• burntool needs to be modified to do the fitting from the first non-saturated pixel.
• Better dark detrend spacing. I don't think we're fully catching the time variations.
• There seems to be a column style crosstalk between OTA01xy*2 and OTA21xy*2 and to a lesser extent to OTA11xy*2. This appears to be reversable as well, with a bright star on OTA21 printing to the other OTAs. OTA41 doesn't appear significantly affected. I don't see a similar result on OTA67-OTA47, but that may just be missing a bright star in the sample I'm looking at. Examples: o5677g0123o, o6042g0256.
• There are issues with the fringe correction on a number of OTAs. There seems to be a time based component to this as well, as 2015 data look reasonably good compared to older data (based on ippmonitor jpeg images).
• Fringe construction doesn't fully respect masked pixels due to the convolution. This leads to ringing around those defects.