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

Let's Get Real About Field Coverage and Image Quality

  • Please log in to reply
122 replies to this topic

#1 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 05:55 AM

Okay, there seems to be a lot of talk about image quality on CN and vendors sure like to use terms like "pin-point stars" or "fully illuminated and corrected" fields but how often do you find REAL WORLD results presented at full image scale and/or with minimal processing? Similarly, how often do you see FWHM and eccentricity measurements presented in a plot covering a full APS-C field? Not very often I think.

 

In any case, I would like to start a thread/topic that allows users to show how their systems are performing using actual images at a full 1:1 pixel scale and hopefully with plots showing the full-field performance in FWHM (star size, expressed in arc seconds) and eccentricity (star roundness).

 

First some caveats, YMMV, not all tools report the same results, not all cameras will give the same results (pixel size, Bayer patterns, mono versus one-shot-color, filtered or not), and individual results may vary by both user, event or session, technique, and sample of equipment (who, when, how, what).

 

In any case, here is my first contribution with results from an approximately two-year-old Tele Vue NP127is at prime focus (660mm with no correctors or reducers) on an unmodified Sony NEX-5R camera (APS-C format). The image scale at capture was 1.5 arc seconds per pixel and the image matrix showing the system's performance is shown at that same reproduction (1:1, or 1.5 arc seconds per pixel). This image is based upon a stack of 16 subs, each exposed for 30 seconds under red-zone light pollution with NO filters, with all processing done in PixInsight version 1.8 (core Core 01.08.03.1123). The image had a very slight crop (approx. 100 pixels) from each edge to remove processing artifacts from the registration and stacking. The cropped image is 4704 x 3116 pixels in size and each sample in the corner-to-corner and edge-to-edge matrix is 396 pixels square (the matrix image is 1196 x 1196 pixels so that it just fits within the posting guidelines on CN).

 

First the actual image matrix (processing included dynamic background extraction, color calibration, and a simple histogram stretch). This is an RGB image that was processed as a 32-bit FITS file and then saved as a JPEG (after conversion to an 8-bit image in PixInsight). The target field was centered on the globular cluster M5 and the field angular size is approximately 118 arc minutes by 78 arc minutes in size (2.0 x 1.3 degrees).

Attached Thumbnails

  • Aberration Matrix NP127is.jpg

Edited by james7ca, 11 June 2015 - 03:16 PM.

  • coinboy1, StarDust1, Guest7564635 and 3 others like this

#2 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 06:30 AM

And here are the FWHM and eccentricity measurements as taken with PixInsight's FWHMEccentricity script with a log star detection sensitivity of 0.00 (the default is usually -1.0). The script was run on the 16-image stack after crop, dynamic background extraction, color calibration, and conversion to a grayscale image (but still in linear format without a histogram stretch).

 

Once again, the image scale is 1.5 arc seconds per pixel and the script reports in pixels, so the median FWHM is 2.044 pixels x 1.5 arc seconds or 3.0 arc seconds. According to the PixInsight documentation most people consider a star to look circular at an eccentricity value of 0.42 or less. So, it should be apparent that some parts of this field do not have circular looking stars (the upper left and lower right being the worst).

Attached Thumbnails

  • FWHM.jpg
  • Eccentricity.jpg

Edited by james7ca, 11 June 2015 - 02:53 PM.


#3 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 06:45 AM

Finally, here is a reduced sized version of the entire image field to give you some idea about the image scale and size (reduced from the original capture scale to fit within the maximum allowed image size on CN).

Attached Thumbnails

  • NP127is Full Field M5 Small.jpg


#4 terry59

terry59

    Voyager 1

  • *****
  • Posts: 10,457
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 11 June 2015 - 08:26 AM

Scope: SW 120ED; Reducer: Orion 0.8x; Camera: Atik 460EX

 

This is just the red channel from my last imaging session with only a STF applied

Attached Thumbnails

  • Capture.JPG


#5 terry59

terry59

    Voyager 1

  • *****
  • Posts: 10,457
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 11 June 2015 - 08:27 AM

FWHM and Eccentricity

Attached Thumbnails

  • Capture.JPG


#6 Alex McConahay

Alex McConahay

    Voyager 1

  • *****
  • Posts: 10,410
  • Joined: 11 Aug 2008
  • Loc: Moreno Valley, CA

Posted 11 June 2015 - 09:41 AM

>>>>>>FWHMEccentricity script with a log star detection

 

How are you getting those thumbnails. When I run the Script/Image Analysis/FWHMEccentricity script in PI I get a text response, but no graphic display. What am I missing?

 

Alex



#7 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 13,186
  • Joined: 14 Jan 2013

Posted 11 June 2015 - 10:46 AM

Alex: Hit the support button.

 

As for image quality, FWHM values are going to depend a lot on seeing, and somewhat on image scale. Also, meridian flips will significantly affect the results.



#8 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 13,186
  • Joined: 14 Jan 2013

Posted 11 June 2015 - 10:57 AM

Here are results of 8 thirty minute subs binned 1x1 on IC 5068 last night.

 

AberrationInspectorIC5068.jpg

 

eccentricityIC5068.jpg FWHMIC5068.jpg

 

Scope used was a Stellarvue SV80ST-25SV. Polar alignment was poor (6' off in azimuth, 4' off in altitude). Image scale is 2.28 arc-seconds per pixel. Filter was Ha. Seeing was fairly poor.

 

Flattener used was a TS Flat2 at 124mm spacing.

 

The camera is a SBIG STF-8300m. The aberration inspector plot is shown at full size.


Edited by Madratter, 11 June 2015 - 02:33 PM.

  • DivisionByZero likes this

#9 Alex McConahay

Alex McConahay

    Voyager 1

  • *****
  • Posts: 10,410
  • Joined: 11 Aug 2008
  • Loc: Moreno Valley, CA

Posted 11 June 2015 - 11:12 AM

>>>>Alex: Hit the support button.

 

Okay, thanks. Ive tried than and get "Insufficient Number of stars fitted for supporting contour image generation."  Which is strange, since I am shooting the "ET Cluster." Figure there should be enough stars in an open cluster!!!

 

I'll give it more of a go. 

 

Alex



#10 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 01:55 PM

Alex, you're going to have to have an image (stack or clean single shot) that has between 1000 and 2000 stars that can be detected and fitted or you may get the "Insufficient Number of stars fitted..." error. I'm not sure whether there is a hard cutoff in the number and/or how well the stars need to be distributed in the field. Generally, once I start drifting below 1000 stars then the plots may fail (but I think I've seen success with numbers around 800). You may want to try a different (lower) log star detection sensitivity but if you go too low then you may pick up noise or pixel defects in the image or get stars that are so small that the quality estimators will be skewed. I've seen cases where I've gotten ridiculously good numbers for FWHM (very low, around one pixel) that were nothing but noise.

 

As for the images that have thus far been posted, have all of these been presented at full scale? I notice that some of them seem a little small even when zoomed up from the preview thumbnail. CN allows images that are up to 1200 pixels square and that is what I targeted when using the AberrationInspector script in PI, but I guess if your sensor is smaller than 1200 pixels square then you obviously can't post something that large. I used a 3 x 3 mosaic with a panel size of 396 pixels.

 

Also, it would be nice if everyone reported their image scale in arc seconds per pixel and the size of their sensor in millimeters (or just full-size -- 36 mm x 24 mm -- or APS-C -- assumed near 23 mm x 15 mm). I'd also like to know the sensor size in pixels (i.e. 2750 x 2200 or whatever) which would allow someone to calculate the field coverage in arc minutes or degrees. Thus, if your image scale was 2.50 arc seconds per pixel and your sensor was 2750 pixels wide you'd have 2.5 arc seconds x 2750 = 6875 arc seconds or 115 arc minutes (approximately).



#11 David Ault

David Ault

    Gemini

  • -----
  • Posts: 3,424
  • Joined: 25 Sep 2010
  • Loc: Georgetown, TX

Posted 11 June 2015 - 01:56 PM

The script cuts each image up into tiles and then looks for a minimum number of stars per tile (30 I think).  If any more than 2 rows or columns don't have enough stars it reports that warning.  It's mostly an indicator that the the support plots may not be accurate.

 

EDIT: Like James said, somewhere around 1000 stars evenly distributed is a good target.  Ideally you want to use short enough exposures to remove tracking effects and long enough exposures to even out the seeing.

 

Regards,

David


Edited by David Ault, 11 June 2015 - 01:58 PM.


#12 Goofi

Goofi

    Cosmos

  • *****
  • In Memoriam
  • Posts: 8,137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 11 June 2015 - 01:58 PM

Good topic, James!

 

Here's my last real imaging project ... Jones-Emberson1 in Oiii taken last month. NP-101, no reducer, Atik One 6.0 with Astrodon 3nm Oiii filter.  My image scale is 1.69"/pixel, pretty average seeing for me.

 

 

 

 

Attached Thumbnails

  • JE1_O3_Mosaic.jpg
  • JE1_O3_FWHM.jpg
  • JE1_O3_eccentricity.jpg


#13 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 01:58 PM

Here is a discussion on FWHM and eccentricity results that I started a little over one year ago on the PI support site. There are plots and values from several users with different equipment and a pretty wide range in the results. For those that are interested it could be a good read.

 

 http://pixinsight.co...hp?topic=6796.0


  • tkottary likes this

#14 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,929
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 11 June 2015 - 02:23 PM

Interesting stuff. I'll try to add some data from my setup. I have some tilt, I'd like to correct that before sharing, but I may not be able to for a while. 

 

I was pretty surprised how much the TSRCFlat2 did for my field when I first collimated it (before I fixed the focuser). It flattened the stars throughout most of my very, very large 5D III full frame field. The corners showed elongation and vignetting, but the majority of the field, and area much larger than an APS-C sensor, did produce pretty nice stars. 



#15 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 02:46 PM

Goofi, I think your image and mine are the first such results that I've seen on Tele Vue NPis scopes. The only real difference in our results may be because I'm working in Bayered RGB using a significantly larger sensor (APS-C, 23mm x 15mm, approx. after my crop) while you are imaging mono at a sensor size of 12.5mm x 10mm (no crop?).

 

Another thing that I've notice even in my own results is that the eccentricity measurements MAY not necessarily reflect how the stars actually look after the image processing is done. What I means is that you can have good looking eccentricity values with rather average or fair looking stars or somewhat poor eccentricity values and stars that still look very good (round). I think this has something to do with the actual shape of the star distortion and the processing that you apply when you actually get to the point of being able to see the stars (that is, the image is no longer in a linear format). Of course, this is only true in a moderate sense, because once the eccentricity values get really bad the stars will also begin to look really bad. 


Edited by james7ca, 11 June 2015 - 02:52 PM.


#16 Goofi

Goofi

    Cosmos

  • *****
  • In Memoriam
  • Posts: 8,137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 11 June 2015 - 03:14 PM

James, that's right .... no crop on mine.  The size of the Sony sensor helps with these sorts of results.  One thing ... my NP-101 is not the new/"is" version, it's the older original model.  If I'm not mistaken, you took your image the other night when it was kinda clear, but the seeing that night wasn't great here.  I took mine almost two months ago and the seeing was pretty good that night.  Your mount should guide better than mine, so that just leaves the difference between a mono CCD with Oiii filter and your OSC with bayer matrix.

 

There's a huge difference I found between a single sub and the stacked results.  I ran a test a few months ago with my subs and was amazed at the differences.  I was actually looking at noise reduction and seeing what was going on, but I noticed the difference between a single sub and stack of 13 subs on FWHM and eccentricity as well.



#17 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 03:44 PM

Since most of the posts thus far have been from medium-sized CCD sensors I've created this graphic which shows the relative field coverage of a Sony ICX694 sensor (Atik One and Atik 460EX) against my Sony APS-C camera. The rectangle in the center of these FWHM and eccentricity plots is what would be covered with the ICX694, which could be used to predict the performance with that sensor (on my NP127is). I may go back and crop my actual image to this same size and then re-run PixInsight's AberrationInspector script. The pixel size on the Sony NEX-5R is about 4.78 microns, while the Sony ICX694 is 4.54 microns (fairly close, although my Sony NEX has a Bayer matrix so the effective resolution is lower than what would be suggested by the pixel size). These differences in the pixel size were taken into consideration when I created the simulation (i.e. they were scaled to the physical size, not just the pixel count).

Attached Thumbnails

  • Simulated Sony ICX694 Sensor Coverage.jpg


#18 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 04:00 PM

Okay, here is what the simulated field of the Sony ICX694 would look like on my NP127is (it looks pretty good, but still RGB with a Bayer pattern versus a mono CCD).

Attached Thumbnails

  • Simulated Sony ICX694 Field.jpg


#19 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 04:16 PM

...There's a huge difference I found between a single sub and the stacked results.  I ran a test a few months ago with my subs and was amazed at the differences.  I was actually looking at noise reduction and seeing what was going on, but I noticed the difference between a single sub and stack of 13 subs on FWHM and eccentricity as well.

Yes, that can be very true. However, I've found that subs can vary from minute to minute either because of changes in the seeing or because the mount or tracking/guiding did a little hiccup (or something like wind or a vibration moved the scope). The stack I did of 16 subs was selected sequentially, I didn't try to hand pick the best subs but I didn't use any of the few subs that I rejected for clouds or for being very obvious outliers when I ran the SubframeSelector script. In any case, I suspect that stacking can either make the results better or worse, probably in most cases slightly better (assuming that you don't compare to an obvious outlier in the individual subs).

 

However, since very few of us do NOT stack our images it's probably fair to be measuring and looking at the stacked results (this also helps to build up the image data if you are using short exposures). Maybe one thing I should mention is that my subs were taken with NO guiding, but they were only 30 second subs done on a Mach1GTO mount so tracking shouldn't have been too much of an issue. In any case, if you are measuring subs that were guided over long exposures and you have good measurement results then you are obviously doing everything right.



#20 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 13,186
  • Joined: 14 Jan 2013

Posted 11 June 2015 - 04:36 PM

My experience with my both my Atlas and my MyT and my other equipment (i.e non mount equipment) is that stacking normally results in markedly better eccentricity values. It usually results in a FWHM value that is close to the average of the individual subs in the stack.

 

I believe the reason eccentricity normally gets better with my equipment when I stack is that a significant amount of the error causing the eccentricity is essentially random in nature and not systematic (i.e. something like differential flexure).

 

My experience is also that any deconvolution done on a stack will emphasize errors in eccentricity while improving FWHM.


Edited by Madratter, 11 June 2015 - 05:07 PM.

  • Goofi likes this

#21 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 04:46 PM

I've even read reports that high-altitude winds and atmospheric disturbances can cause poor eccentricity values. Some claim that if you have a consistent pattern in the winds (jet stream) that can result in misshaped stars that are always distorted in the same direction (almost like a consistent tracking error). In that case, the atmospheric-caused distortions would not be truly random. However, I find that claim somewhat hard to accept.



#22 Peter in Reno

Peter in Reno

    Voyager 1

  • *****
  • Posts: 11,887
  • Joined: 15 Jul 2008
  • Loc: Reno, NV

Posted 11 June 2015 - 04:58 PM

FWHM is a good way to determine auto guiding performance. Like MR said, the better quality the mount the better chance for lower FWHM.

 

For example. I have autoguide my TEC 140 scope with Baader 10x60 finderscope and OAG and I found FWHM typically smaller with OAG than finderscope.

 

Peter


Edited by Peter in Reno, 11 June 2015 - 05:01 PM.


#23 james7ca

james7ca

    Cosmos

  • *****
  • topic starter
  • Posts: 9,821
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 11 June 2015 - 05:03 PM

Here is a plot that shows the individual subs from the session I used to create my earlier reports on my NP127is. This shows some of the variability between each sub. Probably typical, although I've not seen many published results from PixInsight's SubframeSelector script.

 

Most of the rejects toward the end where from clouds. However, you can see if I had skimmed off from the very best subs then I probably would have had a better average result.

Attached Thumbnails

  • Eccentricity and FWHM (All Subs).jpg

Edited by james7ca, 11 June 2015 - 05:06 PM.


#24 Goofi

Goofi

    Cosmos

  • *****
  • In Memoriam
  • Posts: 8,137
  • Joined: 03 May 2013
  • Loc: Coastal Southern California

Posted 11 June 2015 - 05:21 PM

Madratter, now that you mention that  it makes sense to what I've been seeing.  When I was imaging with the AVX my eccentricity was showing the limitations of the mount; now that I'm using the AtlasPro it's more random variation and stacking cleans it up nicely.



#25 JamesSober

JamesSober

    Viking 1

  • -----
  • Posts: 880
  • Joined: 05 Apr 2015

Posted 11 June 2015 - 05:22 PM

   

Edited by JamesSober, 12 June 2015 - 12:48 AM.



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