Add mixin for preimaging

sdss / marvin

Data access and visualization for MaNGA. http://sdss-marvin.readthedocs.io/en/latest/

BSD 3-Clause "New" or "Revised" License

55 stars 32 forks source link

Add mixin for preimaging #401

Open albireox opened 6 years ago

albireox commented 6 years ago

Requested in #399. This is a good an easy idea to implement. For DR15 we would need to make the preimaging public but we can cross that bridge later.

albireox commented 6 years ago

Looked a bit into this. To implement we need to:

[ ] Make the current MaNGA Preimaging available in the SAS.
[ ] Modify the tree definition of $MANGA_PREIMAGING to point to trunk/data.
[ ] Modify the sdss_paths.ini definitions for mangapreimfits and mangapreim. Use %definitiondir instead of designgrp.
[ ] If we want this available for public releases, make $MANGA_PREIMAGING available in the dr15 SAS.
[ ] Create a mixin that downloads the file, opens it, and returns the astropy FITS object.

havok2063 commented 6 years ago

I think this should be incorporated into a new Image tool (or vice-versa) that I've been wanting to implement for a while now. The preimaging could be a mixin on this tool. This tool would do things like

[ ] Grab and display the full color png image
[ ] Allow overlays of fiber positions
[ ] Be connected to the Cube tool
[ ] Potentially allow overlays of individual maps

This tool would also deprecate the image utilities that currently exist, which were stop-gap solutions.

albireox commented 6 years ago

That sounds good to me but just to be clear, with preimaging I mostly mean the FITS file that is a cutout from the NSA imaging in the griz bands. The preimaging jpeg image that is created when the preimaging is generated can be included in the Image class. However, there is probably very little difference between that jpeg and the cutout we already have (the one the Image download tools get) so it may be a bit confusing. Or maybe we should have the Image tool switch to use the preimaging jpeg.

havok2063 commented 6 years ago

Is it interesting enough to include in the Image tool? I could see the tool being used to interact with the full color image, and the individual griz bands fits images. Even though they are individual bands, in essence, it's still an image to display. The pre-imaging jpegs are tiny little things. I think the main image could be the larger full color image cutout, but we could include the pre-imaging jpeg for completeness. I can't recall where the Image Cutout service pulls its images from, so there may be subtle differences between the two. I guess it really depends on how closely you want it connected to the other Marvin Tools.

albireox commented 6 years ago

I'm not super concerned with where it lives as long as it is connected with the Tools. I agree the preimaging is ultimately a series of band images, but they contain more information (a header, an ivar, a PSF, etc) so I think we'd need a different class, maybe a sublcass from Image.

Not sure what you mean with the larger full colour image. The preimage includes everything within several effective radii of the galaxy (same as the cutouts) so not sure what else we'd be showing.

havok2063 commented 6 years ago

Yeah that could work as well. I'm not too concerned either if we want to keep them separate.

I mean these images https://data.sdss.org/sas/mangawork/manga/spectro/redux/v2_4_3/8485/stack/images/1901.png compared to these https://svn.sdss.org/public/data/manga/mangapreim/tags/v2_3/data/D0089XX/8980/preimage-1-209232_irg.jpg

albireox commented 6 years ago

Ok, that makes sense. We can provide different "zoom" levels.