Michael's first impressions of the 5D

[DarkPenguin]

Some more pre-prod samples from Japan for those who might not have seen these yet:

http://dc.watch.impress.co.jp/cda/review/2...08/29/2180.html

Regards,
Bernard

Lord have mercy.

Must repeat "Stitching is cheaper than a 5D.  Stitching is cheaper than a 5D...."

[Ray]

I'm concerned that if we push camera manufacturers too hard on noise they will smooth files in-camera in ways we have no control.

I think Canon are smart enough to avoid doing this. There are too many watchful and critical people around who would pick this up immediately. I guess you are referring to the way the D2X handles noise. At high ISOs noise is very low, but only with the extra NR filter switched on, which smooths the image and reduces resolution. However, Nikon have very sensibly provided a choice of using this extra filter or not.

I understand 'on chip' processing refers to processing that takes place before the A/D conversion.

[LeifG]

I have heard some Canon D2s users comment on plasticy tones from the camera and comment that this is due to aggressive noise removal.

Excuse me for asking, but is this a typo, or is there a Canon "D2s"? If so, what kind of camera is that?

Thanks. It's a typo. For D2s read 2Ds. Doh.

Leif

[Ben Rubinstein]

Nope, that one doesn't exist either.

[BJL]

Ray,

   first, thanks for the entertaining fantasy about where you got your faith that 24x36mm sensor format is the future.

But about sensor noise, though this is not directly related to the 5D, I find it interesting that Sony describes the brand new CMOS senosr with 5.49 micon pixel pitch as having only 2.5 times as much dynamic range as the three year old Sony 8MP interline CCD with 2.7 micron pixel pitch.

That is
a) more than four times as much sensor area per photosite
 three years of technological improvement
c) the change to CMOS, which is often claimed to be superior to interline CCD for noise levels
and yet instead of dynamic range growing by a factor of more than four as I half expected, it is only up by a factor of 2.5.

This suggests to me that probably either
a) well size and maximum signal level have indeed been increased by the expected factor of over four, but dark noise levels have almost doubled so as to consume about half of the dynamic range improvement,
or
 the well capacity has not grown nearly as much as the available sensor area per photosite, due to a continuing disadvantage of CMOS sensors than a larger proportion of sensor area is taken up by other circuitry.