Fix Version/s: None
Sprint:SUIT Sprint 2018-10, SUIT Sprint 2018-11
Team:Science User Interface
Based on a discussion in the Tea Time HipChat room today, this is a "note to ourselves" to take a look at the behavior of Firefly when visualizing images with negative flux values.
This is important for difference imaging and is therefore highly relevant to both LSST and ZTF (if ZTF difference images become visible through Firefly at some point).
The behavior of the asinh stretch, in particular, should be looked at.
I propose that we understand why a problem with difference images in Firefly with the asinh stretch would have been present in April 2016; that the problem has been addressed; and that difference images from the AP pipeline display fine now in Firefly, as shown by the images attached to this ticket.
Thanks, David, that's really good news.
Do we have a straightforward procedure for visualizing signed image data and applying a color map that explicitly distinguishes positive from negative? E.g., a mapping where "0" values are always mapped to a neutral hue and positive and negative values are mapped, respectively, to different saturations and/or brightnesses of two specific colors – say, blue and red?
Firefly has a color table for difference imaging, that has the tooltip text "For difference imaging", which maps negative values to blue and positive values to red. Using firefly_client it is possible to specify the color table as a number, but it is not straightforward. Implementing color table support in display_firefly is in ticket DM-15427 where we (Gregory Dubois-Felsmann and I) need to settle on which colormaps to expose and how to name them.
David Shupe There should be a different story to capture this and then we can decide what to do. I think this ticket can be closed now.
The DECam difference image referenced in the previous comment has been displayed using the afwDisplay interface with the matplotlib and FIrefly backends, with a linear stretch from -20 to +20 DN, and an asinh stretch from -20 to +20 DN. Additionally, the FITS image was displayed with ds9 v7.6. The images display similarly, with the exception of asinh in ds9 which is known to have a rathe different implementation of the asinh stretch.
Given that the HipChat discussion occurred in April 2016, and that later a wrapping problem with Firefly asinh and powerlaw_gamma stretches was fixed in
DM-14778, I propose that we may safely conclude that the issues raised in HipChat have been addressed.