QUOTE (Brian Swift @ Aug 22 2022, 01:14 PM)
Interesting. So is the 102 bias I see in the raw data a "black level" that is normally present but gets subtracted off before companding?
Yes, all of our "digital CDS" CCD systems going back to the original MS98 MARCI have had this, it seems to be intrinsic to the CCD output. The DC offset has to be compensated for if you use sqroot companding. You can perhaps find indications of this if you look at papers, the PDS docs for MSL and M2020, etc. hard enough (or not.)
QUOTE
Has the current "black level" been used for the entire mission?
Yes, and this image shows that it is unchanged since pre-launch.
QUOTE
Looking at this, if there was a mode that just kept the lower 8-bits (no div 2, no clamping), I think it could be turned back into a
full range non-commanded image, since Jupiters brightness varies fairly slowly.
There's a mode like that for MSL/M2020 though AFAIK nobody has ever used it. There's no such mode for Junocam, which uses a piecewise-linear compander instead of a table lookup compander to save parts and logic in the JDEA.
The reconstruction method you describe might work for a smoothly-lit bland object. The TV system on Mariner 6/7 had a "digital video" stream that consisted of truncated samples like these, but the reconstruction method had problems and the whole concept was generally thought to be a bad idea and was never used again IIRC.
https://nssdc.gsfc.nasa.gov/nmc/experiment/...id=1969-030A-01