Notes from interview with Aaron Roodman


We should talk to Tony Johnson, too who is the lead for camera control systems, which need to have image display.

We are still heavy users of ds9.

We use single amps, sensors, or rafts. Scaling is important.


Use cases:

Looking at overscan is important

Need to compare images.

Use zscale and change limits by hand

Are heavy users of ds9's  "smoothing."


We use matplotlib too:

The Software uses matplotlib for display. Test report.


Sometimes look at raw images.


The camera team started using JupyterHub as well.

Aaron is probably the biggest user.

No xwindows, and not having ds9 is a pain


What's missing:

Can't load a whole focal plane or even a raft.

Full focal plan viz is very important.  ds9 struggles with a whole raft.


Therefore, Tony has gotten working another image viewer that can change the resolution on the fly!  This thing is super fast. It's So much faster than ds9. He started with the general viewer and has been adding features to it.

Median signal and noise. Overscan and bias subtraction. Local zscale.

e.g., Bias images show correlated noise visible to eyeballs.


We 'don't care about zooming in to individual objects so much.  We 'don't care about sources. But zooming out is very important.


What have you used for other surveys?

DES

I worked on active optics for their camera, fit donuts, worked on commissioning. Have experience with engineering images, which has different needs. DECam did have a heartbeat image, that flashed up as you took data. It was only good to check that everything looks OK. e.g., tracking. 'Couldn't zoom in. But they would have if they could.

We used lots of DS9.

During commissioning, you look at objects. But not much after that.

Erin Sheldon organized an effort to scan images with eyeballs: https://arxiv.org/abs/1511.03391

We showed lightly processed single post-ISR CCDs at a time.  There was a graphical interface where you could click and flag regions that look weird.

e.g., little diffraction spike along the serial and parallel. Noticed reflection. "Why do we get the reflection pattern from this bright star?" ""satellite trail finding code is not up to snuff."


It helped us discover where we needed black paint.


People on DESC would be happy to look at images for you.