Rubber Soul wrote:
I've downloaded the RAW file you provided elsewhere in this thread. And I got the impression that the scene was originally shot under a strong red light, which was later fixed with click white balance.

Wouldn't that explain why the image held up so well after being pushed up 3 stops? The Achilles Heel of these sensors is typically the red channel. At high ISO, the red channel is always the first one to fall apart. It has always been the most problematic of the 3 color channels. Therefore, shooting a scene under red light will effectively compensate for the sensor's weak spot, allowing it to produce RAW files that can be pushed up to higher ISOs than it normally can.

It's just a theory, of course. Whether or not it's a valid theory, I don't know...

Great point. I definitely have noticed that the colour of stage lights makes a big difference to the result when pushing. I have always assumed that the blue channel is the most troublesome, but maybe that's because when I shoot in very low light there is often tungsten in the balance, often predominantly, so the red and green channels aren't nearly as sparse as the blue.

Here's another file which had mostly green and blue lights. I pushed it to ISO 12800 out of interest, but it's probably too much. In this shot the red channel was the worst. I think it just comes down to shadow, i.e. very low light levels in any of the channels, will cause a problem there. I will try to shoot some ISO 12800 in daylight to see which channels are the worst. In this image horizontal streaks are worse than in the first image.

> You and I both know there have been many references to this supposed "additional detail" and other such claims of easily recognizable distinguishing features. I ain't sayin' the camera isn't better. I am just sayin' that it ain't as much "better" as some owners try to make it out to be. If it was really that much better, you, as a photographer who makes his living selling photos, would NOT sell it. You would want another one.

First, the improvement in resolution between 5D and 1Ds3 is what it is, 8 megapixels of extra data. A doubling of apparent resolution from the 5D's 13 megapixels would be 52 megapixels. Set your expectations accordingly. Here's a shot where I was happy to have the modest advantage in detail:

ok - we got a bit closer to answering the question on this thread: http://www.fredmiranda.com/forum/topic/682831

ISO 12800, NX (with NR): DPP/Neatimage