From: John Sheehy on
Rich <none(a)nowhere.com> wrote in
news:6pWdnW5PHdl826jWnZ2dnUVZ_jdi4p2d(a)giganews.com:

> John Sheehy <JPS(a)no.komm> wrote in
> news:Xns9CE8B493E8E58jpsnokomm(a)216.168.3.70:

>> So, I would say there is some value in showing small images - *IF* it
>> actually means something.
>
> The bird shot looks pretty good. There can be some value in showing
> any image, at any size. But, not to illustrate noise control or
> resolution. If I were to simply take that shot at face value, I'd
> say, yes, it looks like 6400 ISO on a D300. But that would be at
> 100%.

Uh, no. This is so far from reality; I don't know what to say. The D300
is very good at base ISO; there it has more DR and less patterned deep
shadows than any Canon at the pixel level, but the read noise is almost
proportional to ISO and grows very quickly as ISO is raised. By 6400 it
is well behind any recent Canon. If you're used to looking at
conversions from Nikons, they have traditionally removed all high-
frequency chroma in the shadows, and virtually removed all color in the
deepest shadows. Nikons have just as much chroma noise relative to
luminance noise as any other camera using a CFA; luminance and chroma
noise are virtually inseparable, as they have the same source. They tend
to differ only when one camera has pale CFAs relative to another, then
conversion boosts chroma noise. The D300 and 7D seem to have nearly the
same RAW color space, though. I don't see any noticeable color response
difference between them. I just did an unbiased "conversion" of the
Imaging-Resource 6400 versions of the table with the crayons and black
mug for both the D300s and the 7D, and while the D300s does quite well,
without any visible pattern noise, its noise is more intense and coarser-
looking at the pixel level. When you scale the images to the same image
size, the difference becomes greater. The D300s is a bit duller-edged,
and clearly noisier, including chroma noise. This image you have of
Nikons having extremely low chroma noise is an illusion created by the
default style of NR for Nikons.

> Knowing that is a reduced-size crop, I have no way of actually
> understanding how noise looks in the camera for comparative purposes.
> I'd literally need another shot, taking at the same exposure levels
> with a camera I did know in order to determine what the Canon's actual
> noise levels were.

That's not possible. The closest you will get is to downsample the IR
samples, in the RAW color space.