Curious about why this happens?

[dmcclain]

I have two images shown below. Same calibration masters used on both of them. On the left is an image that was taken (I later determined) through high clouds. On the right an image through a relatively cloudless sky. Same exposure duration in each of the image subframes.

What I'm curious about is why the cloudy tainted data show strong edge artifacts (top, right, bottom edges), while the cloudless images show no such artifacts. Indeed, if you examine the residuals of the flat field master, after removing the overall dome, you see edge features just like seen in the image on the left.

So, why do you think that cloudy tainted data bring out those faint master flat artifacts?

[dmcclain]

I suppose that the presence of high thin clouds inherently diminishes the contrast in the image, as well as starkly increasing the apparent noise level. I think both of those encourage excessive stretching, which results in bringing up to visible levels those "features" in the underlying flat field master.

I was thinking about this process and realized that there doesn't seem to be any inherent mechanism in PI to limit the amount of contrast stretching to prevent excessive growth of noise features. And noise, when overly stretched, will tend to show plausible structures in the nebular background - just like when staring at a cloudy daylight sky we see "animal shapes" in the clouds.

My data from the cloudy sky shows enormous background levels in the raw frames, compared to more typical dark sky exposures, for the same exposure duration in each. I suppose that should serve as a warning symptom of clouds.