I ran both sets of data through the cp_verify code, comparing all exposures to the current set of certified calibrations. The results seem to match the expectation that there would be no changes to the images. I've attached comparisons of the gains and read noise determined from the PTC data, along with the list of processing commands. Verification notebooks are available at the SDF: /sdf/home/c/czw/dev/
The median gain difference is 0.002, with stddev of 0.003. Median read noise difference is -0.017 ADU, stddev 0.344 ADU.
The flat verification is dominated by dust spots/illumination differences. If there are concerns about the flats, I can generate a combined calibration from one of the firmware sets for verification, but there do not seem to be any other differences.
The residual ISR-processed bias and dark images look as I would expect, and residual mean plots by amplifier are attached. These plots add a small x-shift so different amplifiers do not overlap. C11 and C14 are the two problematic amplifiers in the dark residual plots. C11 has historically had large scatter (it is the amplifier with the hot column), but C14 was corrected well in the 30s data used to generate the current combined dark. This likely means an updated dark calibration would be good.
I've also attached an image of a section of the residual bias image. There are chains of hot/cold pixels that appear to be imprinted on all amplifiers at the same amplifier-coordinates. I don't think I've seen this before, and so wanted to report this feature. These chains are visible in both the bias and the dark, and with both sets of firmware. They do not repeat in the same location on different exposures, but all of the images I looked at had at least one chain per amplifier. The signal seems to be at most +/- 50 ADU, with a rough by-eye sigma ~ 25 ADU.
Let me know if anyone would like more analysis, but as mentioned above, the firmware change seems to have no effect on the images produced by LATISS.