Author Topic: mosrange: **CAMERA ERROR** Camera missed planet or SPICE data off. (SOLVED)  (Read 2828 times)

bierhaus

  • Anubis (Very Active Member)
  • ****
  • Posts: 57
Hello ISIS,

I am attempting to mosaic the Cassini ISS sequence "ISS_121RH_REGMAP001_PRIME", which consists of 15 images.  I developed a control network using qnet, and after several iterations with the control network (ignoring the higher-error measures, ignoring some points completely, setting method=svd, increasing maxits and increasing tol to 0.9), I was able to get jigsaw to converge.  However, mosrange gave me an error I do not know how to address:

mosrange fromlist=gapfill_list.txt to=rege001_v1.map
mosrange: Working
**CAMERA ERROR** Camera missed planet or SPICE data off.

Any suggestions on how to proceed?  Thank you,

Beau
« Last Edit: November 17, 2010, 01:51:36 PM by kbecker »

kbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 536
    • http://isis.astrogeology.usgs.gov
    • Email
Re: mosrange: **CAMERA ERROR** Camera missed planet or SPICE data off.
« Reply #1 on: November 04, 2010, 11:45:30 PM »
Hi Beau...

Quote
mosrange fromlist=gapfill_list.txt to=rege001_v1.map
mosrange: Working
**CAMERA ERROR** Camera missed planet or SPICE data off.

This indicates that the camera FOV in at least one of your images included in your list does not intersect the target.  You must determine which image that is and eliminate it from the list.  You may be able to better guess which one this is by which percentile the program reports complete.  If it doesn't make it to 10% complete, the offending image is within the first 10% of your file list.

I have submitted a change to the code to report the actual offending image file.  It will be available in the next few days when we release a new system.
Kris Becker
Computer Scientist
Astrogeology Team
U.S. Geological Survey

kbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 536
    • http://isis.astrogeology.usgs.gov
    • Email
Beau...

The new 3.2.1 version of ISIS should help as it now includes a more complete error message.  Please update your ISIS system.

See the release notes for more details.
Kris Becker
Computer Scientist
Astrogeology Team
U.S. Geological Survey

bierhaus

  • Anubis (Very Active Member)
  • ****
  • Posts: 57
Hi Kris,

Thanks for the explanation, and I'll use the latest version of ISIS to track down which image causes the error.

A follow-up question: since all the images clearly contain Rhea as a target, how can the pointing be off such that jigsaw thinks one of the images doesn't actually contain Rhea?  Is that a problem in the control network?  But even then, the most offending points in the control network are a handful of pixels off, which shouldn't that error.  So what could cause the error to begin with, when all the images are clearly high-resolution images of Rhea?

Thanks,

Beau

kbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 536
    • http://isis.astrogeology.usgs.gov
    • Email
Beau...

Ahh, so you are using mosrange after a run of jigsaw?  Your struggles with jigsaw could be the cause.  Chances are if the image pointing prior to running jigsaw had valid geometry for all images containing your target (Rhea), allowing it to update the pointing moved the camera FOV off the target.

A quick test of this would be to rerun spiceinit on all your images and run mosrange on them prior to jigsaw.  If that is successful, then you need to refine (or eliminate) your control net points that are exhibiting high errors in jigsaw.

If you have a failure in mosrange immediately after spiceinit, then the offending image(s) never had good pointing to begin with and should be eliminated from the list (even though Rhea is visible in the image).  This happens from time to time.




Kris Becker
Computer Scientist
Astrogeology Team
U.S. Geological Survey

tbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 653
Proceed with Kris' suggestion...in addition, it might be helpful to see all the details of your jigsaw command and the rms results.   

Depending on the total and distribution of your points and measures across the images, be cautious when ignoring either one.  You want to retain a good distribution of points across the overlaps for each image.   A good distribution of points in a single image (for a frame camera) would be a point in all corners and at least one towards the center (where possible...obviously, the image overlaps drive the distribution).  Also, if possible, a single point should have more than 2 measures.  Avoid

If you have an image in your list that has reasonable camera pointing, you could 'hold' this image (pass the filename in a txt file and pass to the heldlist parameter in jigsaw) to constrain the solution; this would help the images that have very inaccurate pointing.  Your solution cause the pointing of all your images included to be converged to the pointing of the image held..which might or might not be accurate or consistent with other datasets or maps.

Tammy Becker

tbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 653
oops, didn't finish a sentence I was adding.....Avoid a distribution of points/measures that are clustered..e.g., too close together without other points/measures spread out across the overlap areas and across the images.  For example, 5 measures on an image at the lower left corner and no other measures in the other corners or other areas of overlap with other images..this will influence the outcome of the solution.  I bring this up with the idea of ignoring measures/points where it could introduce a cluster type scenario.

Tammy


bierhaus

  • Anubis (Very Active Member)
  • ****
  • Posts: 57
Hi Kris and Tammy,

Thanks for the responses.  I successfully reran mosrange on the pre-jigsaw cubes, so from now on I will be sure to run mosrange before jigsaw.  (This might be a good tip to include in the documentation.) 

However, I find I now have another error that I cannot decipher:

To simplify a first-cut mosaic for this sequence, I am now using just the first nine images, which were acquired in a roughly 3x3 grid.  I have created a control network spanning all available overlap areas (and avoiding control-point clusters as Tammy suggested).  After ignoring some of the larger error points and setting tol=0.6, jigsaw successfully converged.  I then ran cam2map:

cam2map from=\$1.gapfill.cub to=\$1.map.cub map=reduced_v1.map pixres=map -batchlist=reduced_list.txt
cam2map from=N1637518901_1.gapfill.cub to=N1637518901_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519058_1.gapfill.cub to=N1637519058_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519176_1.gapfill.cub to=N1637519176_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519283_1.gapfill.cub to=N1637519283_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519392_1.gapfill.cub to=N1637519392_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519501_1.gapfill.cub to=N1637519501_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519610_1.gapfill.cub to=N1637519610_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519768_1.gapfill.cub to=N1637519768_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed
cam2map from=N1637519875_1.gapfill.cub to=N1637519875_1.map.cub map=reduced_v1.map pixres=map
cam2map: Working
100% Processed

So far, so good.  I made a text file (reduced_map_list.txt) that is a simple list of all the *.map.cub files, and then tried to run automos:

automos fromlist=reduced_map_list.txt mosaic=regmap001_reduced.mos.cub
**I/O ERROR** Unable to open [N1637518901_1.map.cub]

I double-checked the existence of N1637518901_1.map.cub, and indeed I can even look at the file with qview.  If the file exists, and I can look at it with qview, why is automos unable to open the file?

Thanks,

Beau

tbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 653
hmmm, that's a frustrating error....without actually seeing the contents of your input list for automos, are you in the same directory as the .map.cub files? if the input list does not contain a directory path, automos is looking for the filenames in your current directory.

The only suggestion at this point is to double check the contents of your input list and what directory you are running automos from.

One more idea... :-/

 automos figures out the lat/lon boundaries that would include the coverage of all your input images and then mosaics them.  One work around (and less convenient) is to run mapmos and use the lat/lon boundaries that mosrange figures out.   

mapmos from=1st input map file.cub   mosaic=outputmos.cub create=yes minlat=   maxlat=    minlon=  maxlon=
mapmos from=2nd input map file.cub   mosaic=outputmos.cub
mapmos from=3rd input map file.cub mosaic=outputmos.cub
etc....

You really should be able to run automos with a list....if you are in the current directory, etc....maybe attach your input list so that we can check it out.

Tammy

kbecker

  • Isis Support Team
  • Isis (Extreme Power Member)
  • *****
  • Posts: 536
    • http://isis.astrogeology.usgs.gov
    • Email
Perhaps there is some funky character at the end of the line.  You can check the contents of the file using the od command:  od -cv reduced_map_list.txt.  You should see only '\n' at the end of each line including the last line.
Kris Becker
Computer Scientist
Astrogeology Team
U.S. Geological Survey

bierhaus

  • Anubis (Very Active Member)
  • ****
  • Posts: 57
Aha!  There were indeed some "funky" characters in the text file.  The "cat" command did not reveal them, but viewing the file in vi did, in fact, uncover some trouble-makers.  I removed them from the file, and successfully ran automos, with a beautiful Rhea mosaic.  Thanks for your help!

Beau