Opened 15 years ago
Closed 15 years ago
#56 closed defect (fixed)
Problem with NCP projection
| Reported by: | MatthewWhiting | Owned by: | MatthewWhiting |
|---|---|---|---|
| Priority: | normal | Milestone: | |
| Component: | Input | Version: | 1.1.7 |
| Severity: | normal | Keywords: | |
| Cc: |
Description
Something with the NCP projection in a file from Attila is causing a crash at the point of making the detection map.
May need to look at the way pixels outside the coord system are dealt with. Attila will provide the FITS file to look at.
Change History (4)
comment:1 Changed 15 years ago by
| Status: | new → assigned |
|---|
comment:2 Changed 15 years ago by
comment:3 Changed 15 years ago by
Found a bug in plotting.cc, where the moment map was being calculated for blank channels as well. This meant that for the blank channels outside the universe, the transformation was being applied and causing the error. By limiting to only non-blank channels, we can avoid that.
Still need to check that choosing channels based on the first pixel is OK -- I think this has been dealt with elsewhere. (i.e. what if there is a good channel that has the first pixel blank?)
comment:4 Changed 15 years ago by
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
This approach matches what I've done elsewhere, so for the purposes of this ticket, I'm treating it as fine. Ticket is therefore closed.
OK, part of the problem arises from the fact that wcslib converts NCP projection to a SIN projection. It is not obvious whether there are checks to make sure going from pix to world coordinates involves a point that is in the world.
One thing that may need to be fixed is that this conversion is done by wcsset(). However, it is likely that the various parameters need to be cleaned up by wcsfix() after this. Currently this is done before wcsset(). Make sure it is done appropriately.