Open bnord opened 1 year ago
As soon as we can, we should let go of FITs.
Do we need store in butler, or should we think about just making pngs and sending them around (between rubin and zooniverse) without storing in butler?
The case I can think of for why we'd want to store in the bulter: would this is be a "good" way of keeping the original fits files that the pngs came from for analysis after processing?
There is a very vague reference to retrieving pngs "stored in the butler" here: https://dmtn-237.lsst.io
This has been discussed between myself and the former head of EPO, and DM (Frossie, Tim Jenness, John Parejko). Last I heard they had promised to deliver this functionality, but the work has been stuck in development limbo. Relevant artifacts:
https://jira.lsstcorp.org/browse/DM-31836
Snippets of my conversation with John P.:
"Basically, I wrote most of the 3-color cutouts code, Frossie told me “we’re not scoped for doing that much Zooniverse processing in the image cutouts service so I don’t think it can live there, and where would they store the output”, and I’m not equipped to answer any of those questions." ... "My proposal was to output PNG or JPEG, but part of the pushback I got was that the vo-cutouts specification only allows FITS. I think that we should improve the VO spec when it is not sufficient to our needs, but this is part of that broader discussion I suggested above."