Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Question about WinJuPos - derotate before or after processing each stack

  • Please log in to reply
10 replies to this topic

#1 dcaponeii

dcaponeii

    Mariner 2

  • -----
  • topic starter
  • Posts: 255
  • Joined: 01 Sep 2019

Posted 04 July 2020 - 08:39 AM

Good morning,

 

Been out most of the night with better weather again.  I took a series of images so i could do derotation in WinJuPOS.  I've played around with it a couple of times with older images on my drive.  My question is:  Which is better to derotate the stacked images and then do any image processing on the final derotated stack?  OR Do processing on each stack and then derotate the processed images?

 

I guess I'm also wondering if you wait and do wavelets on the derotated stack as well or do wavelets before performing derotation.

 

Inquirying minds want to know.

 

Thanks

 

Don

Waxahachie, TX


Edited by dcaponeii, 04 July 2020 - 09:00 AM.


#2 jesco_t

jesco_t

    Sputnik

  • -----
  • Posts: 34
  • Joined: 23 Jan 2020
  • Loc: Hamburg, Germany

Posted 04 July 2020 - 10:27 AM

I do a third variant:

I do mild wavelet sharpening with very little to no noise reduction for each video sequence. Then I derotate and stack the resulting images in WinJUPOS. With that single image, I do the final sharpening, noise reduction and color processing.

 

As far as I understand, derotation is a very clever variant of „shift & add“, so derotating several images should reduce noise similar to regular stacking. At least, I hope so... still new to this. wink.gif 
 

Not sure if that is ultimately the best way, but so far I achieve the best results with it.


Edited by jesco_t, 04 July 2020 - 10:34 AM.


#3 dcaponeii

dcaponeii

    Mariner 2

  • -----
  • topic starter
  • Posts: 255
  • Joined: 01 Sep 2019

Posted 04 July 2020 - 10:57 AM

I do a third variant:

I do mild wavelet sharpening with very little to no noise reduction for each video sequence. Then I derotate and stack the resulting images in WinJUPOS. With that single image, I do the final sharpening, noise reduction and color processing.

 

As far as I understand, derotation is a very clever variant of „shift & add“, so derotating several images should reduce noise similar to regular stacking. At least, I hope so... still new to this. wink.gif 
 

Not sure if that is ultimately the best way, but so far I achieve the best results with it.

Thank you.  I'll get to it later today.  I shot entirely too many frames overnight.  It's taking awhile.  Her's one that I finished up in a quick run through of Jupiter.  3x drizzle reduced back to 50% of capture size (my posted images keep coming out huge and I'm hoping this makes the posted image closer to reality.  This is a 50% stack of a 15000 frame capture.  Gain = 250 12ms exposure 120s capture.  High-pass sharpen in post processing and a bit of a contrast bump.

 

2020-07-04-0853_3-L-Jup_3xAS_50675w.jpg


  • roelb, Tulloch and jesco_t like this

#4 DMach

DMach

    Surveyor 1

  • -----
  • Posts: 1,509
  • Joined: 21 Nov 2017
  • Loc: The most light-polluted country in the world :(

Posted 16 July 2020 - 03:36 AM

Interesting that there hasn't been more discussion on this post.

 

I have always applied wavelet sharpening to the stacked image (erring on the side of under-sharpening - you can always add a little more later, but you can't take it away) and then de-rotating the results.

 

But out of curiosity, I recently tried de-rotating the unsharpened image stacks. This seem to work quite well for Jupiter, and it would mean I could apply sharpening one time to the final result. It does seem to result in some edge artefacts however, so I'm still in two minds as to the best approach.

 

Those edge artefacts became a real problem when I tried that method on Saturn, due to the rings.

 

So overall, I think I'd still lean towards stack, sharpen then de-rotate.


  • yock1960 likes this

#5 happylimpet

happylimpet

    Aurora

  • *****
  • Posts: 4,822
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 16 July 2020 - 03:46 AM

I do a third variant:

I do mild wavelet sharpening with very little to no noise reduction for each video sequence. Then I derotate and stack the resulting images in WinJUPOS. With that single image, I do the final sharpening, noise reduction and color processing.

 

As far as I understand, derotation is a very clever variant of „shift & add“, so derotating several images should reduce noise similar to regular stacking. At least, I hope so... still new to this. wink.gif 
 

Not sure if that is ultimately the best way, but so far I achieve the best results with it.

This is exactly how I do it.  For example, as we speak*, i am taking the 9% drizzle 1.5x stacks of Mars, copying them into another directory, running all of them through identical registax wavelets (with minimal noise reduction, so all show a little noise). I then combine these in Winjupos, and then do final deconvolution and noise reduction. In fact I think this is fairly standard.

 

 

But out of curiosity, I recently tried de-rotating the unsharpened image stacks. This seem to work quite well for Jupiter, and it would mean I could apply sharpening one time to the final result. It does seem to result in some edge artefacts however, so I'm still in two minds as to the best approach.

 

Those edge artefacts became a real problem when I tried that method on Saturn, due to the rings.

 

And this is one very good reason why i dont derotate un-waveletted stacks!

 

Also, doing a first pass of sharpening allows one to throw out the crap and just include the best runs.

 

EDIT: *UNLESS YOU SPEAK TO MY BOSS, IN WHICH CASE IN WORKING ON THAT REPORT.


Edited by happylimpet, 16 July 2020 - 03:47 AM.


#6 DMach

DMach

    Surveyor 1

  • -----
  • Posts: 1,509
  • Joined: 21 Nov 2017
  • Loc: The most light-polluted country in the world :(

Posted 16 July 2020 - 08:17 AM

This is exactly how I do it. For example, as we speak*, i am taking the 9% drizzle 1.5x stacks of Mars, copying them into another directory, running all of them through identical registax wavelets (with minimal noise reduction, so all show a little noise). I then combine these in Winjupos, and then do final deconvolution and noise reduction. In fact I think this is fairly standard.

And this is one very good reason why i dont derotate un-waveletted stacks!

Also, doing a first pass of sharpening allows one to throw out the crap and just include the best runs.

EDIT: *UNLESS YOU SPEAK TO MY BOSS, IN WHICH CASE IN WORKING ON THAT REPORT.


Yes, I did forget to mention that: I realised I still needed to sharpen first to omit any fuzzy results from the de-rotation, so there's no time saving anyway.

#7 dcaponeii

dcaponeii

    Mariner 2

  • -----
  • topic starter
  • Posts: 255
  • Joined: 01 Sep 2019

Posted 16 July 2020 - 09:22 AM

All very helpful discussions.  I'm sorry if I ask off the wall questions but I've been doing a lot of experimenting behind the scenes as to what works better/worse/easier/harder and wanted to see if I'm repeating stuff already well known in the field.  Thanks again for the disucssion


  • happylimpet likes this

#8 DMach

DMach

    Surveyor 1

  • -----
  • Posts: 1,509
  • Joined: 21 Nov 2017
  • Loc: The most light-polluted country in the world :(

Posted 16 July 2020 - 09:11 PM

Always good to question and experiment I think!

#9 KTAZ

KTAZ

    Viking 1

  • *****
  • Posts: 655
  • Joined: 09 Apr 2020
  • Loc: Scottsdale, AZ

Posted 17 July 2020 - 10:49 AM

I am also working through the learning curve. Last night I captured several runs of both Jupiter and Saturn, but have not processed them yet.

 

A few questions:

For Jupiter, I was running 120sec exposures and getting about 40FPS around 350 gain. I had read that anything over 3 minutes would run into rotation issues, so I kept it at 2. I ran 5 AVI's for a total of around 4800 frames per AVI. Are 5 AVI's enough for a decent image?

 

I guess the process that I'm seeing is to stack each AVI, process wavelets, and then stack the final 5 images after de-rotation?

 

Does Saturn even require de-rotation?



#10 RedLionNJ

RedLionNJ

    Soyuz

  • *****
  • Posts: 3,934
  • Joined: 29 Dec 2009
  • Loc: Red Lion, NJ, USA

Posted 17 July 2020 - 12:17 PM

I am also working through the learning curve. Last night I captured several runs of both Jupiter and Saturn, but have not processed them yet.

 

A few questions:

For Jupiter, I was running 120sec exposures and getting about 40FPS around 350 gain. I had read that anything over 3 minutes would run into rotation issues, so I kept it at 2. I ran 5 AVI's for a total of around 4800 frames per AVI. Are 5 AVI's enough for a decent image?

 

I guess the process that I'm seeing is to stack each AVI, process wavelets, and then stack the final 5 images after de-rotation?

 

Does Saturn even require de-rotation?

You didn't specify a particular camera, so the gain of 350 doesn't mean much.  On top of that, only you can be the best judge of whether the signal to noise ratio is sufficient for your liking. It's also very, very seeing-dependent. If you can stack 20% of each AVI, that adds up to around 5,000 frames total. That may or may not be sufficient, depending on the quality of the data itself. It's not quite a "black art" but experience is a huge factor in repeatable success.

 

For derotation purposes, Winjupos works best on pre-sharpened data with a very precise measuring outline. Any variation from this is going to result in a slightly (or greatly) "softened" result.

 

If you want any chance of bringing out minor variations in Saturn's subtle belts, or any polar features, yes, it requires derotation. It's not uncommon to derotate up to half an hour or forty minutes of data to pump the signal to noise ratio to an excellent value.



#11 dcaponeii

dcaponeii

    Mariner 2

  • -----
  • topic starter
  • Posts: 255
  • Joined: 01 Sep 2019

Posted 17 July 2020 - 12:27 PM

I am also working through the learning curve. Last night I captured several runs of both Jupiter and Saturn, but have not processed them yet.

 

A few questions:

For Jupiter, I was running 120sec exposures and getting about 40FPS around 350 gain. I had read that anything over 3 minutes would run into rotation issues, so I kept it at 2. I ran 5 AVI's for a total of around 4800 frames per AVI. Are 5 AVI's enough for a decent image?

 

I guess the process that I'm seeing is to stack each AVI, process wavelets, and then stack the final 5 images after de-rotation?

 

Does Saturn even require de-rotation?

I'm looking at your equipment list in your signature and am wondering why you are only achieving 40 fps.  Are you using USB 3.0 port?  Are you reducing the ROI of the camera and using the cut box in Firecapture to track the planet in an even smaller ROI.  All these help increase frame rate dramatically.  I too was in the 40 - 50 fps range until starting to learn from the folk in this forum.  Last night with my ASI290MC I was achieving 320 fps at 250 Gain (3.2 ms exposure).  My big lightbuilb moment was when Andrew made me realize that although you have to see the planet in the preview window in order to focus, etc. Once it's focused you do NOT have to see much at all in order to keep the object centered in the ROI.  I've been shooting with about a 30% (or less) histogram as a result.  It's made a huge improvement for my images.  I'm still not above average yet but am getting there.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics