theSuede
Offline
Upload & Sell: Off
|
Andre: Wrong on most accounts except for the direct quotes, actually.
Base ISO is chosen to be located just above where the sensor fills the well with charges, without charge bleeding or other sensor faults. This may be at ISO 16, ISO 72 or ISO 160, that doesn't matter. It has NOTHING to do with amplification "1". Please define amplification "1"! What do you mean? 1e>1ADU? That's called "unity gain" by some people, and is usually higher, around ISO500 on FX cameras... It seems as if you have absolutely no idea what you're talking about. ISO100 as defined by the (one of three) definitions of ISO in digital cameras that you just reiterated will result in VERY different amplifications depending on variations in sensor size. "Amplification" in these circumstances is usually defined as the ratio of how many electron charges it takes to fill the raw to the amount of levels available (IE how many collected/converted charges/electrons it takes to max out the AD-conversion stage).
Small sensor needs lots of amplification to make "ISO100" setting correct (few electrons > full raw-value).
Large sensor collect more "photons" and need a comparably small amount of amplification to make "ISO100" setting correct (many electrons > full raw-value).
You MIGHT universally call it amplification "1" if you set the maximum linear storage capacity of the C in the sensor well to "1" and say that the maximum raw-value is "1" - then the amplification is "1". But this is very unlikely, and to tell the truth very impractical for almost all purposes. Most people that have a clue about what they're talking about would say that amplification is usually around 4e/ADU at base ISO for an FX-camera. Bigger sensors almost always have a higher figure of merit here, and that means less shot noise per pixel per ISO.
I do know the ISO-spec, I've read it, and know it well. The OTHER choices in how to define ISO is also a bit off, choosing to do it by S/N ratios or purely by FWC and quantum efficiency merit is also a bit misguiding, so I guess we're stuck in the middle of something we don't really know how to define in a way that is "right" for all scenarios.
Small sensors are to affected by shot noise to make the S/N ISO definition really valid for them (they wouldn't get over ISO400!), they are best characterized by the FWC/QE definition - they're almost always tuned to expose as high as they can to get away from their inherent noise.
Large sensors are to limited by the FWC/QE definition, as their maximum charge per pixel is usually a lot higher (which also by the same reasons excludes the "ISO by S/N" definition - it would constantly underexpose!), so the REI definition seems to be the best way we know of so far. Most companies use the DC-004 definition, and not the ISO, by the way.
And if you cant see the reasons why the same ISO gives different amounts of noise (as integrated over a complete picture frame) when you vary the sensor area, I do in all honesty think you should be very careful with throwing that PhD around. Me, with a meager double masters did that connection a long time ago.
The D300 has a base amplification of ~1 (0.8e per bitlevel, 14-bit res), the D700 has an amplification of ~0.25 (4e per bitlevel) - and they're both pretty correctly stated as "ISO200", base ISO. Different amplifications - same ISO! A compact takes even more amplification to get to ISO200, ~10 (1e gives 10 bitlevels, 14-bit res) for a correct ISO200 rating.
Shot noise, which is a variety of the uncertainty principle built into the wave-theory that governs measurement of light-energy, is always equal to the square of e, which gives that noise on a certain ISO varies with the amplification needed to get that ISO, and that varies with sensor (and pixel) size. Then the electronics add a very small amount of noise (small, on low ISOs at least) but that another part of the story.
Sorry to rain on this parade, but this is the reality.
|
ISO, shutter speed and f-stop are all transformable to linear scales. Since digital is less lossy as you say, it is more linear and true to the ISO scale. In film there would be adjustments due to non-linearity. This was especially true with long exposures. Dude, go back to your padded cell. 
). But I seriously think you need to read a decent book on how to set exposure manually and learn to understand it. "Understanding Exposure" by Bryan Peterson is an excellent starting point.
