Realigning after new lights are obtained
If I have it right, when lights are aligned, they are resampled, so a little of good data is lost.
In MaximDL, for example, one can sift through lights based on criteria, and save the good ones to a new folder, after alignmet.
If later later collect new lights, sift, and add to the same folder, and then process that good group, that would be a convenient workflow, but is it a bad practice, or is the dataloss insignificant (pretty pictures, not photometry).
TIA for your thoughts.
- --- "Robert" <robert.hertel@...> wrote:
> If I have it right, when lights are aligned,That's a somewhat complicated topic that hinges on sampling (pixel size as a function of fwhm), number of frames, and method of resampling.
> they are resampled, so a little of good data is lost.
But essentially you will always have to resample all but 1 frame as a part of the registration process. If you try to avoid registration by not offsetting (dithering) the sub-exps then you are asking for much more trouble in regards to persistent noise and outliers; and even if you do not deliberately dither, it is difficult to avoid field rotation and drift, so you should still register/resample.
Maybe you are asking about double-registration? (e.g. register frames from night 1 then register night 2 frames to a different master, then have to register master 1 to master 2). Double registration should generally be avoided and it is easy to avoid. If you have several nights of sub-exp then simply process all frames for all nights at the end of the total run. Or use the combined image from night 1 as the registration master for night(s) 2+. Generally it is best to create a master from all sub-exps rather than create a master from sub-masters unless you really know what you are doing visa-vie weighting the master combine.