Cmos 4945

[Peter Ward]

I did some more OSC CMOS tinkering in the backyard last night

the result is here (http://www.atscope.com.au/BRO/gallery489.html)

I wanted to explore longer sub-exposure times (5 minutes) which gave the result above. Processing the data I noticed a subtle "ripple" in the CMOS dark frame data
...which I've also highlighted via an animation on the page linked above.

I suspect this is intrinsic to CMOS chips, but would be interested to hear from other CMOS camera users as to their experience with this low level noise pattern.

[gregbradley]

That's a good image regardless of noise floor. I find the same pattern noise (the banding which I assume is an artefact of the backside illuminated sensors) drops away easily with even a short exposure. 2 hours is amazing and shows how high the QE is.

To get a CCD image that good is going to take more than 30 minutes each of LRGB I would think.

I did some flats last night and they showed none of that fixed pattern noise.

I first noticed that banding shifting on some ASI6200 darks shown on a thread in the equipment section. I compared them to my FLI Microline 16 darks and they don't show that at all. So its a CMOS thing at this stage and perhaps a backside illuminated thing.

I think its something you could live with though as it should disappear with a decent exposure. Also dithering is probably important for this.

Greg.

ps. totally jealous of the 130 GTX. I would love one of those. Like an FSQ130.

[Atmos]

It’s come up pretty well, I was hoping to get some time on this with the Mewlon but the skies keep clouding over!

I guess that as long ad it remains randomised it’ll stack out pretty cleanly. It shows that it is a camera that requires a fairly heavy dither I think.

[Peter Ward]

Ta Greg.

I did some digging.

Richard Wright at Sky & Tel touched on this in the May 2020 edition.
I don't think it's a BSI sensor issue: its a hallmark of CMOS i.e. non-repeating fixed pattern noise...hence won't totally dark subtract out.

The fix being dithering with lots of frames...but we already knew that ! :)

The GTX will probably get a more serious work-out soon, the CFW for my KAF16200 camera is finally on its way :cool:

Apologies in advance for the inclement weather that will surely follow...

[codemonkey]

I don't think it's inherent in all CMOS sensors, but it's common. I've seen it on 2/3 of the CMOS sensors I've used.

The IMX178 was the worst for it; I didn't do much NB because of it and stuck mostly to LRGB.

The 183 is similar to the 178 but jacking up the gain seems to get rid of it, which is great because I use higher gain for NB anyway.

I don't think dithering helps with this because it shifts every frame. Also, it doesn't seem to average out very well; even with many frames I'd still see it in low-signal images. Key thing is to swamp it with enough signal. I had no problems with the 178 and my L frames, but background areas on NB subs was something else...

The ASI1600 suffers from no such banding.

[peter_4059]

Do the medium-bright stars have dark centres or is it just me?

[Peter Ward]

Not your imagination.

It's simply a re-bound artifact from a shallow wavelet application....

I'm still getting a handle on OSC CMOS imaging....the lack of tidy-up is not something I'll commit seppuku over ;)