In message ,
Alan Browne wrote:

wrote:

In message ,
Alan Browne wrote:

Can you find the "natural" (neutral gain) ISO of the sensor?

Define it, and I will find it.

Simply put, that region where the sensor needs only unity gain (analog
and digital).

I don't think any ISO uses unity gain. There's nothing magic about
unity, anyway. Still has to pass through silicone junctions.

For example:
http://www.dpreview.com/reviews/koni...lumi-graph.gif

...has the 7D noise lowest at ISO 200. This suggests that the unity
gain ISO is somewhere in the region of ISO 200 (though not exactly there).

My guess is that it was a mistake, or an anomaly of the in-camera
RAW-JPEG conversion. I wouldn't read too much into DPReview charts.

Take a blackframe at each ISO, and look at the histogram of the RAW
data. How wide are the bell curves? How does it look when converted to
8-bit space, with the mean offset to 128? How does a grey card,
externally metered, register level-wise above the middle of the noise
curve for that ISO? These are the things that are relevant, IMO;
comparing JPEGsfrom cameras that shoot RAW, based on the camera's
metering quirks, is useless and misleading.

--


John P Sheehy