Comment by wisty | Hacker News Reader

Comment by wisty | Hacker News Reader

Some random camera stats:

The iPhone has a 1/3.2" sensor (9.7x crop factor), a 4.28mm length f/2.4 max aperture.

That's like a 43mm f/23 full frame (basically, apature scales with crop factor just like length); or a 26mm f/14.5 APC-C.

Compared to a 1/2.3" sensor (which the S800 is likely to pack) it's like a 6mm f/3.3 lens (43mm full frame, as they don't like to mention how apature scales). The advantage of a S800 will be the range (25mm -> uselessly long without a tripod and good light), compared to the 43mm prime iPhone; apature and image quality will be similar. Also, the Nikon will have niceties like vibration reduction, a flash (maybe), smarter focusing / metering, better processing (maybe).

The real question is, who would want one? I'd say - people with dumb phones who carry cameras; and hipsters who want to be seen with a camera while they play Angry Birds on the bus. It might be bigger in non-US markets, where phones don't appear so cheap (due to hidden carrier costs).

Does aperture scale with the crop factor? I don’t think that’s the case. f/2.3 will always let the same light in. As you (hypothetically) scale up the focal length (the numerator), you also scale up the pupil size (the denominator), so the aperture remains constant.

The effect on DoF is different, sure, but not on light – and I think with those kinds of cameras you are more interested in their ability to let in light than their ability to produce shallow DoF (that’s pretty much a given for any small sensor camera).

If you scale up everything, the aperture stays constant, correct.

But since the larger sensor can work with much more light, it therefore has a much higher S/N-ratio. If you switch that camera to a higher aperture, you get less light, leading to a lower ratio again.

So the lower S/N-ratio of the smaller sensor could be estimated as a higher aperture. Whether the numbers calculated by wisty are correct, I don't know, though.

Please don’t do that. That’s just confusing the issue and highly misleading.

The f-number is a property of the lens and nothing else. The S/N ratio of the sensor is independent of that.

It’s correct: smaller pixels tend to have a lower S/N-ratio, which mostly means that you can’t turn up the gain as much as with sensors that have larger pixels (you can’t increase the ISO by as much) and you will see more noise at the same gain.

But that’s no law of nature. It’s possible to have a crappy large sensor and an excellent small sensor. It’s the sensor that matters here – and there are many more variables involved there. Sensors with pixels a certain size don’t have a certain S/N-ratio only depending on pixel size. And any exact mathematical formula (like f-number scaling with crop factor) doesn’t make sense there, not at all.

Depth of field does scale perfectly with crop factor, and that's a big thing.

f-number is a ratio of apature to length. A 400mm f/5.6 will have a 71mm apature. An iPhone lens does not (despite having a better f-number).

The light is spread out more, so you'll a higher ISO to take fast photos, but let's say ISO is proportional to sensor size. A larger sensor captures more photons, so it can have a higher ISO. That is a law of nature.

As you say, ISO is neither deterministic nor linear, but I'm more confortable considering ISO to scale with crop size than pretending it's constant.

If we are ignoring ISO altogether that's fine, but then there's no point talking about the amount of light hitting the sensor.

That leaves diffraction, but I don't really understand diffraction.

Actually the iPhone 4S has a 35mm equivalent lens, while the 4 had a 29mm equivalent: http://www.imaging-resource.com/PRODS/IPHONE4S/IPHONE4SA.HTM

I'm not sure what you mean by aperture scaling with crop factor either?

> I'm not sure what you mean by aperture scaling with crop factor either?

DoF, the ability to take fast photos (due to better ISO) and diffraction can scale with crop factor. Maybe it's not a perfect equivalence, but neither is focal length.

Reply on news.ycombinator.com