dpreview pixel density metric

Paul Furman wrote:
> timeOday wrote:

>> I don't see the point of the pixel density metric since dpreview already
>> measures noise and resolution at different ISO settings. That's what
>> matters. Bigger pixels tend to perform better on those respects, but
>> it's not a direct measure so why bother when you already have the real
>> thing?
>
> Dpreview only lists ISO and provides jpegs to assess, there's no number
> other than highest offered ISO, which is an OK general indication but...
> If they calculated SNR, full well saturation or some such from raw
> files, that'd be useful but they don't.

No, check out the noise graphs on, e.g., this review:

<http://www.dpreview.com/reviews/canoneos450d/page19.asp>

"timeOday" <timeOday-UNSPAM@theknack.net> wrote:
>>
>> Dpreview only lists ISO and provides jpegs to assess, there's no number
>> other than highest offered ISO, which is an OK general indication but...
>> If they calculated SNR, full well saturation or some such from raw
>> files, that'd be useful but they don't.
>
> No, check out the noise graphs on, e.g., this review:
>
> <http://www.dpreview.com/reviews/canoneos450d/page19.asp>

Those graphs are for jpegs, and Nikon and Olympus are applying much stronger
noise reduction at higher ISOs, and that's reducing detail. So the Nikon and
Oly noise graphs are not representative of the actual noise from the sensor
(look at the ISO 1600 images: the Canon cameras are capturing a lot more
detail).

The page with the -3.0 EV compensation in ACR is interesting, though.

*: http://www.dpreview.com/reviews/canoneos450d/page21.asp

--
David J. Littleboy
Tokyo, Japan

David J. Littleboy wrote:
> "timeOday" <timeOday-UNSPAM@theknack.net> wrote:
>>> Dpreview only lists ISO and provides jpegs to assess, there's no number
>>> other than highest offered ISO, which is an OK general indication but...
>>> If they calculated SNR, full well saturation or some such from raw
>>> files, that'd be useful but they don't.
>> No, check out the noise graphs on, e.g., this review:
>>
>> <http://www.dpreview.com/reviews/canoneos450d/page19.asp>
>
> Those graphs are for jpegs, and Nikon and Olympus are applying much stronger
> noise reduction at higher ISOs, and that's reducing detail. So the Nikon and
> Oly noise graphs are not representative of the actual noise from the sensor
> (look at the ISO 1600 images: the Canon cameras are capturing a lot more
> detail).

It is a bit disappointing that they test noise with jpegs.

On the other hand, they used the same conditions for the resolution test:

<http://www.dpreview.com/reviews/canoneos450d/page33.asp>

So a camera that cheats on the noise test should suffer on the
resolution test, and vice versa.