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

YAMM - Yet another moon mosaic

  • Please log in to reply
13 replies to this topic

#1 sushidelic

sushidelic

    Lift Off

  • -----
  • topic starter
  • Posts: 6
  • Joined: 04 Jun 2019

Posted 13 February 2020 - 05:56 AM

Hello,

 

here's my first "real" moon mosaic, done in 20 panels with a ZWO ASI 183MM Pro, C14 Edge HD and 610nm longpass filter.

It was taken the night of 7th february, two days to full moon.

Seeing was good on that night for my usual conditions here, low frequency turbulence high as usual, but inbetween, some fine detail was visible.

The .ser files took 800GB of my SSD, and Autostakkert run for 36h in batch mode.

I did heavily struggle with grid artifacts after stacking, and there are a lot of things I learned and will do differently next time.

Final result was scaled down to 75%, equaling ~0.17 arcsec per pixel.

 

First some craters (heavily compressed):

 

Aristarchus

aristarchus_graded_compressed.jpg

 

Clavius

clavius_graded_compressed.jpg

 

Gassendi

gassendi_graded_compressed.jpg

 

Pythagoras

pythagoras_graded_compressed.jpg

 

And a (also heavily compressed) preview at forum size:

Moon_2020-02-07_1200px_compressed.jpg

 

Here's the full res:

 

https://live.staticf...84c6ce6f1_o.jpg

 

Hope you enjoy!

Best regards,

Michael


  • D_talley, eros312, deepwoods1 and 6 others like this

#2 Wouter D'hoye

Wouter D'hoye

    Surveyor 1

  • *****
  • Posts: 1,714
  • Joined: 27 Jun 2003
  • Loc: Belgium

Posted 13 February 2020 - 06:21 AM

some outstanding work here. Thanks for sharing.

 

Wouter.



#3 dcornelis

dcornelis

    Mariner 2

  • -----
  • Posts: 282
  • Joined: 09 Oct 2012

Posted 13 February 2020 - 08:47 AM

Good work, that was some fine seeing there!



#4 Astroman007

Astroman007

    Voyager 1

  • *****
  • Posts: 10,132
  • Joined: 27 Oct 2017
  • Loc: Northern Ontario, Canada

Posted 13 February 2020 - 08:58 AM

Great work.

 

Thank you for sharing it with us! Much appreciated. I love our Moon more and more.



#5 AG Optical

AG Optical

    Vendor -AG Optical

  • *****
  • Vendors
  • Posts: 141
  • Joined: 09 Sep 2010
  • Loc: Alabama

Posted 13 February 2020 - 08:30 PM

Terrific!  

 

Thanks for sharing.

 

- Dave



#6 sunnyday

sunnyday

    Skylab

  • *****
  • Posts: 4,247
  • Joined: 23 Sep 2019
  • Loc: Ottawa,Canada

Posted 13 February 2020 - 08:50 PM

one word wow 



#7 sushidelic

sushidelic

    Lift Off

  • -----
  • topic starter
  • Posts: 6
  • Joined: 04 Jun 2019

Posted 14 February 2020 - 04:06 AM

Thank you for your replies!

 

It was really outstanding seeing on that night compared to the conditions that usually exist here in my area. Often I can be happy if I'm just able to make out the "big four" craterlets inside of Plato.

Nevertheless, seeing varied inbetween the different panels, which can be seen on close inspection in the final mosaic. I tried to circument this, by stacking different amounts per panel (6, 9, and 15%), and chosing afterwards, which one gave the best result after sharpening. That again resulted in different noise levels, so finding a middle solution was the way to go.

Also my SSD should have per specs easily been able to handle the full frame rate of the ASI 183, which it didn't in practice. So I had to restrict the framerate to 7 fps, to not drop frames or run out of buffer and nearly stall. I have to research that issue, but I guess my PC is just to plain old to handle modern SATA III drives accordingly, or it might be a WIN7 issue with modern hardware. Running the ASI at 18fps would have given me the chance to collect 4000 frames per panel, what would have contributed to a better result (I'm just imaginig Autostakkert would run for three days in a row then, but hey, if it's worth it!).

 

Best regards,

Michael



#8 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • Posts: 2,105
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 14 February 2020 - 04:22 AM

Michael, you have produced a nice result.  The greatest challenge of these lunar mosaics is getting consistency across all the panels.  Unlike in planetary imaging in which many video captures can be taken in succession at a very fast frame rate, with large lunar images you are pretty much confined to one single attempt per panel.  And the slower frame rates make it even more difficult unless the seeing is extremely good.  And you chose a lunar phase near the Full Moon, which only increased the number of panels required.  So all in all, this is a nice result.  Most of the more interesting regions (for this phase) near the Western limb and the poles have good detail.  

 

Looking at your quoted figure of 800GB and 20 panels, you must have been capturing in 8 bits, so I'm not sure what the cause of the speed decrease is.  With my ASI183mm, I can get 19fps when capturing in 8 bits, but the speed drops considerably down to 7-9fps if I capture in 12 bits, which I almost never do.  There is basically no improvement whatsoever in quality with 12 bit files when you factor in the loss of frame rate and reduced number of frames per panel.  

 

You mentioned grid artifacts after stacking.  I also got this with my ASI183mm, but I found that if I enabled the Gaussian blur feature (set to 1.5 pixels both horizontal and vertical) in AS!3, this eliminated it completely.  This is available under the menu of experimental features in AS!3, and the blur is only added for alignment, but not stacking.  Essentially it makes it much less likely for AS!3 to mistakenly align APs based on pattern noise.  Also worth pointing out is that I have been testing Rolf Hempel's stacking software on my lunar videos, and it performs at least as good as AS!3, and works 2x to 3x faster.  I would consider testing that out for your large files.  The huge amount of time it takes to stack these files is a major impediment in producing mosaics, and it's always a tragedy if you find a stacking error after waiting that amount of time and you need to start over.  Rolf's software is the topic of a huge thread here on CN, although maybe you've already read this.  

 

https://www.cloudyni...ysystemstacker/



#9 sushidelic

sushidelic

    Lift Off

  • -----
  • topic starter
  • Posts: 6
  • Joined: 04 Jun 2019

Posted 14 February 2020 - 12:00 PM

Hello Tom,

 

thanks for your comprehensive reply!

 

For the issue of not beeing able to constantly record at the highest frame rate, I am pretty sure that the problem lies somewhere in the SATA connection of the SSD. You were totally right that I am at 8 bits, as 12/16 doesn't make much sense for lunar imaging. Using firecapture, USB3 live preview constantly runs at 19fps, no stuttering. If I start a recording at full speed, I can watch the buffer diminishing, and when it's used up, recording continues at very low frame rate. I can also watch the difference between "Frames captured" and "Frames saved" instantly when recording full speed. So it seems to be clear that my computer is not able to transfer the needed 400MB per second to the SSD, what it should easily achieve per spec. I have an up to date laptop running WIN 10, and I will install Firecapture on it and see how it performs in comparision to my old but still ok workstation (which is an Intel Core i7 970 hexacore @ 3.6GHz, 32GB Ram).

 

I was using horizontal and vertical blur in Autostakkert!3 to prevent the grid artifacts, but the problem was, as soon as I turned it on both ways, Laplace delta returned an unusable quality graph. Switching to classic gradient solved this issue, but I still have artifacts. Not even close as bad as without blurring, but visible. I was capturing at Gain 0, and from what I've read, this might likely contribute to the issue.

 

I was testing out Planetary System Stacker, but my results were not as good as with Autostakkert!3 and it crashed on batch processing. That said, I want to make clear that this is absolutely due to my inexperience with the software. I was using it for the first time on this project, and I read the manual wink.gif. I will need to get more comfortable with it and get a feeling on how it behaves, then I am sure it will return excellent results, I'm not just there yet. So I lost patience and returned to AS, that I have way more experience with.

 

Which brings me to a point I completely forgot - I was looking if there's any software that helps me capturing my panels and stumbled across "Moon Panorama Maker", also by Rolf Hempel, which is just INSANE. I was hoping for a program, that somewhat simplifies that process, but MPM is incredibly feature complete! Just as an example, the build in Drift Correction works even better than Pulse Guide, that I am thinking of "misusing" this feature for unguided DS photography. So my big thanks and kudos go to Rolf for giving away freely this complete package to us fellow moon addicts!

 

Best regards,

Michael

 

EDIT:Typo (non native ;) )


Edited by sushidelic, 14 February 2020 - 12:03 PM.


#10 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • Posts: 2,105
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 14 February 2020 - 01:22 PM

Yes, Rolf deserves a lot of credit for his work, which is amazingly good, simple to use, and his documentation and user guides are also very excellent.  I don't use his panorama maker, mostly because I image on a Mac, but also I purchased the ASI183 specifically because it would require fewer panels for Moon mosaics.  I typically image nearer to the Quarter phases, and with my C9.25 Edge I can cover the entire Moon in only 4 panels.  Your C14 Edge will require more panels than that, but you will at least be far fewer than the 20 you needed for the Full Moon.

 

What program did you use to stitch the mosaic panels?  I've always found it interesting that the panels never perfectly align, which is especially obvious if you try to align everything by hand.  Even when using panorama stitching software, if you are able to inspect the alignment before blending, you will see that even here, the panels will never perfectly align either.  I was initially surprised by this, since I had naively assumed that aligning panels would be trivial.  But it is actually quite interesting that it is not. Fortunately, you can almost never detect the discrepancies in the final blended result.  This does have implications for using the image downstream though or making measurements, such as the distance between two points on the lunar surface.  I also use Rolf's LRGB program quite a lot, and it does an amazing job of perfectly aligning two images.  You can even repurpose the program for uses other than LRGB generation.

 

One thing you could try, if you haven't already, in Firecapture is to enable the RAM buffer.  This is under the settings menu, and you can try adding additional size to the RAM buffer.  I found that this did help increase my frame rates.  Initially, like you, I experienced a 19fps for a few seconds, and then the buffer would fill and I would drop all the way down to 2-3fps, which is basically unworkable.  But I'm also imaging on a Mac, and was using the beta version of FC, which has since been replaced.  In any event, all of the issues you describe are challenges that lunar mosaic imagers can relate to.  For something that seems very simple, it is not so simple! 



#11 sushidelic

sushidelic

    Lift Off

  • -----
  • topic starter
  • Posts: 6
  • Joined: 04 Jun 2019

Posted 14 February 2020 - 01:44 PM

I am using Microsoft ICE (Image composite editor) for stitching. Not a "Pro" solution, but really easy to use, free and it can output PSD files containing the layers, so you can modify the blending down the road, what I did for this picture to use as much pixels of the better tiles as possible.

I have been using PTGui at work for spherical HDRI stitching, and maybe this would also do a good job. Never tried though.

 

I also experienced that panels do not exactly line up. I guess that is because of the reference frame, as all the subsequent frames are not "normalized" in position, but aligned to that one single reference, as far as I understand. So the seeing distortion of this picture will be kept all down to the final stack. For smaller details, this can surely be significant, if I watch how craterlets are sometimes jumping around on the screen. But as you said, i have a hard time finding those faults in the final stitched image.

 

Thanks for the tip with the RAM buffer, I have only checked "aggressive RAM recovery" till now and did not see that option, I'll try that out. Imaging at ~8FPS helped me to prevent underflow, but still there were frames not saved to disk when I switched to the next tile and I had to wait for all of them been written before continuing. So this setting might not help with my actual problem, but if I manage to get my SSD to run faster, more RAM buffer would be very much appreciated.



#12 Tom Glenn

Tom Glenn

    Vanguard

  • -----
  • Posts: 2,105
  • Joined: 07 Feb 2018
  • Loc: San Diego, CA

Posted 14 February 2020 - 02:26 PM

I am using Microsoft ICE (Image composite editor) for stitching. Not a "Pro" solution, but really easy to use, free and it can output PSD files containing the layers, so you can modify the blending down the road, what I did for this picture to use as much pixels of the better tiles as possible.

I have been using PTGui at work for spherical HDRI stitching, and maybe this would also do a good job. Never tried though.

 

I have use ICE, but wasn't aware that you could export the individual panels.  That's a good option, because my problem with it was that when I had ICE blend the panels for me, it created black clipping along the terminator that discarded important details and increased noise.  The current version of Photoshop is also doing a very good job of stitching.

 

 

I also experienced that panels do not exactly line up. I guess that is because of the reference frame, as all the subsequent frames are not "normalized" in position, but aligned to that one single reference, as far as I understand. So the seeing distortion of this picture will be kept all down to the final stack. 

I think this is probably the single greatest contributing factor.  I've wrestling with trying to find an explanation for the inability to align panels, and have thought of a number of possibilities.  Rolf had suggested lens distortion, although for an f/10 SCT at these focal lengths, pincushion and barrel distortion are essentially nonexistent.  I recently took a look at a few videos in which I had captured the exact same image (same framing in the camera, same lunar region) in succession, separated by just a few minutes.  Even here, the images would not perfectly align, even though they should be identical, and so I think it simply has to do with how the APs are stitched together in each stack, which in turn is related to the seeing.  If you play around with different AP sizes in AS!3, and then make a layered stack in Photoshop of the results, you can see small features jumping by a few pixels in some places.  So distortion is the answer, but it is atmospheric distortion.  Another interesting contributor, at least in theory, is that the libration of the Moon is changing, mostly due to diurnal libration as the Earth rotates during the period of time required to capture the images, but also somewhat because of the orbital motion of the Moon.  If you look up the exact sub-observer point (for your specific location) on the lunar surface, you will see that in one hour, that sub-observer point will change by about 4", which doesn't sound like much, but corresponds to about 20 pixels with my setup.  This introduces an apparent rotation of the Moon, which although very small, would make even "perfect" images fail to align pixel for pixel if they are separated by a period of time, especially if they include both the limb and central portions of the Moon.  This is why LRO mosaic images are not registered to each other, but rather each individual image tile is registered to its exact location on the lunar surface, accounting for the sub-satellite position and camera pointing angle at the moment of the image.  



#13 lakeorion

lakeorion

    Apollo

  • -----
  • Posts: 1,410
  • Joined: 03 Aug 2010
  • Loc: Lake Orion MI

Posted 14 February 2020 - 03:15 PM

If you play around with different AP sizes in AS!3, and then make a layered stack in Photoshop of the results, you can see small features jumping by a few pixels in some places.  So distortion is the answer, but it is atmospheric distortion. 

I came to that conclusion also.  The biggest lens we use is the one we can't control.  Fortunately that lens allows us to breathe.

 

I have had fair luck with the freeware panorama program Hugin.  You can specify non-linked lens parameters for each panel, and let the program decide iteratively how to distort each panel to minimize mismatch.



#14 Tulloch

Tulloch

    Viking 1

  • *****
  • Posts: 779
  • Joined: 02 Mar 2019
  • Loc: Melbourne, Australia

Posted 14 February 2020 - 04:20 PM

Michael, you have produced a nice result.  

 

Looking at your quoted figure of 800GB and 20 panels, you must have been capturing in 8 bits, so I'm not sure what the cause of the speed decrease is.  With my ASI183mm, I can get 19fps when capturing in 8 bits, but the speed drops considerably down to 7-9fps if I capture in 12 bits, which I almost never do.  There is basically no improvement whatsoever in quality with 12 bit files when you factor in the loss of frame rate and reduced number of frames per panel.  

 

Agreed, nice result.

 

I too had issues with frame rate and USB throughput, I could only get a maximum of 60fps with my (much smaller sensor size) ASI224MC. I had a play around with the settings and found I had set the USBTraffic setting to 40 (since I had been getting problems before, and I was using a smaller ROI cutout for planetary). When I changed that up to a value of 100, my max frame rate at full sensor size is now around 150 fps. I wonder if this might be your issue also?

 

Andrew


Edited by Tulloch, 14 February 2020 - 05:25 PM.



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