Hi,
I am having problems with the image rotating after a meridian flip.
It is as if the camera has rotated, which it hasn’t, so the image taken after the flip is several degrees rotated compared to the image after the flip (this is after rotating the image 180 degrees to get both the same way up).
It’s not usually much of an issue as I crop the overlap out but I’m imaging chip filling targets at the moment and I can’t use the flip which is hitting my imaging time.
Thanks
Jon
I know exactly what you mean. I never noticed it before this effort at a mosaic. The screen capture shows what is happening.
The image centers are as per the F&M wizard - but from one side to the other they don’t line up on a vertical line.
Top right is Panel 1 - original start position. Panel two is bottom right. Panel 3 & 4 (top left & bottom left) have been rotated 180° by me to join up…but clearly 180° is not exactly correct.
There is horizontal & vertical displacement of panels 2,3, & 4 from #1 (apart from the specified overlap) … I have not thought about it enough to ‘know’ exactly what is going on with that.
That’s interesting, so panels 1 and 2 were taken before the meridian flip and 3 and 4 after the meridian flip ?
I haven’t used the flip recently due to this issue but looks like a good night here tonight so will try again with a view to investigating further.
Seems to me the “Rotate selection” feature of the Mosaic Wizard needs to be set correctly to avoid this rotation issue.
@jmacon has it - this is normal, especially on a mosaic spanning something the size of sh2-240.
it’s kind of like a map projection problem. what we are imaging appears to be a sphere to us and we’re mapping it onto a flat, square sensor. so as you progress away from the center of the mosaic, you need to rotate the sensor to avoid following the curved lines of RA and DEC.
you can see in the PI mosaic planner that you need to rotate two of the frames a tad. i had to put in 0.1 degree of rotation in the root frame because i think there’s a bug related to PA=0.
now having said that, the offset in @kinch s mosaic is somewhat unexpected. maybe the center coordinates for each pane were not offset by SGP? take a look at what PI’s script has computed - the RA and DEC of left-right and top-bottom pairs are not exactly the same, again owing i think to the spherical -> flat transformation that has to occur.
rob
Thanks Rob - nice explanation and the PI info is interesting - I had not seen/used that feature. I have to admit to one error on my side that I only realized yesterday when looking at my overall mosaic - I planned it in advance but when shooting the frames, the camera was orientated about 5° different from the input into the planner.
I am not sure what actual effect that has had - but once the full ‘subject’ was covered, I guess I can manage. As for the horizontal & vertical shift in the frames: I have little doubt that that might be a problem on my end - it is too big to have gone un-noticed before this. All a learning experience this end with my first 4 panel mosaic. I won’t be rushing to do a second 
If I ever get this one finished, I’ll be sure to post it here.
yeah please post it; sh2-240 has been on my list for a while but i have never managed to do it (the example image is from rick stevenson…)
i did this 12-panel mosaic of the NA nebula a year ago and intended to finish the SII and OIII this year but the weather was terrible and i never finished… so one more year i guess. on that one i didn’t want to deal with different rotator angles and flats, so i just made the overlaps large and dealt with it…
rob
I think there are two issues here, the issue of rotation after meridian flip and the issue of rotation in Mosaics, but I don’t think they are necessarily related. Although I have observed it myself I’m not entirely sure what causes the rotation after the flip.
The rotation observed when doing mosaics is (as has been mentioned above) a problem related to projection and translating one coordinate system into another. In fact the problem is not so much related to the fact we project a sphere onto a flat surface, it’s more related to ‘where’ we carry out this projection on the coordinate system, in this case it is dependent on the declination of the mosaic. The higher the declination, the more pronounced this problem becomes. You can easily visualize this by imagining doing a 4 pane mosaic centered on the pole. If you were to use the same camera rotation angle for each panel you would be having a really tough time 
The PI mosaic planner does have a checkbox which can calculate this rotation error for each panel but the SGP Framing and Mosaic Wizard does not. So for mosaics at higher declination it would be worth finding out what the rotation difference would be for the panels to avoid having to add big overlaps. Of course this also means that for each panel you will have to change the camera angle, which is fine if you have a rotator but a bit of a pain if done manually.
Since each panel of a mosaic is a separate target in the sequence file, any required change to the rotation (position angle) of each panel could be specified – either automatically by SGP or manually by the user. It would likely be a rather small adjustment between panels. Possibly do-able if using a high resolution rotator but problematical if done by hand.
Charlie
Agreed, unless you are creating a mosaic at high declination, in which case the angles can become larger. Luckily there aren’t many/any mosaic targets that way 
good point about the declination. most map projections wildly exaggerate the size of features at the top and bottom of the earch, which is a similar problem.
rob
