DM-6652: Remove database hack #5
Conversation
| # we don't have astrometry_net data (yet) so astrom and photo cal are impossible | ||
| config.doCalibrate=False | ||
| else: | ||
| elif False: |
There was a problem hiding this comment.
Is there any need to leave this in, if we don't do calibration now anyway? We can always add it back in when we do the calibration step, right?
There was a problem hiding this comment.
It's a convenience for the user. A dirty convenience I grant you, but the whole package is dirty...
| @@ -37,3 +38,7 @@ | |||
| config.calibrate.astrometry.solver.useWcsRaDecCenter = False # It's off for some reason dunno yet | |||
| config.calibrate.astrometry.solver.useWcsParity = False # I doubt I guess right | |||
| config.calibrate.astrometry.solver.useWcsPixelScale = False # DGM says it's 0.4, but.... | |||
| else: | |||
There was a problem hiding this comment.
Were there actually Sloan filters available? Same as above, it seems like master should have a clean config and we can add on in a branch as new features are available.
| @@ -6,6 +6,10 @@ | |||
| config.isr.doFringe = False | |||
| config.isr.doLinearize = False | |||
|
|
|||
| config.charImage.repair.doCosmicRay = False | |||
| config.charImage.repair.cosmicray.nCrPixelMax = 1000000 | |||
There was a problem hiding this comment.
There's no need to set nCrPixelMax if doCosmicRay is False.
| @@ -84,6 +86,10 @@ def run(self, ccdExposure, bias=None, dark=None, flat=None, defects=None, fring | |||
| ampExposure = ccdExposure.Factory(ccdExposure, amp.getBBox()) | |||
| self.updateVariance(ampExposure, amp) | |||
|
|
|||
| # Don't trust the variance not to be negative (over-subtraction of dark?) | |||
| variance = ccdExposure.getMaskedImage().getVariance().getArray() | |||
| variance[:] = numpy.where(variance > 0, variance, 100.0) | |||
There was a problem hiding this comment.
How is the value 100 motivated? Seems like it could be configurable or calculated from the data.
There was a problem hiding this comment.
I'll try setting it using a robust stdev of the image.
| @@ -1,4 +1,4 @@ | |||
| from lsst.obs.monocam import MonocamIsrTask | |||
| config.isr.retarget(MonocamIsrTask) | |||
| config.dateObs = "date" | |||
| config.darkTime = None | |||
| config.darkTime = "DARKTIME" | |||
There was a problem hiding this comment.
FYI, this now comes in the VisitInfo, but that's for another ticket.
There was a problem hiding this comment.
Yeah, this was written quite a while ago, and hasn't been modernised.
Now that the FITS files have been "sanitized" (updated with the appropriate metadata), the hack is not necessary, but it actually gets in the way because the database connection prevents use of multiprocessing.
Might be useful for selecting calibs when combining.
This allows us to work with a modern ip_isr, following DM-5462.
It's now available in the sanitized FITS files.
We have biases and darks, so activate the correction that uses these. I'm concerned that this can oversubtract sky data, and so added a guard against having negative variance.
There may be something in the calibs that causes us to find a LOT of CRs, so disable that until we can figure out what it is.
Merlin says that the noise problems have been fixed.
It gives better PSF models and is more robust than our PcaPsf.
Until astrometric matching performance can be verified.
PaulPrice
force-pushed
the
tickets/DM-6652
branch
from
58fdb7f
to
ec81aa7
Compare
No description provided.