drizzlepac
drizzlepac copied to clipboard
spacetelescope
Updated WFPC2 drizzle parameters for HAP
Issue HLA-1052 was created on JIRA by Jennifer Mack:
https://github.com/spacetelescope/drizzlepac/tree/master/drizzlepac/pars/hap_pars/svm_parameters/wfpc2/wfpc2 JSON files for WFPC2 drizzling were delivered in Dec 2022: N=1 N=2 N=4 N=6 TOTAL
These should be replaced with JSON files attached here. (Note that the prior N=6 JSON should be removed.) N=1 N=2 N=3 N=4 TOTAL
The primary change includes changing skymethod='localmin' from the prior 'match' which did not work well for the overlapping chips.
NOTE these parameters work well for the WF chips only. If treating the PC chip separately (see HLA-1031), only one parameter driz_sep_scale needs to be changed from its current value of 0.0996 (for WF) to 0.0455 (for PC).
Comment by Jennifer Mack on JIRA:
Based on the 27July2023 regression test, it looks like I will need to update the final_bits parameters for N=1 and N=2. https://hla.stsci.edu/HAP/hapwfpc2view.html
The choice of parameters will depend on how easy it will be to implement HLA-1061 which allows for a different final_fillval option. Currently, we treat DQ=2 as good, which is ideal for bad pixels but not for the region between chips. ( Example dataset attached for N=1 input frames. )
Instead of final_bits=65535 (1+2+4+8+16+32+64+128+256+512+1024+2048+4096+8192+16384+32768), which ignores ALL dq flags, we want final_bits=65533 which treats only DQ=2 as bad. Similarly for N=2 frames, 61439 will be changed to 61437.
On the other hand, if HLA-1061 is implemented, the final_bits parameter will be set to 1024, as for N=3+ and the final_fillval will be use to set pixels values with zero weight.
Shall I file a new ticket or append to this one? I'm still testing, so its possible I may uncover other issues...
!image-2023-08-02-16-37-36-154.png!
Comment by Steve Goldman on JIRA:
Thanks for looking at this. I would suggest creating a new ticket linked to this one.