 ...
·
·
6
likes
|
|---|
Hi folks, http://vizier.cds.unistra.fr/ is currently down, which means that PixInsight cannot solve.  I'll make sure pending AstroBin images solve automatically without you needing to do anything once Vizier is up again, which I hope is tomorrow (!) Thanks for your patience! Salvatore |
4.07
...
·
|
|---|
Hi Salvatore, It would seem Vizier at Strasbourg is up again (at least it works when I use it annotate an image in my Local PI), but advanced plate-solving on AB still fails. Do you see the same? Thanks, Francesco |
...
·
|
|---|
No, it’s just your image in afraid. Have you tried adding a FITS file in the advanced plate solving settings?
|
|
4.07
...
·
|
|---|
Thanks, let me try that.
|
|
4.07
...
·
|
|---|
No, looks like it still fails even after uploading a XISF file. It happens for two images: https://app.astrobin.com/u/frankz?i=cm7351 https://app.astrobin.com/u/frankz?i=j6gvmr The strange thing is that, for the second image, which is older, advanced plate solving used to work fine for previous revisions. |
|
4.07
...
·
|
|---|
One commonality between the two images is that, in both cases, the latest and final revisions have different pixel size than the previous revisions. Do you think this could be a cause for failure?
|
...
·
|
|---|
No, AstroBin uses the relevant revision data to plate-solving revisions. I can try to isolate this to see the exact logs.
|
|
4.07
...
·
|
|---|
Thanks, that’d be great. Whatever the problem is, it seems limited to those two images: I tried to re-start the advanced plate solving for other images, and it worked.
|
...
·
·
1
like
|
|---|
@Francesco Meschia (and mentioning @Joachim Stingl who is having the same problem), I don't really know what to do with this one, seems like PixInsight just doesn't like these particular images of yours (log will follow at the end of the message). The only card I can try playing, which I can do this week, is sending a higher resolution version to PixInsight, in case subsampling makes it more likely to fail this way. I'll keep you posted! (Unfortunately I couldn't copy/paste from the remote session, so here's two screenshots of relevant parts of the log)   |
|
4.07
...
·
|
|---|
Very interesting Salvatore – thank you! A question for you: PI not being a blind solver, I assume the Solver script is “seeded” with the coordinates and the image scale determined by the Astrometry.net blind solver? Is there a way to double-check the values being used for those images that fail? |
|
4.07
...
·
|
|---|
One more bit of strangeness: I have in my staging area a very old image of the Leo Triplet, which had a correct PI plate solution and advanced annotation. I tried to restart the advanced plate solving, and it failed just like in my more recent images. And they all happen to be in the same area of the sky. I tried to do the same for two other images in my staging area, in different parts of the sky (M106, the Cygnus Veil): it worked in both cases. It would be very strange, but could it be something related to sky coverage of the catalog used by PI? Maybe some part of it got corrupted? What part of the sky is your image that fails solving, @Joachim Stingl? |
|
4.07
...
·
|
|---|
Another data point: I just uploaded a new image of NGC 3628. It also fails to advance-solve. It does seem something related to this particular corner of the sky… |
...
·
|
|---|
Francesco Meschia: It's possible. If we search for recent NGC 3628 with a small field radius (so it's only that galaxy and not the whole triplet), we can see that none of the recent one have PixInsight solutions. The most recent one to have a solution is this one from last June: https://app.astrobin.com/i/nwtagk And I received a new version of PixInsight to run on AstroBin in July! This version came with changes to the plate-solving algorithm, because of of the items in the changelog said: - Astrometric solutions are orders of magnitude more robust and accurate. We are now capable of modeling much wider fields and much more complex arbitrary distortions. The failure rate will be considerably reduced, and the generated data will be much more accurate and better adapted to the geometry of each solved image in AstroBin's web interface. So I believe something might have broken a very specific cases. Francesco Meschia: This is correct, and yes, this is the information that AstroBin sends PixInsight: Please note that `imageResolution`, which is the pixel scale, is not your native resolution but it's scaled up because PixInsight receives a scaled down version of your image. Your NGC 3628 is 2400 pixels wide and has a pixel scale of 0.65"/px, which, when scaled to 1824 pixels wide becomes 0.85"/px. I will now try with sending the entire image and see what happens (the concern is with memory usage of very large image and what happens to rendering the annotations). |
...
·
|
|---|
Salvatore Iovene: Unfortunately it fails the very same way. @Francesco Meschia, have you checked if your local version of PixInsight can solve your NGC 3628? |
|
4.07
...
·
|
|---|
Yes it solved with no issues. This is really weird.
|
...
·
|
|---|
Can you see if this solves: https://cdn.astrobin.com/thumbs/8Rs3-bAc0hKx_1824x0_esdlMP5Y.png by seeding this information: RA = 170.072000 Dec = 13.588000 Pixel scale = 0.853947 ? |
|
4.07
...
·
|
|---|
I just ran it, it solved just fine:  |
...
·
|
|---|
Then we reached the end of what I can do. We need to reach out to Juan Conejero to see if he can help. Btw what version of PI is that?
|
|
4.07
...
·
|
|---|
Thanks for trying, this is really a good mystery. Good point about the versions. PI is 1.9.2 build 1633, and the version of ImageSolver is 6.3.1. Running on macOS. Oh by the way: I don’t know if you can see it in the logs, but yesterday (or the day before yesterday depending on the time zone) I did get an advanced plate solution for a revision of this image: https://www.astrobin.com/j6gvmr/K/. Unfortunately I didn’t keep it because of issues with the content of the image itself, but it did solve... |
0.00
...
·
|
|---|
Francesco Meschia: Hi Francesco, the particular image I´m trying plate-solving is roughly around the Virgo constellation with the Markarjan Chain and galaxies like M 87. |
|
4.07
...
·
·
1
like
|
|---|
Thanks for sharing the info. So, not too far from Leo. Makes me think of a corrupted catalog.
|
