Nikon D5300 Test of Low Photon Rate

Started 1 month ago | Discussions thread
ForumParentFirstPrevious
Flat view
sharkmelley
sharkmelley Senior Member • Posts: 1,969
Nikon D5300 Test of Low Photon Rate
7

There was discussion (actually a disagreement) on another thread about how a sensor behaves for very low photon rates. You can find the discussion starting here: https://www.dpreview.com/forums/post/62303278

Roger's prediction (by simulation) is that with a gain of 2e/DN and a read noise of 3e, quantization issues will prevent faint signals being recorded. You can see this in the top row here:

Figure 10 from http://www.clarkvision.com/articles/image-stacking-methods/

By coincidence (as discussed on that thread) the NIkon D5300 camera at ISO 100 has a gain of very close to 2e/DN and a read noise very close to 3e. So I designed a real test very similar to Roger's simulation. I took 100 real exposures of a test chart displayed on a computer monitor.

Here is the test data in linear form:

Linear values used to generate test target

But this can't be used as a target on the screen because it is not displaying properly. To display properly, I assigned an Adobe RGB profile and applied a gamma of 2.2 to the data:

Actual test target in AdobeRGB colour space

This now displays properly i.e. it gives the correct relative photon rates to a camera.

The brightest numerals (i.e. "25") correspond to 2.5e per exposure. I placed solar film over the camera lens and played with the exposure until I obtained the correct photon rates - that is the purpose of the large discs in the test target - the data can be blurred to reduce the noise and then measured accurately.

The target was displayed on a monitor in a darkened room and the exposure parameters using solar film were the following: ISO 100, 1 sec, f/11.

I took 100 raw exposures and stacked them in PixInsight without using any calibration frames. Only the green (G1) raw data pixels were used. Three quarters of those green (G1) pixels were then thrown away to reduce the image size down to something approximating the original image.

Here is the result, as linear data scaled up by a very large amount:

Stack of 100 images at ISO 100 with very low photon rates.

Certainly there are lots of artifacts in the data which mean I wouldn't naturally choose ISO 100 if I was interested in low signal levels. However, it doesn't look as if we are seeing the quantization issues which Roger suggested.

As always, comments, criticisms, disagreements and questions are welcome.

Mark

-- hide signature --

Takahashi Epsilon 180ED
H-alpha modified Sony A7S
http://www.markshelley.co.uk/Astronomy/

 sharkmelley's gear list:sharkmelley's gear list
Sony a7S +1 more
Nikon D5300
If you believe there are incorrect tags, please send us this post using our feedback form.
ForumParentFirstPrevious
Flat view
Post (hide subjects) Posted by
ForumParentFirstPrevious
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow