This only happens when parallel processing is enabled (well, obviously it is always enabled so this is no alleviation).

So this bug really has very limited implications. However, this is a serious error in our development tools so I'm going to assign a high priority. Build 1084 of the PixInsight Core application is being built now for all supported platforms.

[TheMadOne]

Thank you ever so much Juan for your extremely quick response and work around for this problem!  Just one of the many reasons I am a major fan of Pixinsight....  It has done great things for my Image processing, and I still have loads and loads to learn with it!

Mark Jordan
Greenfield, IN

www.thestardeckobservatory.com