Software for stacking high volume of frames

Hi Jason,

I am just starting to write some astro software and am interested in tasks that can not be solved currently.
If you like you could send me your raw data (perhaps a DVD or something … and I could do some analysis and try to get the best results for you.

Best regards

Martin.

CCDstack works great in 64 bits for over 1000 images for me and provides great calibration.

Thank you all for the repliesreplies! I am currently downloading CCD stack to give it a try, although it is unlikely I will cough up funds in the near future.

Many of the alternate stacking algorithms in DSS have offered me a ridiculously long "producing final image" with an estimated time in the literally hundreds of hours

My best results have been with 100 frames at Adaptive weighted average or with 400-500 frames at Average

In DSS you can first register the images, and then stack them later. This way you should be able to process more frames.

In the 'Register Settings' screen, uncheck 'stack' to first process the images  without stacking them (only registration will happen).
Registered images have all informations displayed in the list: score, angles, etc.

Later you do only the stacking of the already registered images. Unchek 'Register already registered pictures'.



Also, for a large number of frames you should only stack a narrow selection of the best images.
Stacking the best 10% of 5000 images should give you much better results than stacking all the frames, especially if many of the frames are 'less-than-perfect'.
This way you reduce the CPU load for the computer, the processing time, memory usage, etc.

There are many programs available for doing stacking, a few of which have been mentioned already.  Like Martin, when I first started, I tried writing my own image calibration, registration and stacking software.  It did work for my specific FITs files, but I quickly realized that my work was terribly outclassed by existing tools.  My favorite is PixInsight (there is a free 45 day trial).  It covers all of the pre-processing (like stacking) as well as post-processing.

There is a way you can do this with DSS though.  Stack only 200 at a time.  Take the 1st 200, stack them, then save the result as a 32-bit TIFF with the processing changes embedded.  Then do the next stack, etc.  Eventually you will have 25 stacked images.  Now take those 25 images and load them into DSS and stack them.  It is a little time consuming, but should work.

OrionRider's suggestion of only stacking the best frames is excellent.  It reduces the number of frames to stack but you will get more detail in the final result.  I don't know that I would drop down to 10% though.  At 1.2s you aren't getting much signal and with only 500 frames that would be around 10 minutes of exposure time total.  Unless the object is very bright you aren't going to see much.  Doing 80% will allow you to toss out the worst offenders but you will still have 1.33 hours of exposure time in the end.  In this modified version, you would load 250 frames into DSS at a time and tell it to select the best 80%.  That should get you about 200 frames that will actually be stacked.  You should end up with 20 stacked files that you can then stack together.

Regards,
David

Jason,

If you are using PixInsight, is there a reason you aren't using the SubframeSelector script to identify images you want to discard?  You can even weight the results using an arbitrary expression to adjust your stacked results (i.e. targeting FWHM over SNR).

Regards,
David