Dithering Applied to Groups of Images?

[Ronald Crouch]

As a part of my search for improved cadence with modest 60sec subs I am excited to see the ability to dither as part of a loop rather than for each subframe is sitting there for me to use! This awesome. Why does it matter? Consider a typical request for 6 hours of rub subs with 60s exposures. That is a total of 360 subframes. I see with an excellent Mach 1 mount that guided dither with 1sec guide exposures seems to take 20 to 30 sec either with Maxim or Phd2. So th total overhead is 150minutes added to the desired 6 hours!

if instead the rib are repeated in a loop with 3 subs for each filter and dither happens only once after each loop the total time spend dithering becomes 17 minutes rather than 150 minutes. With 120 subs per filter total dithering each 1/3rd of the time does nothing to data quality.

I think it would be a great addition to support setting up such a sequence for targets built with the Scheduler web interface. Just better leverage what we already have! At the same time allowing a plate solve position check/correction to be evoked on the same time as the looped dither would also improve overall results. Fortunately my mount tracking is excellent so though desirable the plate solve can be dropped.

If I use unguided dither instead the time savings for the entire 6 hours of data goes from about 72minutes dither overhead to 8 minutes. These are really significant time savings with clear skies being pretty precious.
best to all,
Ron

[Bob Denny]

As a part of my search for improved cadence with modest 60sec subs I am excited to see the ability to dither as part of a loop rather than for each subframe is sitting there for me to use!

You may be misunderstanding the application of dithering. In live ACP sequences and also those coming from the scheduler, dithering is applied to each image.

Consider a typical request for 6 hours of rub subs with 60s exposures. That is a total of 360 subframes. I see with an excellent Mach 1 mount that guided dither with 1sec guide exposures seems to take 20 to 30 sec either with Maxim or Phd2. So the total overhead is 150minutes added to the desired 6 hours

I should point out that it's not MaxIm that is calculating and performing the dither, it is ACP.

Also cutting down on dithers will reduce the effectiveness of dithering when combining. It does cut down on the data quality. The effect of suppressing hot and cold pixels is reduced in proportion. I will chat about this with Doug George, whose opinion I value. I may also chat with Adam Block on this. I have a feeling that group dithering this might be a "free lunch" idea.

The Astro Imaging Channel just had a program All About Dithering that I found interesting. In particular the idea of doing direct dither was covered. ACP does it with tiny slews. However I think it's worth looking at doing it with a different call PulseGuide(). Is that going to be faster than SlewToCoordiantes()? Probably, and maybe a win. The big question is this: Is your mount going to provide you with good enough tracking to do 60 second unguided images? If not then direct dither unguided may be a loser. One of Doug George's "principles" is "More money gets you more data per unit time". So yeah if you have a perfect (expensive) mount, you can do 60 sec unguided without any star shape lossage, and use direct dither.

Since this is a general astronomy engineering discussion (at this stage) I'm going to move this thread to the General Astronomy section. if we end up with a specific wish list item (like PulseGuide dithering), I'll take it back to ACP and file a Gemini ticket for it.

[Ronald Crouch]

Hi Bob,
Thank you for the thoughtful response. I assure you I do under dither and the ramifications of not dithering effectively. I cannot tell 60sec guided from 60sec unguided subs as the mount with apcc mounting model is quite good. No subs lost to guiding either way. I like the idea of pulse guide dither rather than micro slews.

If I group acquisition to be one sub each with sequential lrgb the dither the effectiveness will be identical as if dither was done on each filter individually. Loop on that to build data. That is the central idea. With 100 subs dithering every 3 is the sort of common application being used.

just talking and thinking about this is a great thing.

best,
Ron

[Bob Denny]

OK let me research the PulseGuided dither. However I may need to chat with you to understand what you're saying in the second paragraph.

ACP-1945 - Unguided Dither using PulseGuide() Instead of Micro-Slews

[Ronald Crouch]

Hi Bob call most anytime!
consider this sort of thing..
loop 50 times
1 Ha sub
1 S2 sub
1 O3 sub
dither
end loop

This would dither only once rather than 3 times to get exactly the same dither performance for each set of filter data.

With 50 subs there are so many that something like this would not changes things at all in the result.

loop 50 times
2 Ha subs
2 S2 subs
2 O3 sub
dither
end loop.

This time dither is reduced to 1/6th of the overhead.

with CMOS 100+ subs is very common so looping something like this to get 300 subs..
loop 100 times
3 L subs
3 G subs
3 R subs
3 B subs
dither
end loop

... creates excellent data with dither time reduced a factor of 12.

These are the sort of constructs that are very commonly used. Obviously there is more intelligence regarding the looping.. My examples are just cartoons.

Best,

Ron

[Bob Denny]

I understood that, but how do you do this with ACP? Dithering is applied before each image. Maybe you misunderstood how #dither is applied. I'll do some research on this.

[Ronald Crouch]

I’ve done staggered dither many times. No complaints about results. One good thing is for the last few years automatic cosmetic correction in PixinSight tends to kill stray hot pixels even without dither but that is no excuse to not dither effectively! When I get a good clear reasonably moon free night I’ll just ask Nina to run three 60 sub L frames. One with dither each sub, one with dither every 3 subs, and one with no dither at all.

Then process all three both with and without cosmetic corrections. We would have 6 images to pixel peep!. Does this sound like a good idea to you?

Best,
Ron

[Bob Denny]

OK, meanwhile I am looking into PulseGuide dithering. Are you doing guided or unguided dithering (in ACP I don't care about NINA)?