Dark Frame Noise Levels

[cjdawson]

Finally found time to process the info that I took from my camera.

 

Here's the raw data for my 70D.  It's interesting that I've been imaging at ISO 800 in the past.   Looking at this, graph, I think that's the right figure to settle on, it's just as the Mean starts to increase in a curve.

 

Here's the data for a series of stacked frames.  Over the stack of images, it appears that ISO 800 gives a good balance. It is interesting to see that the mean figure is higher than for a single frame.

The thing to note about this is the the total exposure time for each stack is the same amount of time.  So for ISO 100, that's 2 frames at 256 seconds each, for ISO 800 that's 16 frames at 32 seconds each.

 

For those that want to see the numbers, here's the raw data.  The temperature is the temperature of reported by BackyardEOS for the first image in the sequence.  Each image had a 10 second gap between.  The whole set of images were produced in a single run.  I did this deliberately as this is how I use the camera for a real image sequencing run.

[Starwiz]

Very interesting stuff.  I've been running some more of my own trials, but need to run some more still.  I'm finding that even a relatively small change in the temperature can have a significant effect on the noise so need to show this effect somehow in the results.

I've also settled on ISO 800 for the time being as it 'feels right' looking at my current data.

John

[cjdawson]

Yeah.  I think that's an issue in my data as well.   I'm wondering about finding a way to stabilise the temp so that it can be controlled.  Think in reality it's beyond that capabilities of what I can get my hands on.

[Adam J]

Do not forget that dynamic range is reduced as ISO increases, its not all about noise . 

[Starwiz]

On 12/7/2016 at 13:33, Adam J said:

Do not forget that dynamic range is reduced as ISO increases, its not all about noise . 

Yes, that's something else I've learned since initiating this post.  Any thoughts on how Dynamic Range can be measured?  

John

[Adam J]

2 hours ago, Starwiz said:

Yes, that's something else I've learned since initiating this post.  Any thoughts on how Dynamic Range can be measured?  

John

Measured? I would guess you could just take a picture of a white object and increase the exposure untill the pixels values reached their maximim then repeat at different ISO. However, essentially dynamic range reduces by a factor of 2 for each ISO step past unity gain.