The currently implemented donut-deblending approach is a non-iterative approach, i.e., a least-square fit where a template is used to model the donut images on the image stamp. The input parameters are the intensities/magnitudes of the donuts and their centroids. The initial values come from the  bright-star catalog. The template could be a nearby isolated donut, or a pre-generated phosim image.

Completion criteria:

• run cwfs on the deblended donuts, and compare the results to Zernike truth (from Phosim)
• when we add these wavefront solutions (using a TBD weighting mechanism) to those from isolated donuts, do we get improved wavefront measurements?
• for high/medium/low density fields, can we achieve good performance using these tools?