Fix Version/s: None
Sprint:SUIT Sprint 2017-4, SUIT Sprint 2017-5, SUIT Sprint 2017-6
Team:Science User Interface
SLAC has delivered imageServ APIs to retrieve the image cutouts but they have questions about how to provide the cutouts for coadds (per Kenny Lo). They would like our team to provide how-to.
Thanks to Tatiana who points out that IRSA has been using the cutouts algorithm in http://irsatest.ipac.caltech.edu//ibe/cutouts.html. Basically users provide the center coordinates and the size of the cutout. "The size parameter consists of one or two (comma separated) values followed by an optional units specification. Units can be pixels (px, pix, pixels) or angular (arcsec, arcmin, deg, rad); the default is degrees. The first size value is taken to be the full-width of the desired cutout along the first image axis (NAXIS1), and the second is taken to be the full-height along the second axis (NAXIS2). If only one size value is specified, it is used as both the full-width and full-height. Negative sizes are illegal."
Should we suggest SLAC team to consider to adapt the same algorithm for LSST image cutouts, both single epoch and coadd? Need David C. and Gregory to consider this and approve it.
Please see the full document in the link below.
- mentioned in
Write a document to combine IRSA's cutouts strategy doc and comments from D. Ciardi and Gregory.
I have combined IRSA cutout page with D. Ciardi's comments here into a Confluence page, https://confluence.lsstcorp.org/display/DM/Image+Cutout+Description.
Could you all take a look and let me know any changes I need to make.
IVOA's SODA 1.0 has provided guidelines on image cutouts. Search for 'cutout' for use cases and examples in http://www.ivoa.net/documents/SODA/20170310/PR-SODA-1.0-20170310.html.
The draft version is at https://confluence.lsstcorp.org/pages/viewpage.action?pageId=54856011&src=jira
There are a few key aspects that need to be kept in mind in making this decision:
1. Image Edges: the cutout service must be able to cross image edges. The focal plane array is 189 separate CCDs and the co-adds will be stored in some form of tiling across the sky. If the cutout size reaches a boundary it must be able to cross that boundary and build the remainder of the image from the next image over.
2. The images will not (in general) be oriented in coordinates (RA and Dec) and will likely cut across at some angle to the xy-orientation of the images.
3. Related to the two items above: the cut out service needs to be able to handle gaps in the coverage (e.g., gaps between CCDs) and set those pixels to some value that can be read by the visualizers and set to 'black'.
4. The cutout service, as viewed by the user (user here means a person and/or the application that calls the cutout service) must be able to handle multiple calls - the primary use case here is time series. Whether this is a set of serial calls to the cutout or an upload of positions and whether this is specifically a requirement on the cutout service or not is left for the development team to think through.