Why are my photos noisy - discussion

David, good source of information.

www.rp-photonics.com/image_sensors.html

By "Poisson distribution", do you mean frame to frame for the same photosites, or do you also mean spatially, within a single exposure?

The former, I would assume, substituting pixels for photos. Too many outliers for the latter.

Sorry for being unclear.

I meant that each pixel on the sensor has its own activation energy, depending on nearby crystal defects such as dislocations and out-of-lattice doping atoms.

So each pixel has its own (Arrhenius) rate constant for the accumulation of dark current electrons. This rate constant, multiplied by the exposure time, gives the mean of the Poisson distribution of the accumulation of dark current electrons for that pixel.

Every pixel has its own Poisson process.

But...

Most pixels will have very similar (low) rate constants. A small proportion will have higher rate constants. A smaller proportion will have even higher rate constants. The distribution of such rate constants is presumably studied by sensor designers, but is unknown to me.

These per-pixel rate constants aren't fixed over time. Some crystal defects fix themselves - again following an Arrhenius rate law. New defects can be created by radiation.

I guess it's possible that recording 8k video for a few hours would reduce the dark current noise in your astrophotograpy images. I have no evidence for this. 😀

I noticed when using my A7Rii for exposures of several minutes that it didn't seem to need LENR. Shots without it seem fine despite many complaints to the contrary. Then one day I got caught photographing LE on a hot sunny day. And the camera exhibited all sorts of hot pixels. Perhaps the complaints originated from hot countries.

I have followed this thread and others about ISO/exposure (they can be identified by their length) and find that I am more confused than I was before about how I should set ISO on my Canon R6. I am not totally thick, but under Beginners' Questions I would expect a more prescriptive answer than a discussion of Physics fundamentals, particularly when the participants appear not to agree on them and when some take an adversarial stand towards others.

What can be done to give simple answers to simple questions? I personally have had enough of the difficult answers, though I may be in a minority in this.

:) :) :)

Edit Perhaps a thread entitled Beginners' Answers would be the appropriate answer!

David

Unfortunately the posts after the first few should have been moved to the Beginners Questions Discussion Forum where the more technical discussion is meant to be had.

Maybe if we just stick to my OP which I tried to explain noise and indirectly how that affects how to set ISO.

Is there anything in my OP that is not clear?

You mean, apart from it being too long and involving concepts that I dont believe the average beginner is familiar with?

As I wrote above, Beginners' Answers would be more appropriate.

:)

David

I didn't discuss any concepts that in my experience average beginners have had any significant difficulties in understanding.

Which concepts in my OP are you referring to?

IMHO "Signal to Noise Ratio" may need an explanation, a separate article, a link, something. The article might benefit from including images to illustrate the problem and to prove the proposed solution works.

Yes, that is a good idea but I know the likes of you, bobn2, JimKasson, Michael Fryd (not necessarily in any order) are far more qualified and experienced than me to write such an article.

To be honest, being a layman enthusiast photographer my interpretation of Signal to Noise Ratio doesn't go beyond words to the effect of "Ratio of Good Data to Bad Data".

Signal includes noise, it's good + bad data.

ok, ok 😊 Now you are really over complicating things. I'm sticking with just Good data to Bad data ratio for the sake of maintaining what sanity I still have at my 60+ years of age 🙂

You don't need to stick with wrong. It is very important to realize that signal is everything, including noise.
As to over-complication, I'm not suggesting that you dive into why, under certain circumstances and for certain purposes, noise isn't bad data.

Let's see if at my 68 years I understand this. Signal is overall data. Noise is bad data. SNR is an indicator of how much bad data in overall data or how much good data in overall data.

How do you define “bad data”?

I'm not far behind you 🙂

SNR - Signal to Noise Ratio - is a ratio or a fraction if you like - S/N. If the numerator (top number, S) is larger than the denominator (bottom number, N) then there is more Signal than Noise in the data.

It's the data I hope my noise reduction app changes to good data 🙂