R6 MKI high ISO performance...

John Sheehy wrote:

Ferenc MOGOR wrote:

Dunno why this guy is going against the laws of physics. On the same size real estate sensor (aka full frame in this case) a lower MPxl count will always collect more light than a higher Mpxl count coz the lower Mpxl count has larger photosites.

That is not true. The only thing that is true is that in FF sensors of the same technology level, and FF sensors only, high-MP FF sensors seem to have a little bit more visible electronic READ noise than low-MP FF sensors. It might seem that the noise difference is larger than it is because of the way images are magnified for inspection (100%), and the fact that smaller pixels get sharpened more and have lower pixel-level contrast with the same lens. Noise has nothing to do with pixel size, per se, because 1" sensors (2.7x crop) of recent manufacture have slightly less noise at high ISOs than a 2.7x crop from the R6. There is no "Law of Physics" that says that you're stuck with more noise when you use smaller pixels.

What it seems to be with FF sensors is some kind of small "electric storm" that comes with both a lot of pixels, and a large FF sensor. This is a read noise quantity/character difference, not a shot noise difference. As far as shot noise is concerned, the whole "small pixels make for more noise" idea has no validity, whatsoever. Take the often-used analogy of rain falling into buckets. Yes, you get more vertical variation from bucket to neighbor bucket with smaller buckets, but you have the ability to recombine the buckets into larger ones if you want. No information was lost by initially capturing the rain in smaller buckets, but if you start out with larger buckets, you can never split each bucket into more, smaller buckets, and get any extra resolution.

Fortunately it's not rain or buckets and I don't think it's the best of analogies.

We do see as we tend to a smaller structure size that it becomes more difficult to keep those structures isolated and to keep an active detention area constant. (If we have one pixel Vs an infinite number of pixels that ratio problem becomes more evident).

This is going to happen especially in low light conditions, thus allowing higher ISOs with same noise factor in the end. Please tell me where I've gone wrong with this approach.

I think I personally am unsure of the exact questions wishing to be answered but if it's at what point do we see a reduction in the detection area ratio then that sounds more like a question relating to those structures not the amount / density.

This doesn't include what can be usefully done with whatever area we area talking about.

As an end user I'm interested in system performance, as a designer it may be other things.

I would be cautious of claims of specific designs from those without evidence or a clear path to it. That doesn't of course stop us enjoying debating / enjoying what may be rather than what is.

No exceptions for "low light regarding shot noise, or read noise with smaller sensors. Only with read noise with FF sensors, and you really don't see that difference until you get pretty high up on the ISO scale. You're not going to see it much at ISO 6400 on a FF sensor. 51K, 102K, especially in deep shade or with incandescent light, and it becomes more significant.

-- hide signature --

Beware of correct answers to wrong questions.
John
http://www.pbase.com/image/55384958.jpg