How Signal Noise Ratio and DR work.

Started 4 weeks ago | Discussions thread
ForumParentFirstPreviousNext
Flat view
Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 4,621
How Signal Noise Ratio and DR work.
1

This is a small exploration in Signal Noise Ratio (SNR) and also touching on DR a tiny bit.

The comparison is based on a setup I shot back in 2017, with three different cameras.

A scene with a bright side and a dark side, dark side barely visible.

This was shot with three different cameras, Phase One P45+ that was introduced in 2007, Sony Alpha 900 that was introduced 2008 and Sony A7rII that arrived 2015. The Sony are 24x36 mm on Sony CMOS and the P45+ is 37x49 mm on Kodak CCD.

Both sides have a ColorChecker and it has 6 neutral fields. I opened the files in RawDigger and the statistics for the 2x6 grey neutral fields. Let's compare the P45+ with the Sony A7rII:

On the X-axis we see the data numbers in the raw file. This is essentially our signal. The signal devided by the measured noise gives us SNR that is plotted on the Y axis. This is a log log plot needed to be able to handle the large range of value. What we see here is that the P45+ and the A7rII are pretty close in noise at high signals (highlight), but signal noise decreases more for the P45+ at low signals (darks).

Ideal behaviour would be like this:

For a sensor without other sources of noise, SNR would be proportional to the square root of the number of captured photons.

We see that both P45+ and Sony A7rII are pretty parallel with the ideal line but they drop down at low values. That drop depends mostly on readout noise.

So what it tells us is that P45+ and Sony A7rII have similar noise at high exposures, but Sony A7rII has cleaner darks. That is the result of something like nine years of development.

The quantities plotted here are per pixel, but the P45+ and the Sony A7rII are very close in pixel count.

Now, let's add the Sony A900 to the mix. The A900 arrived back in 2008, so it is near contemporary with the P45+.

Here we can see that the A900 is a bit below the P45+ in image quality per pixel.

But the A900 has just 24 MP, let's compensate for that in math:

So, the A900 was a bit behind the P45+.

So, what can we learn from this?

  • One lesson may be that things really work as they are expected to work.
  • Another lesson is that making the sensor larger increases Signal Noise Ratio.
  • Improvements on readout noise affect the dark part of the image.

Something we also may see is that sensor development make small strides with regard to image quality. The great progress is done other fronts, like pixel count, frame rate, video and accurate focus. There are improvements in image quality, but at slower pace.

Some reservations...

There is a bump on the curves when switching between two ColorChecker cards. I would think that my measurement may pick up some noise from the ColorChecker and that noise may depend on the kind of illumination.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

Nikon Coolpix A900
If you believe there are incorrect tags, please send us this post using our feedback form.
ForumParentFirstPreviousNext
Flat view
Post (hide subjects) Posted by
ForumParentFirstPreviousNext
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow