New Canon EOS 5D

In article , says...
Nono... it wasn't pushed a stop afterwards. I was referring only to
the fact that ISO3200 itself is ISO1600 pushed digitally in-camera.

The raw is at:

http://www.resonant.org/images/episo...ning_skate.cr2

Dammit, I forgot Adobe is forcing me to upgrade to CS2 if I want to use
ACR 3.1!

Could you do a JPG on default settings?

Of course it looks much worse on the scaled down version. On the full
size picture it's much harder to discern.

Well, a little, maybe. I still see it pretty clearly, but then, I
spent a long time trying to fix it, so I know what I'm looking for.

Well, it's really obvious on the resampled images due to the spacing of
the noise.
--
http://www.pbase.com/bcbaird

Brian Baird wrote:
Dammit, I forgot Adobe is forcing me to upgrade to CS2 if I want to
use ACR 3.1!

You can use the DNG 3.1 converter to make CS load it using ACR 2.4

-Mike

In article ws.net,
says...
Dammit, I forgot Adobe is forcing me to upgrade to CS2 if I want to
use ACR 3.1!

You can use the DNG 3.1 converter to make CS load it using ACR 2.4

-Mike

But I'm lazy!
--
http://www.pbase.com/bcbaird

Brian Baird wrote:
In article ws.net,
says...
Dammit, I forgot Adobe is forcing me to upgrade to CS2 if I want to
use ACR 3.1!

You can use the DNG 3.1 converter to make CS load it using ACR 2.4

But I'm lazy!

You're just looking for an excuse to spend more money. :-)

-Mike

In article ws.net,
says...
You can use the DNG 3.1 converter to make CS load it using ACR 2.4

But I'm lazy!

You're just looking for an excuse to spend more money. :-)

That was called the BG-ED3!

--
http://www.pbase.com/bcbaird

In article , says...
Could you do a JPG on default settings?

Heh, oddly, I can't. I've gotten rid of CS1, and the ACR in CS2 is
doing a much better job on conversion. So much so, that I think I'm
going to go back and reprocess that shot properly sometime soon. It's
not completely perfect, but drastically improved. I also no longer
have the out-of-camera jpg.

Does CS2 seem to solve the banding problem? If so, it would lead me to
believe the conversion has a lot to do with this effect.
--
http://www.pbase.com/bcbaird

In message ,
I, wrote:

In message 5DcMe.938$sw6.378@fed1read05,
"Skip M" wrote:

Tough question, the 5fps of the 20D is an attraction, as is the $1400 price
vs $3500 (est).

The speed of the 20D might be the reason why the banding occurs. Maybe
they sacrificed readout quality for speed.

A little more experimentation has shown me that the 20D's readout isn't
any worse than the 10D's; it is simply more visible because it is the
main shadow-quality limitation with the 20D's much lower noise. Taking
10D and 20D blackframes at various ISOs, and replacing each horizontal
line with the average value of its pixels, and offsetting the entire
image to have an average of 128, the two cameras created very similar
line patterns at ISO 100, and then diverge up to ISO 800, where the 10D
has about 33% more banding (measured by histogram width). Without
averaging the lines' pixels, the 10D has about 3x as wide a noise bell
curve as the 20D. At ISO 1600 and 3200, the 10D's banding is less than
at ISO 800, and those strange patterns I've seen in the RAW data (lines
of all odd values, all even values, and alternating odd and even values)
might be put there intentionally to offset banding in the 11-bit
digitization. If this is the case, maybe they should have done the same
with the 20D at ISO 3200, which is all even numbers in the RAW data
(except for interpolated bad pixels).

In any even, I am optimistic that future software can get rid of banding
(I can get rid of it myself in RAW data in many cases, but don't have
the facility yet for a quality converter to utilize the fixed data); if
the 20D were to lose its banding, it would probably get a stop or more
of usable shadows, and would be pushable at least one more stop.

I firmly believe that the 20D has a much better sensor than it seems to
have; the quality bottlenecks are inconsistent readout and the generic
12-bit digitization limit.

--


John P Sheehy