help on measuring noise on DSLR and IPhone please

[Spik]

test method:
I took a completely black photograph at night in a dark room on my Canon 1200D with a lens cap on to capture an absolute black image. this image was later put into post processing software to show the peak blacks. the following image is the result, ofcourse the remainder of the image is the noise in the image.

noise pattern 800D

• Spik
• 1 May 2020

800D noise pattern

later I repeated the same test with my iPhone 11 and got a completely abstract noise pattern. the white sections being the noise and the black sections being the complete black.

iphone 11 noise

• Spik
• 1 May 2020

iPhone 11 noise

the Canon shot in RAW and gave a natural noise pattern (highly compressed may I add) whilst the IPhone shot in JPEG and gave blocks of noise...

I'm assuming that I'm looking at noise due to the fact that both photos were taken in absolute darkness.

can anyone help explain the result from the iPhone? is it due to the fact that it had been shot in JPEG or is there another factor I am unaware of?

I used some PSNR software to determine the noise level also but it pinged back with a decimal number which I don't know how to interpret?? any helpful links or advice?

this is part of my university assessment if you are curious... any help would be greatly appreciated.

many thanks
Bradley.

 

[Canon Bob]

Welcome Bradley......

I'm not a phone camera user and the following is simply based on papers/reports that I've read.

The 1200D captures a single image and you can process the raw file as you see fit,,,changing the noise profile alongside many other properties.

When it comes to the iPhone, things works quite differently. The phone continuously captures a series of images from the moment that you enable the camera function. When you take the photo it already has a whole buffer full of images to work with and starts processing. This entails picking the sharpest, the brightest, the best dynamic range, the cleanest etc.
After applying various filters based on the perceived scene....eg, skin smoothing if a face is detected......it then combines the stack to produce the final jpeg. The noise reduction is selective and not global....ie, it may apply different NR in the dark areas of one of the stack whilst smoothing the lighter areas of another.

In short, the 1200D gives you the image that you took whilst the phone camera gives you the image that it thinks you wanted. I can't explain the block noise other than to theorise that the phone doesn't have enough data to make its normal intelligent analysis.
HTH
Bob

 

[Karl.t1965]

There are a few things that will affect noise levels, ISO setting is probably the most obvious but exposure length also has a massive impact on digital noise.
In general (ish) the tighter packed the sensors the more prone they are to noise, a 5DS-R will not fair as well as a 1DX but both are designed for their specific job, design ethos of the manufacturer will also play a big part too.
Sometimes if you want to get away from the impacts of noise you just have to take a step back to old school film. 100 asa film is 100 asa film if it's shot at 1 second or 20 minute exposures. I don't know many digital sensors that will survive a 20 minute exposure and give a clean image