f-number equivalent between m43 and FF

Started Mar 25, 2014 | Discussions thread
dwalby Veteran Member • Posts: 5,257
Re: SNAP! Where it is?
1

crames wrote:

dwalby wrote:

crames wrote:

Upsampled noise looks noisier, not due to any change in the standard deviation but due to the way magnification shifts the visible frequencies in the noise to more sensitive regions of the eye's contrast sensitivity function.

By upsampling don't you create a condition where the noise in adjacent pixels is no longer uncorrelated?

Yes, the new pixels created when upsampling will be correlated with the surrounding pixels they were interpolated from.

When you bin a block of pixels you only achieve the sqrt(N) SNR improvement if the noise is uncorrelated over all the samples. Depending on the magnitude of the cross-correlation components of the noise you lose some or all of that processing advantage.

I don't follow you. When you bin you downsample. Are you talking about upsampling then downsampling? Why would you do that?

I'm here trying to learn myself.  I have a signal processing background and most of the concepts being discussed seem to overlap into optical processing as well, so I'm trying to apply what I already know to understand exactly what they're talking about here.

From what I gather so far, the SNR improvement in these two cases seems to be achieved through pixel binning, either directly or by your eye's perception.  I believe the actual SNR of each recorded pixel is the same, as they are the same pixel with the same illuminance in both cases.  So the only way one can become less noisy is if some additional processing is performed, and that would be binning because one image contains 4x (or whatever multiple) pixels compared to the other.  Its analogous to having a pixel that's 4x larger and collects 4x the photons, thus has a better SNR.

When you bin, the processing advantage only applies if the noise samples are uncorrelated, which they are at native resolution.  If you upsample the smaller image to the larger image pixel count, your noise samples after upsampling are no longer uncorrelated.  Therefore if you later bin both images, one will enjoy the sqrt(N) SNR improvement, and the other will not, because the larger one has uncorrelated noise samples and the (upsampled) smaller one does not.

Post (hide subjects) Posted by
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow