D300 sample at ISO 6400

Not particularly useful, but here it is - posted on Japanese
"slr club" forum, presumably from a pre-production model at
the launch:
http://www.slrclub.com/bbs/vx2.php?i...c=asc&no=20929

frederick wrote:
Not particularly useful, but here it is - posted on Japanese "slr club"
forum, presumably from a pre-production model at the launch:
http://www.slrclub.com/bbs/vx2.php?i...c=asc&no=20929


I'd say the noise looks similar to a Sony A100 at 800 to 1600 (raw, that
is - the JPEGs have much less noise) and the sharpness on the Nikon lens
badging is very similar to Sony's 'don't blur it all by noise reduction'
approach - which the D80 does not have. The D80 is the lowest noise
application of the Sony 10 megapixel CCD sensor (Pentax and Samsung
next, and Sony - last) and fine detail sharpness seems to be reciprocal
- Sony sharpest, Pentax/Samsung middle, Nikon least so but aided by
kit/low cost lenses which are generally much better. I found the Nikon
D200 to be much the same as the Sony A100 though fewer comments are made
about noise - nothing like as noise-free as the D80.

This is probably a JPEG shot and does not look as if fierce NR has been
implemented, which bodes well for exceptional detail sharpness - which
is exactly what Nikon went for on the D2X, and got slated for having too
much noise. Personally I prefer the D2X approach to the D80 approach. I
don't mind film-grain like noise, and it's quite useful to be able to
get it on demand for certain types of shot. It can always be post
processed. Fine detail is very important. This shot gives a clue that
the new Nikons may be emphasising fine detail and not using excessive NR.

David

--
Icon Publications Ltd, Maxwell Place, Maxwell Lane, Kelso TD5 7BB
Company Registered in England No 2122711. Registered Office 12 Exchange
St, Retford, Notts DN22 6BL
VAT Reg No GB458101463
Trading as Icon Publications Ltd, Photoworld Club and Troubadour.uk.com
www.iconpublications.com - www.troubadour.uk.com - www.f2photo.co.uk -
www.photoclubalpha.com - www.minoltaclub.co.uk
Tel +44 1573 226032

David Kilpatrick wrote in
:

This is probably a JPEG shot and does not look as if fierce NR has been
implemented,

Not luminance, but the chroma noise reduction is pretty strong. Not that I
would complain; chroma noise is the only type I really want an in-camera
JPEG to remove (other than things that are obviously not signal, like line
noises and hot pixels).

--


John P Sheehy

On Aug 27, 4:35 am, John Sheehy wrote:
David Kilpatrick wrote :

This is probably a JPEG shot and does not look as if fierce NR has been
implemented,

Not luminance, but the chroma noise reduction is pretty strong. Not that I
would complain; chroma noise is the only type I really want an in-camera
JPEG to remove (other than things that are obviously not signal, like line
noises and hot pixels).

--


John P Sheehy


There's aggressive chroma NR. Nikon's is quite intelligent, it uses
some kind of masking technique (or that's what it looks like). The
downside is that after chroma noise has been "removed" like this,
either in jpegs or nikon capture, you can't use neat image or
something like that effectively because the noise distribution is
nonuniform. It seems to try to reduce colour bleeding, and it mostly
works, but complicates further NR; eg take a look at the gold
lettering and the yellow bleeding out of it. There's also luminance
smearing; it may not be obvious if you haven't spent as long as I have
playing with NR in nikon capture (I use a D200 and mostly shoot at
night, so I'm an expert at this ), but it would have been easier to
see if there was low-contrast texture (or maybe there was on the
camera body's metal; Nikons aren't smooth).

It's quite impressive if it's 6400, though. It seems a stop or so
better than the D200 (although I'm not sure how its jpegs look at
3200, to be honest).

John Sheehy wrote:
David Kilpatrick wrote in
:

This is probably a JPEG shot and does not look as if fierce NR has been
implemented,

Not luminance, but the chroma noise reduction is pretty strong. Not that I
would complain; chroma noise is the only type I really want an in-camera
JPEG to remove (other than things that are obviously not signal, like line
noises and hot pixels).


There has to be a catch somewhere...
For D2x with same pixel density, base iso stops at iso800,
D300 at iso 3200.
I doubt Nikon would be dumb enough to put iso 3200 in as a
base iso if it was going to be completely unusable.
Even that "Hi 1" shot looks quite usable - not exactly great
exposure as it looks exposed for the shadows.
So, a new trick pulled out of the hat? I dunno - I thought
that there weren't any tricks left in the hat.

frederick wrote in news:1188179600.655799@ftpsrv1:

There has to be a catch somewhere...
For D2x with same pixel density, base iso stops at iso800,
D300 at iso 3200.
I doubt Nikon would be dumb enough to put iso 3200 in as a
base iso if it was going to be completely unusable.
Even that "Hi 1" shot looks quite usable - not exactly great
exposure as it looks exposed for the shadows.
So, a new trick pulled out of the hat? I dunno - I thought
that there weren't any tricks left in the hat.

No doubt, the image is very well exposed; the young woman's face looks
almost clipped, and the black camera is pretty grey. Well-exposed images
don't vary as much from camera-to-camera as poorly exposed ones do. For
sensors of equal size, shot noise varies little from camera to camera;
maybe 1/2 stop difference over the range. For under-exposed images, or
deep shadow areas, read noise is dominant, and can vary by up to 4.5
stops at ISO 1600 from camera to camera!

I want to see RAW files; RAW files with pure black areas, RAW files with
out-of-focus flat areas of various intensities, etc. RAW noise is much
simpler than these JPEG guessing games including mystery conversions and
post-processing, NR, etc.

IMO, if a camera has RAW output, then the IQ of "the camera" is "the
RAW".

We're about 10 years into the digital photography craze, and there isn't
a single web-repository of RAW test files from various cameras, and/or
measurements from RAW files, even though such would hold the answers to
all the questions about IQ, but instead, throngs look at JPEGs and
speculate in the dark.

--


John P Sheehy

On Aug 26, 5:55 pm, frederick wrote:
Not particularly useful, but here it is - posted on Japanese
"slr club" forum, presumably from a pre-production model at
the launch:http://www.slrclub.com/bbs/vx2.php?i...&sn1=&sid1=&di...

They've pretty much conquered chroma noise I see.

On Aug 26, 5:55 pm, frederick wrote:
Not particularly useful, but here it is - posted on Japanese
"slr club" forum, presumably from a pre-production model at
the launch:http://www.slrclub.com/bbs/vx2.php?i...&sn1=&sid1=&di...

Additionally, not much detail lost when it is "NR'd"

http://www.pbase.com/andersonrm/image/84544263

In article , John Sheehy
wrote:

We're about 10 years into the digital photography craze, and there isn't
a single web-repository of RAW test files from various cameras, and/or
measurements from RAW files, even though such would hold the answers to
all the questions about IQ, but instead, throngs look at JPEGs and
speculate in the dark.

how about these?

http://www.jirvana.com/raw_large/
http://www.glasslantern.com/RAWpository/
http://www.rawsamples.ch/index_en.php
http://raw.fotosite.pl/

On Aug 27, 3:59 am, frederick wrote:
John Sheehy wrote:
David Kilpatrick wrote in
:

This is probably a JPEG shot and does not look as if fierce NR has been
implemented,

Not luminance, but the chroma noise reduction is pretty strong. Not that I
would complain; chroma noise is the only type I really want an in-camera
JPEG to remove (other than things that are obviously not signal, like line
noises and hot pixels).

There has to be a catch somewhere...
For D2x with same pixel density, base iso stops at iso800,
D300 at iso 3200.
I doubt Nikon would be dumb enough to put iso 3200 in as a
base iso if it was going to be completely unusable.
Even that "Hi 1" shot looks quite usable - not exactly great
exposure as it looks exposed for the shadows.
So, a new trick pulled out of the hat? I dunno - I thought
that there weren't any tricks left in the hat.

Not at all, if they managed to reduce read noise to Canon's levels,
then these results shouldn't be a problem (note that there is heavy NR
on this image, and the dark camera is exposed almost as a midtone, as
you said).