I’ve been fighting a big problem recently with concentric rings of different colors appearing in both my lights and flats. However, the flats do not correct the rings in the lights, and possibly make them worse. Below is a severely stretched stack of over 60 images of M101 that were calibrated with flats and biases.
The rings definitely inhibit my ability to process the images. Some of the parts of the rings are as bright as dim parts of the data, making removing it by a mask or black point correction impossible. Siril’s background extraction mostly just makes it worse. So far, I have been manually fixing it by blurring the image, removing remnants of data, and subtracting it. This still causes me to lose data that I would not if the field was flat.
I’ve done a lot of research about this, and there doesn’t seem to be a definitive explanation for it. It seems to mostly affect Sony and Nikon DSLRs. The effect is enhanced by using lower ISOs. People posit that this could be due to some kind of in-camera cooking done to the image, possibly from the red and blue white balance scaling that apparently can not be turned off, or some kind of vignetting correction that is supposed to not occur without a lens, but may anyways. Others suggest this is simply quantization of vignetting that occurs with the extremely dim, low range data. No matter the explanation, it spoils otherwise good data and makes post processing a nightmare.
Many suggest the best, or only, solution is to increase the ISO and, therefore, decrease the exposure. This would allow finer sampling at the dim and low range values that the sky glow and deep space objects live at. I’m definitely no expert on this, but I imagine shot noise may increase since there is less “smoothing” going on by rounding in the ADC process at a higher ISO.
As an example, take a recent un-debayered raw frame of M101 that was taken in 180 seconds at 200 ISO. I linearly stretched the image to fit the 16 bit histogram in Siril. Even with a decently long exposure, the sky fog and dim target data lie below 1/8 of the range. I can imagine that the vignetting in this small region of values could be quantized to a noticeable effect that would be visible once greatly stretched, as is done in processing astronomical images.
I’ve done an experiment by taking single exposures of my ceiling at the same focus I would image at. I’ve taken images at 200, 400, 800, and 1600 ISO. The exposure was halved as the ISO was increased. The histogram was about halfway across on each image, but the histogram moved left as the ISO went up and exposure went down. I guess the ISO isn’t really linear.
I debayered the images using Siril, did an auto-equalization in GIMP to equalize and stretch the channels, and then decomposed the images into HSV. Below are the H channel of each image.
You can clearly see the presence of the rings in the 200 and 400 ISO images. As the ISO is increased, the ringing effect diminishes.
Many people suggest that this could be caused by internal reflections, filters, or other glass such as a coma corrector. The above images were taken without the LPS filter that I typically use nor with the Baader MPCC coma corrector that I use. The camera was mounted with a simple 2″ T-adapter that was fully threaded inside.
I recall that I haven’t always had this problem. When I first began with a dobsonian and equatorial platform, I took short images of 15 seconds or less at a high ISO, probably at 3200. As I got better at tracking the sky, I would up my exposures, and therefore decrease the ISO. Since I got my new German equatorial mount, I’ve been using 200 ISO and three minute exposures, with the color aberration appearing the worst yet. Next time, I’ll try 1600 ISO and a shorter exposure. It might not be all bad, though, as satellites and planes will not ruin as much data!