Examining the stddev of a single pixel in a series of images

[Nocturnal]

I see. Thanks for explaining. What was it doing with the 0 sized rectangle? Should it have raised an exception or did it start using the entire image?

[Nocturnal]

OK by grabbing a single pixel I get fractions that when multiplied by 65535 result in whole integers. Not sure why that didn't work with the mean of a single pixel rectangle but fine. Trouble is that the data I get is really bad. I took a pixel from the middle of a sequence of 3 minute darks taken at unity gain. The average was 320 and the stddev just about 26. For a sequence of bias frames for the same pixel the average is 270 with an stddev of 24. Looks like all is lost by the time bias is set.

I need to do some more measurements but this doesn't look very good. Thanks for making the script so I could do this test.

[pfile]

you missed what juan said about what happened with a 0 pixel selection:

By passing an empty rectangle to Image.mean(), you are computing the mean of the whole image (actually, of the current selection, which is the whole image by default).

rob

[Nocturnal]

Indeed I missed it, thanks. That explains why the result was so much smoother. It has since been explained to me that the 12b to 16b conversion adds 4 LSBs of data rather than simply padding the MSBs with zeroes. This leads to large swings in the bottom part of the data. I'm still trying to wrap my head around this.