What to do when Photometric Color Calibration Doesn't...

Seldom

Member
PCC appears to be doing a full plate solve, but ends with "Insufficient Photometric Data: Got 0 samples..."
 
Hi

Without knowing more about your image, it is difficult to say. Is it a wide field of view? Small FOV? Deep? NB...etc etc. (In focus??)
All of that being said, the first parameter to adjust as a blind guess is under Photometry. Uncheck the automatic magnitude limit and try something like 12 or 13th magnitude.
-adam
 
Target is a tile field around NGC1909 (Witch Head).  FOV is 106'x140'. Filters are LRGB.  Focus is best I could get with @focus3 on a Cyan night per ClearSky chart.  All tiles have been ImageSolved.  Scope is an NP101is with .8x FL reducer.  FL= 435mm.  Pixel size is 3.8mm. 

Should have added that PCC works with some of my images.

Follow up: Turning off the Automatic Magnitude limits and setting Magnitude to 13 got my image to calibrate.  Tried another problem image and that worked also.  Does this mean the problem's fixed?
 
Yes, this is a good solution for the issue you presented. You want to match the catalog faint threshold with something commensurate with your data so there is a good correspondence between the two of them.  Does this mean I made a good guess? I am "Seldom" correct. I had to do it... it was there...
-adam
 
And I seldom get useable answers to my questions...  So far the PI forum is batting way better than Cloudy Nights.
 
I'm relatively new to PI (only been using it for 2 months) and am having a great deal of difficulty getting PCC to work consistently. Two weeks ago, I captured images of the Rho Oph region with a 135 mm lens and when I used PCC on the stacked images, it worked perfectly. Last weekend, I took more images with slightly different framing (but using the same lens) and no matter what I do, PCC fails the plate-solving step.

Some notes:
- I definitely did NOT generate or apply Drizzle.
- I used Astronomy.net to get the exact coordinates of the image and the pixel size.
- I extracted the date/time and focal length from the image.
- I tried using 4 different Astrometry catalogs.
- I tried disabling "Limit magnitude" and setting it anywhere from 10 to 14.
- I tried various settings of Sensitivity, from -2.0 to -1.0.
- I tried setting Noise reduction to 1 to filter out tiny stars and background noise.
- I tried Triangle and Polygon alignment.

Nothing seems to work - it always fails stating that too few stars matches were found. I checked "Force plate solving" and it gets to attempt #27 and just hangs.

I'm out of ideas and quite frustrated at this point. Any help or advice would be greatly appreciated.

If anyone cares to give it a go, the stacked FITS file is here along with the stats from Astronomy.net and the PCC log (with mostly default settings) for reference.
 
1. It would be a useful diagnostic to see (or provide) the image of detected stars. You will be able to see what PCC sees of your image and you can blink this with your actual image.
2. Make the image available for download. Write in your message the field center, pixel size, and focal length.
 
This image can be solved without any problems:

Code:
lsast

Rho_Oph_Stacked_Normal_DBE2_BN:
Referentiation matrix (world[ra,dec] = matrix * image[x,y]):
-2.47179939e-03  -7.76904600e-05  +7.90274138e+00
+7.82787001e-05  -2.47439259e-03  +4.92613620e+00
WCS transformation ....... Thin plate spline
Control points ........... 2650
Spline lengths ........... l:1467 b:1305 X:1437 Y:1276
Projection ............... Gnomonic
Projection origin ........ [3131.473838 2089.912454] px -> [RA: 16 28 14.268  Dec: -24 14 35.66]
Resolution ............... 8.908 arcsec/px
Rotation ................. 1.776 deg
Observation start time ... 2020-06-20 05:07:55 UTC
Observation end time ..... 2020-06-20 06:14:03 UTC
Focal distance ........... 115.78 mm
Pixel size ............... 5.00 um
Field of view ............ 15d 29' 57.0" x 10d 20' 33.6"
Image center ............. RA: 16 28 14.268  Dec: -24 14 35.66  ex: -0.002710 px  ey: +0.024261 px
Image bounds:
   top-left .............. RA: 17 01 20.045  Dec: -19 09 08.69  ex: -0.005554 px  ey: -0.002723 px
   top-right ............. RA: 15 56 35.041  Dec: -18 41 07.21  ex: -0.001803 px  ey: -0.001189 px
   bottom-left ........... RA: 17 02 39.311  Dec: -29 21 37.38  ex: -0.154853 px  ey: +0.227920 px
   bottom-right .......... RA: 15 52 30.372  Dec: -28 51 11.80  ex: +0.002859 px  ey: -0.004103 px

On the ImageSolver script, I have entered the coordinates of Antares (by clicking the Search button) as an approximate center position. I have selected the PPMX catalog with a limit magnitude of 10.5, with distortion corrections enabled. I have not even downloaded your Astrometry.net data; it is not necessary.

Your image contains asteroids 338 Budrosa (12.72 mag) and 448 Kreusa (12.45 mag). Both are shown as small trails because the image has been acquired during about 1 hour. If the metadata had geodetic coordinates, asteroid annotations would be more accurate with topocentric instead of geocentric coordinates. Here is a nice annotated version of your image:


On a side note, you shouldn't use the FITS format in PixInsight (except to input raw data). It has been deprecated and does not support any of the advanced features that we have implemented in XISF. This will be even much more critical with the new versions that we'll release this year.
 
As an interesting byproduct of the astrometric solution, here is a very accurate distortion map for your image (that is, for the lens you have used to acquire it), generated by the ImageSolver script with data from 2650 stars:


Juan,

Thanks for timely reply. I will try the settings you suggested. I have a few questions, though:

1) In the future, how do I determine what settings need to be altered in order to achieve a successful plate-solving?
2) How do you set the "Limit magnitude" to 10.5? It appears to only allow integers.
3) Instead of FITS, what file format should I be using?
4) How do I read that distortion map and what is it telling you?
5) How do I locate those asteroids in the annotated image? I see the labels, but cannot find the trails...

- Bill
 
Last edited:
OK, I just tried PCC with the suggested settings and it failed again. I uploaded the log file here and attached a screen shot of the settings. What am I doing wrong?
 

Attachments

  • Settings.JPG
    Settings.JPG
    162.4 KB · Views: 198
Hi Bill,

1) In the future, how do I determine what settings need to be altered in order to achieve a successful plate-solving?

The most important factors to take into account here are:

- You need approximate coordinates of the center of the image. Usually this is no problem because these coordinates are part of standard acquisition metadata, or you can search for a deep-sky object represented in the image.

- Also necessary is the approximate scale of the image: focal length and pixel size, or arcseconds per pixel. As above, these data should be included in acquisition metadata, or you should already know them.

- One must select an appropriate star catalog and limit magnitude. This mainly depends on the field size. You can start selecting the automatic options of ImageSolver, which are suboptimal in most cases but let you start with a valid astrometric solution, which can be refined later.

- Wide-field images always pose difficult problems to compute accurate astrometric solutions. In general, it is much better solving them with the ImageSolver script instead of using the plate solving feature of PCC (which uses ImageSolver internally, but with a reduced set of configuration options). This is also applicable to any image when you want to compute very accurate astrometric solutions, since in most cases (especially when there are significant field distortions) you'll have to fine tune solver parameters.

In this case I have used the PPMXL catalog with a limit magnitude of 10.5. I tried first with 12 limit magnitude, but the amount of stars in the field was excessive. Unfortunately, the Gaia DR2 catalog is not useful for bright magnitudes, and the VizieR implementation could be better so it can take a very long time to download a large set of objects. In these cases PPMXL is a good compromise solution.

2) Instead of FITS, what file format should I be using?

XISF is the native file format of the PixInsight platform.

3) How do I read that distortion map and what is it telling you?

The distortion map is a graphical representation of the difference between a linear (distortion free) astrometric solution and the actual solution that has been computed for your image. With no distortions, all red spots should be perfectly circular and your image would be a geometrically perfect representation of the sky under the employed projection system (Gnomonic projection by default). When you see red circles with lines, the direction and length of each line measure the distortion present at the circle location. Typically (as in your image), distortions tend to grow radially from the center of the image towards the corners. Field distortions always tend to be larger in wide field optical systems.
 
Hi Bill,



The most important factors to take into account here are:

- You need approximate coordinates of the center of the image. Usually this is no problem because these coordinates are usually included in acquisition metadata, or you can search for a deep-sky object represented in the image.

- Also necessary is the approximate scale of the image: focal length and pixel size, or arcseconds per pixel. As above, these data should be included in acquisition metadata, or you should already know them.

- One must select an appropriate star catalog and limit magnitude. This mainly depends on the field size. You can start selecting the automatic options of ImageSolver, which are suboptimal in most cases but let you start with a valid astrometric solution, which can be refined later.

- Wide-field images always pose difficult problems to compute accurate astrometric solutions. In general, it is much better solving them with the ImageSolver script instead of using the plate solving feature of PCC (which uses ImageSolver internally, but with a reduced set of configuration options). This is also applicable to any image when you want to compute very accurate astrometric solutions, since in most cases (especially when there are significant field distortions) you'll have to fine tune solver parameters.

In this case I have used the PPMXL catalog with a limit magnitude of 10.5. I tried first with 12 limit magnitude, but the amount of stars in the field was excessive. Unfortunately, the Gaia DR2 catalog is not useful for bright magnitudes, and the VizieR implementation could be better so it can take a very long time to download a large set of objects. In these cases PPMXL is a good compromise solution.



XISF is the native file format of the PixInsight platform.



The distortion map is a graphical representation of the difference between a linear (distortion free) astrometric solution and the actual solution that has been computed for your image. With no distortions, all red spots should be perfectly circular and your image would be a geometrically perfect representation of the sky under the employed projection system (Gnomonic projection by default). When you see red circles with lines, the direction and length of each line measure the distortion present at the circle location. Typically (as in your image), distortions tend to grow radially from the center of the image towards the corners. Field distortions always tend to be larger in wide field optical systems.

I had the exact coordinates of the center of the image and the exact focal length and pixel size, so I'm sure none of those are the problem.

How do you know which star catalog to use?

Can you point me to a resource (e.g., an online tutorial) that describes how to use ImageSolver and feed the results into PCC?

I added a couple more questions to my list above before you posted your reply - can you please take a look at the new questions?

And I tried PCC with the settings you defined - it still failed (see previous post). So I'm still baffled and confused...
 
Last edited:
How does one add additional star catalogs? My PCC only has APASS. Nothing else is in the dropdown.

he's talking about the plate-solving phase of the target image - in there there are several catalogs in the PCC dropdown.

rob
 
The image can be solved and calibrated with PCC without problems:


However, after PCC a green cast remains. I don't know why. Is the image in linear state?
I took your advice and started over with the stacking and registration process, saving everything in XISF format this time. I had no better luck with an XISF image using the parameter settings you described earlier (using Antares as the reference, PPMXL as the catalog, Limit magnitude to 10). It still fails (log capture below). I uploaded the unstretched XISF file HERE.

Can anyone tell me what I'm doing wrong?

PCC Log output:
==============
Reading image:
C:/Users/Bill/AppData/Local/Temp/PCC_T_4GKWRFNSF0CN.xisf
Loading image: w=6264 h=4180 n=3 RGB UInt16
515 FITS keyword(s) extracted.
Seed parameters for plate solving:
Image coordinates: RA = 16 29 24.460, Dec = -26 25 55.21
Resolution: 13.598 as/px
Starting StarAlignment iteration
Catalog PPMXL already loaded.
Catalog PPMXL size: 57418 objects

Using the triangle similarity star matching algorithm.
C:/Users/Bill/AppData/Local/Temp/stars.csv:
Scanning star data: done
9736 stars.
PCC_T_4GKWRFNSF0CN:
Structure map: done
Detecting stars: done
57186 stars found.
* Reference image: Limiting to 2000 brightest stars.
* Target image: Limiting to 2000 brightest stars.
* Distortion correction: Iteration 1 of 50
Matching stars: done
266 putative star pair matches.
Performing RANSAC: done
* Previous attempt failed - this is try #2
useScaleDifferences=true
* Reference image: Limiting to 2000 brightest stars.
* Target image: Limiting to 2000 brightest stars.
* Distortion correction: Iteration 1 of 50
Matching stars: done
413 putative star pair matches.
Performing RANSAC: done
* Previous attempt failed - this is try #3

...

* Previous attempt failed - this is try #20
useScaleDifferences=true
* Reference image: Limiting to 3000 brightest stars.
* Target image: Limiting to 3000 brightest stars.
* Distortion correction: Iteration 1 of 50
Matching stars: done
580 putative star pair matches.
Performing RANSAC: done
*** Error: Unable to find an initial set of putative star pair matches.
*** Error: The image could not be aligned with the reference star field
Please check the following items:
The initial coordinates should be inside the image.
The initial resolution should be within a factor of 2 from the correct value.
Adjust the star detection sensitivity parameter, so that the script can detect most of the stars in the image without mistaking noise for stars.
The catalog should be matched to the image. Choose the appropriate catalog and magnitude filter, so that the number of stars extracted from the catalog can be similar to the number of stars detected in the image.
*** Error: Unable to plate solve image: Alignment failed
This usually happens because the initial parameters are too far from the actual metadata of the image.

** Warning: Process finished with errors:
PCC_T_4GKWRFNSF0CN.xisf: The image could not be plate solved

*** Error: Failure to plate solve image: Rho_Oph_Normal_DBE
Reading swap files...
1808.471 MiB/s
<* failed *>
 
I figured it out - I was entering the pixel size in arc-sec/pixel (8.9 in this case) instead of the size of the pixel on my camera sensor (5.67 um). Once I changed that, it solved right away. A boneheaded error on my part.
 

Attachments

  • Lagoon_ABE.jpg
    Lagoon_ABE.jpg
    408.2 KB · Views: 82
  • Lagoon_ABE_PCC.JPG
    Lagoon_ABE_PCC.JPG
    439.2 KB · Views: 102
  • Lagoon_ABE_PCC_SCNR.JPG
    Lagoon_ABE_PCC_SCNR.JPG
    444.1 KB · Views: 82
You should never have to use SCNR after PCC. The green cast cannot be reproduced with other data sets covering the same region of the sky. The only reason I can figure out is lack of linearity, which may have been caused by your application of ABE.

This image is not at all a good candidate to use the ABE tool. You should use DBE placing as few samples as possible, very carefully located on relatively free sky regions. This is something that ABE cannot do on complex images like this one.
 
Back
Top