Guided tour of a calexp object

[stargaser]

Ready for review.

Addresses #52

• Read in a calexp, show various methods and how to display images
• Show how to make cutouts various ways
• Repeat for a coadded image

[kadrlica]

Thanks @stargaser, this is very useful!

Something that I found confusing about lsst.afw.image.Exposure object hierarchy is that by default everything is done by reference. This makes sense when accessing the image, mask, etc. attributes, but becomes non-intuitive when creating cutouts. For example:

bbox = afwGeom.Box2I()
bbox.include(afwGeom.Point2I(xmin, ymin))
bbox.include(afwGeom.Point2I(xmin+width, ymin+height))
cutout = calexp.Factory(calexp, bbox, afwImage.LOCAL)

This will provide cutout, a new ExposureF that is still connected by reference to it's parent calexp. I think that it would be worthwhile discussing this a bit in the calexp tutorial. Specifically a discussion of the Factory and clone methods, and some discussion about the deep argument in the Exposure and MaskedImage constructors.

[drphilmarshall]

Hi @stargaser ! Is this now ready for review? Great if we can merge it today, if possible. Feel free to request code reviews from anyone from the @LSSTScienceCollaborations/dm-stack-club - likewise others, feel free to volunteer to work through David's notebook!

[stargaser]

I need to address comments from @kadrlica and I think I need to rebase README.md. Let me try to do that quickly, early this afternoon.

[stargaser]

Ready for review now @drphilmarshall

[drphilmarshall]

Thanks @stargaser ! I verified your calexp tour on v16.0, and put a note on what I needed to change. I also standardized the header. Good stuff, thank you!

[PaulPrice]

Much preferred these days:

cutout = calexp[bbox, afwImage.LOCAL]

You can drop the LOCAL because you're not dealing with a subimage, making this even more cleaner:

cutout = calexp[bbox]

[PaulPrice]

Don't need immediate=True.

[PaulPrice]

That's because they're the same class, lsst.afw.image.Exposure.

[stargaser]

Oh, I see now, that makes this result expected!

Coadds have the same methods, but it seems to me that they have different contents...e.g. XY0 is non-zero; many more cosmic rays in the masks; overall masks look much different; PSFs must be an average from the coaddition process? And certainly the dataIds are a lot different. Maybe this part of the notebook needs to focus on those aspects.