Processing Short Exposure Images

[Stub Mandrel]

This is a thread for those whose subs rarely if ever exceed 2 minutes long, often shorter, with relatively low total exposure time to discuss tips and techniques for image processing.

Whether your an Alt-Az imager battling field rotation, you have issues with polar alignment or you have a relatively small, unguided mount you will often be faced with the challenge of getting the most out of relatively sparse data. The No-EQ Challenge thread has shown that even twenty minutes of sub-minute exposures can give good results.

I think some of the things we could discuss are:

We can't get rid of noise, but how can we keep its effects under control

What are the best tools for short exposure images and what settings are best?

Different types of software and their pros and cons

What percentage of subs should we stack? (e.g. can we use a greater proportion of subs to reduce noise while letting sigma delta keep our stars round?)

Ways of boosting faint background detail

What sort of standard should we be aiming for

And anything else

[mikeyj1]

Really good idea for a thread!

Can I offer a book recommendation?  Astrophotography on the Go: Using short exposures with light mounts  (Patrick Moore Series)

ISBN: 3319098306

a good reference with good info

Mike

 

 

[Stub Mandrel]

Thanks Mike, looks useful, I've looked at the preview pages..

[wimvb]

Great idea to have a thread for short exposure (budget) AP. Here are a few of my experiences with a limited setup:

When imaging with my EQ3 Pro and being limited to short exposures, I found that the "law of diminishing returns" should be thrown overboard. The more subs you can stack, the better quality you will get as far as noise is concerned. Doing 90 x 1 minute exposures is still very little time on target, which can be done in one session.

And as you suggested, keeping even less than perfect subs in the stack can give good result if the stacking process is chosen with care. But obvious faults with tracking should be kept out of the stack. The addage "Garbage in, garbage out" holds true.

A good dithering scheme is all important. I have dithered by hand in -10 C cold weather, just because I knew it would be worth it. Dithering 15 pixels in both RA and DEC is THE single best way to decrease the effect of hot pixels and chroma noise (what Tony Hallas calls colour mottle), and lift detail. Noise reduction during processing can improve an image a lot, but never as much as dithering.

BTW, a simple rule for manual dithering: set the hand controller slewing speed to 1 x sidereal. To get 15 pixels dithering, you then slew the mount for as many seconds as the angular resolution of your setup is. E.g. if you have a 2 arcseconds/pixel resolution, then moving the mount at sidereal speed for 2 seconds will move the mount 2*15 arc seconds = 15 pixels. Keeping the slew rate to max 1 x sidereal, will also avoid backlash problems. (If you put the slewing speed to 0.5 x sidereal, if possible, you need to move the mount for 2 x the resolution in order to get 15 pixel dithering.)

While not photography related, but relevant to the EQ3:

When you use the EQ3 at higher latitude (like my 60 degrees), the weak alt adjustment screws are really stressed to the max. The design of the mount is also such that the alt axis will tighten itself more if you turn the mount upwards (towards higher alt). This is because the bolt that is in the alt axis tightens the fork in the bottom part of the mount. Taking off the black cover plates and slightly loosening the bolt will make it easier to do a polar adjustment. I have also made a small wedge for the alt adjustment screw to rest against. Polar alignment is easier with this wedge in place.

 

Hope this can be of use

[Stub Mandrel]

Excellent advice Wim, that's exactly the sort of thing I was hoping people would post.

[Filroden]

19 hours ago, wimvb said:

And as you suggested, keeping even less than perfect subs in the stack can give good result if the stacking process is chosen with care. But obvious faults with tracking should be kept out of the stack. The addage "Garbage in, garbage out" holds true.

I've found my integrated image can improve when I actually reduce the number of subs. I now try and be quite harsh in what subs I allow through into integration to avoid poorer quality subs softening (at best) the final result. I would always recommend visually inspecting each sub at full screen size and then once again zoomed into a corner of each image. The first pass helps spot things like high clouds passing through (which can add unnecessary gradients into the image), the second pass helps spot subs where there has been trailing (poor tracking, vibrations from wind, etc).

I would never recommend using an arbitrary % of subs. I know this approach works when extracting frames from video for lunar and planetary imaging, but in these cases you can be taking 1000 subs from over 10000. I find % limits will either throw away good subs or include poor ones.

If you have good image analysis software, you could choose a parameter (signal to noise ratio, star eccentricity, FWHM, etc) and make a decision on an acceptable cut off, rejecting subs that fall over that limit. PixInsight also allows for subs to be weighted based on a user formula, so subs with higher SNR or smaller FWHM could be given more weight in the integration process.

As with many things, it is worth experimenting. Integrate everything, integrate everything except certain subs, etc and then compare and see which one provides the best result. Time spend on that initial stage could save a lot of time in post-processing trying to remove gradients or tighten star shapes, etc.

[Stub Mandrel]

I must admit, I do a trawl and get rid of the clearly useless subs (cloud, triangular stars etc.) then use 80 or 90% in DSS, which apparently scores based on number of stars detected (i.e. clarity) and how round they are (eccentricity). I do experiment with different percentages when I hit problems or there's clearly a big variation in subs (especially when thin cloud strengthens the LP and means some subs are very light with few stars).

 

@FilrodenKen - specific question for you - what Ha filter do you have and what length exposure are you using - your results look like the 'next level' that I'm hoping to achieve.

[Filroden]

13 minutes ago, Stub Mandrel said:

@FilrodenKen - specific question for you - what Ha filter do you have and what length exposure are you using - your results look like the 'next level' that I'm hoping to achieve.

I got the Astrodon 3nm. I debated getting a set of Baader narrowband (which would have been much less than the sin,e Astrodon) but eventually decided I was likely to want to upgrade later so why spend twice?

I've only taken a few 30s and 60s subs so far but the difference in quality over my L filter is mind blowing. They are much easier to process to a good mono image though a little harder to then blend with LRGB (especially weak LRGB). You really need good L and R data to not be overwhelmed by the quality and quantity of Ha data. 

[wimvb]

I agree, Ken. Number of subs should not be based on %, but on inspection. Inspection of single subs, but also inspection of final result. Increasing the number of subs will decrease noise in the background. But it can also, as you write, soften the final result. In case of very slight star trailing issues, I have found that the final stacked image can have rounder stars. But you can't expect improvement in terms of small scale detail. As always, experimentation and critical inspection is the best way to go. Sometimes you have to face the fact that a night's imaging only produced garbage that should go straight into the bin.

[Stub Mandrel]

Thanks both.

I must admit I'm surprised that  you can use a narrowband filter down to 30 seconds!

I suppose ~95% of all that nebulosity is getting through, so i shouldn't be!

[Filroden]

4 minutes ago, Stub Mandrel said:

I suppose ~95% of all that nebulosity is getting through, so i shouldn't be

And very little else. I think you would need 60-120s for Oiii and Sii based on Gina's experience with wide field imaging so still within range of short subs! Note, this is with the ZWO ASI1600MM which has very low read noise and works really well with short subs.

Here's a stretched single 60s sub:

[Stub Mandrel]

:-)

Well with my skies, guiding is hardly worth the bother without narrowband, so this convinces me that filter first guiding later is worth a try.