Aug 7, 2013

PixInsight Deconvolution

 After commenting recently on a thread in the AZ-Observing mailing list about PixInsight it sparked multiple side conversations with other list members asking some questions about deconvolution in PixInsight.  This animation does pretty much sum up the experiences of many including myself when it comes to getting good results.  After much trial and error and gnashing of teeth I have reached a point where I can get some decent results.  The conversations from the mailing list encouraged me to go ahead and put together a video and post on deconvolution in PixInsight.  I'm not saying it's the best way, but it does work and I've enjoyed some good results with it.

World class imager Manuel Jimenez put together a tutorial some time ago that taught me much about PixInsight's implementation of deconvolution.  His tutorial can be found at his website. There are a few points to remember in order to get the best results.

1 - Do deconvolution on a linear image.  Before you stretch.
2 - There's only so much deconvolution can do.  It can't rescue a bad image, but it can make a decent one really "pop".
3 - Use the Dynamic PSF process module.  It's your friend.
4 - Create a star mask for local deringing support in your image.
5 - Create a luminance mask to limit the process to high signal areas.

The Final Result

Like the image in the video I also did deconvolution on the OIII and SII data that I had collected in my attempt to recreate the Pillars of Creation image.  Shown below is the end result.  This was shot with my Explore Scientific ED152 f/8 Apo triplet from southern Arizona.  It's a total of 37 hours of exposure time and processed entirely in PixInsight.  Thanks for watching.  Please contact me if you have any questions, comments or improvements to my process.

Here's a link to a larger version of this image:
M16 - Eagle Nebula in Hubble Palette
37 hours of exposure


  1. Mike,

    Great Tutorial! I

    Thank you for taking the time to put this one together!

    The final image is beautiful!

    Clear Skies


  2. Nice work, Mike! That's pretty much what I do as well. The only additional step I would suggest is sorting the stars in DynamicPSF by Amplitude (remove saturated and dim stars) by Aspect Ratio (remove outliers) by Rotation Angle (use a Circular PSF if these are all over the place otherwise just remove outliers) and by MAD (remove outliers).


  3. Thanks for the video and the writeup! Very clear what the changes to the settings you've made will do.

    I'm curious, do you do DBE after this process or before it?

    1. Thanks Stephen. I don't know what the "official correct" answer is to your question. My own answer - I do DBE before deconvolution.

    2. I did a quick run through of the process with a star field image and did DBE afterwards. Seemed to work OK for me. Since it's still linear, I'm not sure how much it mattered. Maybe it worked better for my image because it made the stars tighter? I'm using a DSLR so I always have slightly fuzzy stars.

      Quick question: in the video, you how how you drag the STF to the histogram tool. Nice trick, I'm stumped on how you did it. What part of the STF tool are you dragging? The triangle for an instance? Also, what part of the histogram tool are you dragging it into? I've tried with my copy and I'm not succeeding.

  4. Thank you for a superb tutorial and for taking the time to do it.

  5. Is there a way to download the video to my hard drive for future reference?

    1. Hi Steven,

      Try this link. I posted it to my Google Drive for a couple of guys in my astronomy club who requested the same thing, and it's still out there:

      PixInsight Deconvolution mp4 download

      Thanks, Mike

    2. Thanks Mike, got it. Is the mask video posted as well? I have a tendency to collect videos and tutorials of anything PI. Comes in handy when trying new processes as well as passing to to future questions on the forum.

      Thanks again,

    3. Hey Steve,

      Have fun - glad you found this stuff useful. It's encouraging.

      PI Basic Luminance Mask tutorial


  6. Hi, Mr. Wiles. Great images you got there. If you don't mind, I'd like to make you a simple question. I know that you've taken pictures of deep-sky objects using an ES 127ED telescope before. I was just wondering here what these pictures would be like if they'd been taken straight from an ES 68° or 82° eyepiece or what kind of raw image we would get from a camcorder put close to the eye relief. Have you ever performed such an experiment?

  7. Great work, That turned out amazing! Thank you for the tutorial