Do bigger pixels mean less noise?

Started Jan 6, 2012 | Discussions thread
sacundim Senior Member • Posts: 1,111
Ok, let's think about this.

bobbarber wrote:

...the quality of noise changes at some point, around ISO 3200 on the Panasonic G2 I currently use, but at some point with every camera. The qualitative change is from what I call "speckly" noise to "blotchy" noise.

I'm going to take your statement (here and later in the subthread) that this is the G2's raw performance at face value for now. (Maybe I'll check later with my G1 raws, but I'm afraid I gave up on ISO 3200 really quick on that camera.)

I can think of two hypotheses that could account for something like what you describe:

  1. The camera "cooks" the raw sensor data to perform some form of noise reduction. Some cameras have been documented to do this. I however suspect that the G2 doesn't.

  2. The "blotchiness" is non-randomness in the noise.

Noise is, ideally, completely random, so that each pixel is affected completely independently of any other pixel in the photo. To the extent that this assumption is not met, the noise will look like a pattern instead of "speckles." So it could well be the case that at ISO 3200, the noise level of a pixel in the G2 sensor is correlated to that of its neighbors.

Just a thought. Comments welcome.

One of the more curious facts about m4/3 DXOMark scores is that the GH2 doesn't score better than the GH1, despite general agreement that the GH2's low light images are better than the GH1's. Some people have attributed this to the GH1 suffering from pattern noise at high ISOs that the GH2 eliminated; if this is the case, the magnitude of the noise is more or less the same but its distribution is more random for the GH2.

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