Hmm, when looking at my Bias frames info in KStars and comparing to yours, all the values are pretty much the same - except Median. Yours shows 0 mine shows 526 point something in the Histogram tab but 0 in the Statistics tab. Should I be worried?

I don't think so. When I look at the results that I get I'm not worried at all.

I don't know....

I think Andy Grove was on to something when he coined the phrase:

"Only the paranoid survive"

Maybe you should be afraid.... VERY afraid....

Hi,

the problem with empty bias frames can be caused by a wrong offset. The default value of 8 is too low, and the sensor's dark current compensation kicks in and clips the noise to zero. Offset should be chosen in a way that a bias frame has a mean value of about 500-1000 (on a 16bit scale). For my ASI294MC, it was offset 14 (at gain 0, might be different for other gains).

I just noticed this too...
My bias frames used to be not empty before mid-October last year.
Since then they are empty. Totally empty.
ASI294MC. Gain 200. -5C.
I used to use an offset higher than default (40 vs 8 ). Don't remember when I switched to the default of 8.
Just checked:
at offset 35 bias histogram is still clipped.
at 37 it looks good.
Interesting thing is that the histogram (in Ekos) looks exactly the same for exposure times 0.01-0.0001 sec.
Not sure what's going on...

Concerning bias frames I noticed that the Indi driver (or the capture module in Ekos) sets the exposure time to the lowest possible value if you set the frame type to "bias", regardless of what is set in the exposure field in Ekos. Which makes perfectly sense in my opinion. You can check your FITS header data if there is your exposure time (0.01-0.0001) or always 0.000032. That would explain that all histograms look the same.

And yes, offset is an important parameter at least for some sensors. I'd like to have it next to the gain setting in the Ekos window instead of the Indi driver window.

Good point! on the exposure for bias frames being always at min of 0.000032. It does make sense.
So, can I use bias frames taken with the offset of 40 with the light frames taken at lower offsets?
Thanks!

Just did bias frames with a Nikon DSLR through Ekos. Set the exposure time to 0.00025 (1/4000). Resulting nef images look good but exif shows exposure times of some random numbers around 11s. Weird.

In general, bias, darks, flats and lights should be taken at same offsets, and gain, and temperature. Unless you use PixInsight which can do a dark optimisation where the dark temperature (and dark exposure) not necessarily has to match with the lights or the flats.

But if you already have lights with an offset of 8 and therefore too low for meaningful biasses, I would just take darks with the same offset/gain/temp as the lights and only do the dark subtraction. Use many darks (20+). Any fixed pattern noise is part of the darks (if it's not clipped due to the small offset).

If you subtracted biasses taken with a higher offset than your lights have, you would "eliminate" noise patterns which are not present (because they are clipped) in your lights due to their lower offset.

Can you let Ekos create FITS format with your DSLR? Maybe one can trust the FITS header more than the Exif data.

Did some more testing with my ASI294‘s bias frames. After powering on the cam and doing bias frames with shortest possible exposure I get correct frames with a median value as expected (depending on the offset I chose). But once I do exposures > 1 s (all fine with them) and then do bias frames < 1 s again, these frames are all zero. The magic threshold here is 1 s; above it‘s OK, below it‘s all black. Until I powercycle the cam.

In light frames > 1s I notice an all zero pixel column to the very right of the image. Maybe they are opaque covered „dark pixels“ for doing an automatic dark correction? Some CMOS sensors have this feature. Maybe some sort of auto dark offset correction kicks in once the first exposure was taken > 1s? For astro purposes, I think this should be deactivated in the cam‘s firmware (or whatever code is responsible for that)...

Anyway, I now do only use frames > 1s exposure, be it light (of course ), dark, flat or flat-dark. And I don‘t use bias frames at all. The 294 has so high amp glow, it doesn‘t help being able to scale bias-corrected dark frames as PixInsight can (for example). And in the end, my images look quite OK, so no problem... But I would like to understand that behaviour.

CS, Bernd