Author Topic: Something changed in StarAlignment Structure Map?  (Read 1550 times)

Offline steven_usa

  • Member
  • *
  • Posts: 74
    • View Profile
Something changed in StarAlignment Structure Map?
« on: 2017 February 14 16:08:09 »
In using Convolution and Deconvolution, I would use Structure Map in the StarAlignment tool (then Convolute the map, use it as a mask, and then Decon the original image with a PSF).

But the last couple images I've tried processing, I am getting "holes" in the StructureMap.  See attached as an example -- the bright star at the top is filled in.  But the star at the bottom (Mizar) has a hole in the center.

I recently did some updates and my PI installation is reporting: Core Version 1.8 RIPLEY, Version 01.08.04.1195 Ripley (x64).   I don't see any dials on StructureMap settings to help fill this in, is there some other quick way to do so?

Offline steven_usa

  • Member
  • *
  • Posts: 74
    • View Profile
Re: Something changed in StarAlignment Structure Map?
« Reply #1 on: 2017 February 14 17:25:09 »
For reference, here is a link to my integrated (not-yet-stretched) result of Mizar, if that helps.

https://app.box.com/s/jegr7dxvxtkm9aaes2ysyz2xl9vw484d

But again, this has happened in a few of my recent image processing sessions (after a recent update).

Offline Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 6935
    • View Profile
    • http://pixinsight.com/
Re: Something changed in StarAlignment Structure Map?
« Reply #2 on: 2017 February 15 00:59:11 »
Hi Steven,

Nothing has changed in StarAlignment's structure map generation for a long time. Those holes are normal because the stars in this image are large-scale structures. You can fix them by increasing SA's star detection scales.

To facilitate building the mask that you want, you can also reduce the image to one half its current size (with IntegerResample, for example, using a downsample factor of 2). Using StarAlignment for this task may work, but there are different options that can give you more control on the mask. With the downsampled image, you can use MultiscaleMedianTransform with the following parameters:

- 5 layers
- Layers 1, 2, 3 and R disabled

Then you can apply Convolution, STF AutoStretch and HistogramTransformation (by copying STF to HT first, then fine tuning clipping points) to build the mask quite easily. Once you have the mask this way, upsample it to the actual size of the image.
Juan Conejero
PixInsight Development Team
http://pixinsight.com/

Offline steven_usa

  • Member
  • *
  • Posts: 74
    • View Profile
Re: Something changed in StarAlignment Structure Map?
« Reply #3 on: 2017 February 15 04:51:30 »
Thanks, I will try those as soon as I can.   

It was just curious, since in the past it had seemed reliable even for those large structures (like Alnitak or Sirius at .4deg FOV).  But I'm also recently switched from Canon RAWS to ASI1600MC FITS, maybe there is some difference in how they are expressing the data.

Thanks again for the suggestions, looking forward to trying :)