A Few of Questions Re: Stacking in AS2

[gnomus]

I've had another go at imaging the moon using my ZWO ASI120 and my C8 scope.  I'm working on the moon before I attempt planetary work.  But ... every night I see Jupiter, and I'm itching to have a go.

I've learned a couple of things (which may be help others about to start out in this venture), but still have a few questions.

The things I have learned:

1) Switch to a 'Lunar' Tracking Rate if your mount does this.  It made life much easier on my second attempt.  

2) Set exposure using the histogram - do not set exposure using the picture on screen.  In the pitch dark it is very deceiving.  If you set exposure using the screen the image will be way too dark indoors the following day.  You will have to lighten the image, and this will make it very noisy (especially noisy in the black sky). 

3) Focus using a Bahtinov mask  on a nearby star

Now for the questions:

At present I am using AS2.  I "analyse" the SER video.  I then Ctrl-click on the resulting graph, such that I only select images above 80% quality.  This has given me a relatively low percentage of images to stack.  Typically I end up with 2-3%.  The best I have had is 8%.  I suspect that I have not had good 'seeing' on either of the nights I have tried to do this: certainly the on-screen image was very 'liquid' in appearance (hence the need for the Bahtinov).  I took the advice that 'ArmyAirForce' gave me in this thread - http://stargazerslounge.com/topic/231139-first-attempt-with-zwo-asi120mm-s-moon-many-questions/ -  and got an IR pass filter.  That is what I used last night.

My questions are:

1)  Is a figure of 80% quality about right?  Is it too stringent, or should I, in fact, be aiming for higher quality than this?  I have been shooting 4000 frame videos so that my 2-8% would give me a reasonable number of frames to stack.

2)  When I move on to planetary, is this still the sort of quality I should be aiming for, or do I lower my threshold for planetary?  I believe that people usually stack over a 1000 frames.  The best I have done with my lunar work at 80% is 242 frames.

3)  I have been selecting a MAP size of 50 in AS2 - does this seem about right?  

4)  Under Quality Estimator, I have selected "Gradient".  Under Analyze I have "Last stack is reference" unchecked, and "Autosize" checked.  Is this correct?

I attach probably the best of last night's bunch for comments.  The previous effort can be seen in the aforementioned thread.  I'm trying not to over-process, but that damned wavelet slider is so tempting ...  "maybe just a few points more.. ", and so on.

[glowingturnip]

i usually do stacks using about 5%, 10%, 15% and 20% of frames, depending on the shape of the quality graph, then choose manually which one I like best (usually one of the lower %ages).  3) and 4) sound right.  My scope is too short a focal length for planetary (750mm) so I've not really had anything I'd call a success.

yep, a tad heavy with the wavelets 

[gnomus]

i usually do stacks using about 5%, 10%, 15% and 20% of frames, depending on the shape of the quality graph, then choose manually which one I like best (usually one of the lower %ages).  3) and 4) sound right.  My scope is too short a focal length for planetary (750mm) so I've not really had anything I'd call a success.

yep, a tad heavy with the wavelets 

Yes.  I don't really know what I'm doing with wavelets.  I tried again, this time using "Linked Wavelets" and being much less aggressive (I hope). Should I be using linked or unlinked wavelets?

[Bizibilder]

These are the settings I use for the Moon with my ED120 and ASI120mm:

I have more or less abandoned wavelets - It's far easier to use Lucy-Richardson deconvolution - Download Astra Image 3.0 SI (its freeware)  http://www.phasespace.com.au/download.html

It is pretty intuitive to use.

PS the graph shows a really rough .avi!!!  I just picked one to run through the program and picked the worst of the whole session!!

[glowingturnip]

Should I be using linked or unlinked wavelets?

dunno i'm afraid, I don't use the registax wavelets, I use pixinsight.

What processing package are you using ?  If you have one that does deconvolution then that usually works really well on the moon, better than the wavelets.  Especially Van Cittert deconvolution.

[gnomus]

These are the settings I use for the Moon with my ED120 and ASI120mm .....

PS the graph shows a really rough .avi!!!  .................

Thank you for posting this screenshot.  Your "really rough" avi looks significantly better than anything I have captured in my 2 (and a bit) sessions so far!!  I think I have not been lucky with the seeing. 

I'll have a go with that deconvolution program.

Regards

Steve

[gnomus]

dunno i'm afraid, I don't use the registax wavelets, I use pixinsight.

What processing package are you using ?  If you have one that does deconvolution then that usually works really well on the moon, better than the wavelets.  Especially Van Cittert deconvolution.

I have a demo version of PixInsight - the licence expires tomorrow, I think!.  I just downloaded, and am trying the Astra Image package recommended by Roger.  Not really sure what I am doing.  Should I be aiming for: a)  bigger PSF sizes and less iterations, or smaller PSF sizes and more iterations.

Can anyone offer some starting "recipes"?

[Bizibilder]

For starters with Astra image try running L-R Deconv using the default parameters - just click preview and watch the little left hand picture.  It will almost certainly be "overcooked" so lower the PSF to 0.8 and try again (just move the preview using the mouse to "reset" it) and again at 0.6 or even 0.4  You should be able to see where the best position is.  All the above with "aggressive" box checked.

Then try unchecking the aggressive box, set the PSF to 1.0 again and try 60-80 iterations and see what you get.

That should get you started but there is plenty of room for experiment.  The good thing is that once you find a good combination it seems to work for a whole batch of images in a mosaic for example.

[gnomus]

For starters with Astra image try running L-R Deconv using the default parameters - just click preview and watch the little left hand picture.  It will almost certainly be "overcooked" so lower the PSF to 0.8 and try again (just move the preview using the mouse to "reset" it) and again at 0.6 or even 0.4  You should be able to see where the best position is.  All the above with "aggressive" box checked.

Then try unchecking the aggressive box, set the PSF to 1.0 again and try 60-80 iterations and see what you get.

That should get you started but there is plenty of room for experiment.  The good thing is that once you find a good combination it seems to work for a whole batch of images in a mosaic for example.

Thanks again.  I got a reasonable result with the Lucy-Richardson 0.6 x 5 iterations "Aggressive".  It was certainly much better than my second image posted in this thread.  There seemed to be more detail in the crater walls top left-ish. (Sorry I don't know the name of the crater with the two peaks within it.)  Couldn't really get anywhere with the Van Clittert - kept getting a fine checkerboard pattern in the Mare.  I'll keep playing around.  And thanks to both of you for the help.  Incidentally, I found a thread here that suggested I had been going about the Wavelet thing all wrong (I had been pushing Wavelet 1, and backing off the others).  There is a Wavelet "recipe" given in the largest post within that thread - http://stargazerslounge.com/topic/12782-wavelet-settings/.  That also gave a better result than my second image in this thread - but I did not think it was as good as the L-R.

[gnomus]

... (Sorry I don't know the name of the crater with the two peaks within it.)  

I think the crater I'm referring to is called Pythagoras/

[ghebert]

Hey guys. I've a novice with telescopes. I recently got an Orion StarBlast 4.5 Equatorial for Christmas from my wife and I'm amazed by it! Here's a photo of my first night!

-Garret-

Sent from my iPad