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

Stack Count Lesson From Jupiter and a T3i

planet Maksutov dslr imaging
  • Please log in to reply
8 replies to this topic

#1 BQ Octantis

BQ Octantis

    Apollo

  • *****
  • topic starter
  • Posts: 1498
  • Joined: 29 Apr 2017
  • Loc: Outback NT, Australia

Posted 24 August 2019 - 07:55 PM

G'day all,

 

I've been able to improve my planetary imaging significantly this year. The first breakthrough was learning that I could image for 200 seconds without smudging from Jupiter's rotation. This is important to me because AstroDSLR can only pump out LiveView .mp4s at 8.6 fps—so my available frames for a non-derotated stack went from 860 to 1720. So the next question was how many bad seeing frames I could afford to throw away. I concluded that 1024 frames is a good stack target to produce a good result for a final image at 4.45 pixels per airy disk. I just produced a visual illustration of it yesterday:

 

(Click for full size.)

1-1024_RGB_decimated_wavelets.jpg

 

This helps with final stack count targets—with derotation of 3-4 frames (3072-4098 frames), I can even get smooth results at 6 ppAd. This is helpful as Jupiter gets farther and farther away…

 

Cheers!

 

BQ

 

 


  • zjc26138, Starman27, kbev and 6 others like this

#2 petert913

petert913

    Gemini

  • *****
  • Posts: 3228
  • Joined: 27 May 2013
  • Loc: Portland, OR

Posted 24 August 2019 - 09:01 PM

Great example of the benefits of stacking !



#3 RedLionNJ

RedLionNJ

    Soyuz

  • *****
  • Posts: 3661
  • Joined: 29 Dec 2009
  • Loc: Red Lion, NJ, USA

Posted 24 August 2019 - 09:52 PM

You might want to consider some kind of change in your regime if you're getting such a slow frame rate.

 

My (older) T2i gets between 25 and 30 fps with BackyardEOS.  I'm told it could get even more with Magic Lantern (which I already have, but don't usually boot from it).



#4 BQ Octantis

BQ Octantis

    Apollo

  • *****
  • topic starter
  • Posts: 1498
  • Joined: 29 Apr 2017
  • Loc: Outback NT, Australia

Posted 25 August 2019 - 03:37 AM

You might want to consider some kind of change in your regime if you're getting such a slow frame rate.

 

My (older) T2i gets between 25 and 30 fps with BackyardEOS.  I'm told it could get even more with Magic Lantern (which I already have, but don't usually boot from it).

 

I've certainly thought about it. I've tried the following:

  • Sent AstroDSLR output to a RAM disk → no change in frame rate.
  • Upgraded AstroDSLR from v1.3 to 3.10 → took my frame rate down to 5.1 fps.
  • Recorded on-camera at 3× zoom → 10× larger files, less total frames recognized by Lynkeos, much lower quality output.
  • Recorded with my wife's new MacBook Air → no change in frame rate.

When my MacBook Pro finally dies, I might consider getting a Windows laptop and BYEOS for the 2-3× boost in Live View fps. But Andrew (Tulloch) in Solar System Imaging and Processing has demonstrated the limits of 20 fps, and I don't think I'm that far from the absolute limit of a 7-inch aperture even at 8.6 fps…

 

BQ


Edited by BQ Octantis, 25 August 2019 - 03:43 AM.

  • RedLionNJ likes this

#5 BQ Octantis

BQ Octantis

    Apollo

  • *****
  • topic starter
  • Posts: 1498
  • Joined: 29 Apr 2017
  • Loc: Outback NT, Australia

Posted 25 August 2019 - 03:50 AM

Great example of the benefits of stacking !

Exactly! Most planetary imagers refer to the fraction of frames they kept or rejected, but this didn't help me understand what my target number of frames should be (unlike with DSOs, in spite of SNR being SNR in all domains). So this season I decided to figure it out.

 

The more relevant takeaway for me was setting a reference point for imaging other planets. I actually now shim my 18mm eyepiece to get Saturn's SNR at the sensor to be roughly (±10%) the same as Jupiter's so I could stack the same number of frames. This let me resolve the polar hexagon. At 8.6 fps.

 

BQ


Edited by BQ Octantis, 25 August 2019 - 03:56 AM.


#6 Tulloch

Tulloch

    Messenger

  • -----
  • Posts: 486
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 26 August 2019 - 09:25 PM

I've certainly thought about it. I've tried the following:

  • Sent AstroDSLR output to a RAM disk → no change in frame rate.
  • Upgraded AstroDSLR from v1.3 to 3.10 → took my frame rate down to 5.1 fps.
  • Recorded on-camera at 3× zoom → 10× larger files, less total frames recognized by Lynkeos, much lower quality output.
  • Recorded with my wife's new MacBook Air → no change in frame rate.

When my MacBook Pro finally dies, I might consider getting a Windows laptop and BYEOS for the 2-3× boost in Live View fps. But Andrew (Tulloch) in Solar System Imaging and Processing has demonstrated the limits of 20 fps, and I don't think I'm that far from the absolute limit of a 7-inch aperture even at 8.6 fps…

 

BQ

Hi BQ, interesting study. Note that the link you gave was to some non-ideal imaging data of mine, but I thought I'd do a similar thing to you for a better set of data I have. This was a 5000 frame stack from my Canon 700D in good seeing conditions, and I went up all the way to 4096 frames - you can probably see the 4096 frame image gets a bit soft as frames of a lower quality were included, but the noise is low. You'll need to click on the image below, this shows the planet at 1x capture size, f26 (or approx 6x pixel size).

 

It seems my sharpening technique is a bit more harsh than yours, so my best quality frames are at about the 2048 mark which tallies with my normal experience of stacking the top 25% of 10000 frames, or 50% of 5000 frames to get the best result (seeing dependent of course).

 

My early results from the ASI224 seem to indicate I need to stack at least double this number to reduce the noise from this planetary camera - but at 200 fps this isn't a huge problem lol.gif .

 

Andrew

Attached Thumbnails

  • Jupiter_Tv120s_200iso_1024x688_20190703-20h37m02s-loop01_pipp all frames ps1sm label.jpg

  • zjc26138 and Lumix.guy like this

#7 BQ Octantis

BQ Octantis

    Apollo

  • *****
  • topic starter
  • Posts: 1498
  • Joined: 29 Apr 2017
  • Loc: Outback NT, Australia

Posted 27 August 2019 - 02:49 AM

Hi BQ, interesting study. Note that the link you gave was to some non-ideal imaging data of mine, but I thought I'd do a similar thing to you for a better set of data I have. This was a 5000 frame stack from my Canon 700D in good seeing conditions, and I went up all the way to 4096 frames - you can probably see the 4096 frame image gets a bit soft as frames of a lower quality were included, but the noise is low. You'll need to click on the image below, this shows the planet at 1x capture size, f26 (or approx 6x pixel size).

 

It seems my sharpening technique is a bit more harsh than yours, so my best quality frames are at about the 2048 mark which tallies with my normal experience of stacking the top 25% of 10000 frames, or 50% of 5000 frames to get the best result (seeing dependent of course).

 

My early results from the ASI224 seem to indicate I need to stack at least double this number to reduce the noise from this planetary camera - but at 200 fps this isn't a huge problem lol.gif .

Outstanding illustration, Andrew! And consistent with my recent Jupiter results, too:

  • There's a tipping point of detail over noise at 256 frames (16x). So I consider this a bare minimum stack in bad seeing.
  • There's a clear tipping point in smoothness vs detail between 256 frames and 1024 frames (32x).
  • There's a slight improvement in smoothness between 1024 frames and 3072 frames (55x).
  • There's a slight loss of detail between 3072 frames and 4096 frames (64x).

So my target is to derotate just two or three 1024 frame stacks, depending on seeing.

 

Thanks for sharing!

 

BQ

 

P.S. No surprise on the planetary cam stacks—at 200 fps, it's a whole different domain!

 

P.P.S. I certainly didn't mean to toss you under the bus!


Edited by BQ Octantis, 27 August 2019 - 03:00 AM.

  • Tulloch likes this

#8 Tulloch

Tulloch

    Messenger

  • -----
  • Posts: 486
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 27 August 2019 - 06:25 PM

Outstanding illustration, Andrew! And consistent with my recent Jupiter results, too:

  • There's a tipping point of detail over noise at 256 frames (16x). So I consider this a bare minimum stack in bad seeing.
  • There's a clear tipping point in smoothness vs detail between 256 frames and 1024 frames (32x).
  • There's a slight improvement in smoothness between 1024 frames and 3072 frames (55x).
  • There's a slight loss of detail between 3072 frames and 4096 frames (64x).

So my target is to derotate just two or three 1024 frame stacks, depending on seeing.

 

Thanks for sharing!

 

BQ

 

P.S. No surprise on the planetary cam stacks—at 200 fps, it's a whole different domain!

 

P.P.S. I certainly didn't mean to toss you under the bus!

Not at all, just that I don't think it was the best data to use for that purpose, that data was more to do with poor seeing than anything else.

 

I was thinking about your observations above, and wanted to mention that I didn't change my wavelet sharpening parameters in Registax to accomodate the different number of frames in the differetn stacks. In fact, it may well be that the only reason why 2048 frames looks the best (to me) is that this is the number of frames I regularly use to sharpen my images. Had I been restricted to 1000 I would have used different parameters, and perhaps this would have been the "sweet spot" in this study.

 

It might be interesting (not to mention time-consuming) to modify the wavelet sharpening parameters to "optimize" the results for each frame stack. Obviously anything less than 64 frames is not worth the effort, but it might be interesting to try.

 

I'll think about do a similar study with the ASI224MC and post the results somewhere.

 

Andrew



#9 BQ Octantis

BQ Octantis

    Apollo

  • *****
  • topic starter
  • Posts: 1498
  • Joined: 29 Apr 2017
  • Loc: Outback NT, Australia

Posted 28 August 2019 - 08:09 AM

I was thinking about your observations above, and wanted to mention that I didn't change my wavelet sharpening parameters in Registax to accomodate the different number of frames in the differetn stacks. In fact, it may well be that the only reason why 2048 frames looks the best (to me) is that this is the number of frames I regularly use to sharpen my images. Had I been restricted to 1000 I would have used different parameters, and perhaps this would have been the "sweet spot" in this study.

Actually, I did accommodate the stack number. I posit that the sweet spot of parameters for a stack is roughly the same across the frames…based more on gut feel than any solid evidence. The only circumstantial evidence I can produce is that when I switched from processing individual frames to using my "filmstrip" method, I did so because the parameters didn't change all that much across stacks in a 15-20 minute imaging session to warrant processing each individually; this also gave me much better uniformity across an entire rotation animation. But in this study, I found that the amount of sharpening a stack can take is proportional to the square root of frames (much like SNR)—so I simply sharpened all the stacks as a filmstrip using the 1024 stack as the reference, and then reduced the transparency of the sharpening layer on each stack to the square root of the stack ratio.


Edited by BQ Octantis, 28 August 2019 - 08:49 AM.

  • RedLionNJ and Tulloch like this


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





Also tagged with one or more of these keywords: planet, Maksutov, dslr, imaging



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics