Properly deal with bright defocused objects

[cwwalter]

In testing imSim 2.0 I saw we neglected to handle the case where very bright objects are being imaged on out-of-focus sensors. This includes both the wavefront sensors and intra and extra focal focal plane images. This is due to the switch over to FFT imaging for very bright objects which bypasses the ray-tracing.

The following image demonstrates the problem. Note the in-focus bright stars among the donuts:

[cwwalter]

So far discussions have noted:

• The light is spread over many more pixels in a donut. So, the saturation and full well conditions will be different.
• One approach might be to clip the flux and really do raytracing for bright objects on out of focus surfaces. Bright enough objects might then need to be scaled.
• To do this we would probably need to use batoid to fire some test rays to determine if/where stars are out of focus.

[cwwalter]

For reference these are the limits used by the AOS wavefront estimation pipelines for selecting stars:

https://github.com/lsst-ts/ts_wep/blob/develop/policy/task/magLimitStar.yaml

[sidneymau]

One thought would be to do a fractional fourier transform on these objects to draw them out of focus. This might take some work to implement, however (particularly computing the correct fraction for the transform as a function of sensor height)