0

Which of the two embolded phrases below is more correct?

Serindu. 643 points 1 year ago*

Taking a picture is like filling up a bucket of water. The shutter speed is how long you keep the hose on. The aperture is how big the hose is. And the ISO controls how big your bucket is.

You don't want to overflow your bucket (over-exposed image) and you don't want to fill it too little (under-exposes image). You want to fill it about halfway.

So if you have a big hose (wide aperture) you need a bigger bucket (lower ISO) and/or only turn the hose on for a short time (faster shutter speed) in order to get the bucket filled just the right amount.

Similarly, if you have a small hose (narrow aperture) you might have to run the hose for a long time (slow shutter speed), or use a smaller bucket (higher ISO).

Each of these decisions has trade offs in how the picture will turn out, but I don't know how to describe them as ELI5. Also, setting those values depends on available light, which doesn't really fit the analogy. The analogy works best if you assume you have constant lighting conditions, then figuring out what the settings should be can be done with some trial and error, which helps solidify the concepts.

marcan42. 90 points 1 year ago.

With a smaller bucket (high ISO), you can fill it more quickly and easily (shorter time and/or smaller hose), but when you look at it it becomes harder to accurately tell how full it is (the picture is noisier). Bigger buckets take more time or a bigger hose, but you can more accurately look at the water level. This is partially because it's always raining a little bit; the rain will add noise to the amount of water collected (think of the bucket as getting taller but not wider: the taller the bucket, the less effect the rain has on the overall level as a percentage).

Also, if you imagine the image as a lot of buckets instead of just one (like pixels), the bigger your hose (wider the aperture) the harder it is to hit a single bucket only - so the image is less sharp (in particular, there is less depth of field, so out-of-focus parts of the image are blurrier; this analogy isn't perfect).

Available light is like water pressure. The less pressure you have, the bigger a hose you need to get the same amount of water in the same amount of time. So if you have a small hose (small aperture), a big bucket (small ISO), and low pressure (light), it's going to take forever to fill it up properly (very slow shutter speed required).

Also, if the hose is moving, the longer you keep it on, the messier everything gets (motion blur).

OozeNAahz. 15 points 1 year ago

I think of the ISO as putting rocks into the bucket before filling it. You get it filled quicker but at a lower quality.

2 Answers2

1

I think all of the analogies are poor because ISO has nothing to do with digital exposure; it is not sensor sensitivity. ISO is simply amplification; it is just a brightness control/setting, much like the brightness controls for your monitor.

With an **ISO invariant sensor you can use any (lower) ISO you wish and adjust the brightness after the fact with identical results... or better results because the recovery can be done selectively.

With a non-invariant sensor increasing the ISO (analog amplification) improves the SNR of the recorded image relative to downstream noise added by the camera (typically by the ADC). I.e. using a higher ISO reduces image noise, it does not increase it.

(**AFAIK, there are no completely ISO invariant sensors/cameras at the moment. With all cameras you are somewhat better off using the first couple of steps of ISO gain rather than recovering underexposure in post).

Steven Kersting
  • 17,087
  • 1
  • 11
  • 33
0

The rocks are not really any good analogy, but the real analogy depends on the kind of sensor. These days, there are ISO-invariant sensors (and with larger sensor, this is more common) but also, particularly for small sensors, some that aren't.

For an ISO-invariant sensor, the ISO corresponds just to size of bucket fill you are aiming for, but the filling and measuring itself is identical. You just scale the results afterwards according to the level you had been aiming for.

For an ISO-variant sensor, you have different analog amplification. Consider filling your bucket into beakers of different diameter and checking the results with the same ruler. Modeling noise is not really easy here. The bucket may be somewhat leaky, or the hose causes a certain droplet size. As the beakers become smaller, that has more of an effect, but the ruler resolution also comes into play dependent on the ISO.