Summary Vidoe-conference between CFHT and Terapix: Thursday Feb. 22, 2007

Attending: Chiara, Yannick Gregory (terapix), 
 Cuillandre, Albert, Veillet, Arnouts, Salmon, Martin, + 1 CDD (CFHT)

Thank you. This morning's teleconf saw very positive discussions.
Here is a summary of what was discussed.

o Negative xtalk (negative doughnuts) are not properly removed. Terapix
prefers no removal than current removal method. CFHT (Loic and Chi-Hung)
are testing a more promising avenue for removal using the SEx -OBJECT
checkimage. Was decided to test this new method on the D3 Willott images
and iterate with Chiara to produce stacks and evaluate results. To
quote Yannick: "Removing this would would calm 95% of PIs". It is the top
priority.

o Guide window xtalk forming a cross across the chips. Terapix tried
many different ways to mask or remove this feature without much success.
CFHT should try to find methods to do it. The first step is to track the
different evolutions of the problem (the cross pattern tended to follow
the changes in DSP code of which we had numerous evolutions in the first
1.5 year of operation) and address each at CFHT. Second highest priority.

o Vertical lines in chip #60 (ext 4). Is most probably due to on-chip
guiding when resetting the guide box. This has been seen in the lab at
CFHT, is detector dependent and cannot be removed at the source. All
images should show the effect including darks and flats.

o Horizontal lines on chip #60 (ext 4). Origin is unknown. Other chips do
not show this behavior.

o Sky construction. Chiara showed that using a single image rather than
whole cube in producing a sky increases the noise slightly, but within a
tolerable limits (compared to speed gains in processing). However, source
masking should be improved on single images.

o Photometry. CFHT (Stephane Arnouts) will analyse the standard star
measurements to determine zero points of narrow-band filters and
color-term in the Y filter. The relative (chip to chip) zero point
variations seen by Herve Aussel on COSMOS data does not originate from
hardware problems (tests on flat field images showed a very stable --
~10e-5 -- chip to chip offset).

o Request by Terapix that CFHT makes accessible (say on the web) any
engineering test we do. Optics throughput curves are needed for zero
point estimations. Unoftunately, nothing is known about the quantum
efficiency of the detectors.