From: ejmartin on
On Jul 19, 8:15 pm, John Sheehy <J...(a)no.komm> wrote:
> "Roger N. Clark (change username to rnclark)" <usern...(a)qwest.net> wrote innews:48815D3C.2030803(a)qwest.net:
>
> > John Sheehy wrote:
> >>  Then again, I'm
> >> the guy who got 100% on all his math and most of his science tests,
> >> without studying, so maybe I'm expecting too much.
> > So what happened?  I showed you some of your math and conceptual
> > errors but you failed to recognize them.
>
> I haven't read your posts yet, except for a couple.  It takes a long time
> to reply to one of your longer posts, with all the tangents you go on
> acting as if I was implying things that I wasn't.
>
> The things I've seen of yours in other people's quotes have been off the
> mark, and your 4 electron read noise does *NOT* exist in DSLR pixels at
> base ISO.  It's a fantasy.  6 or 8 FZ50-like chips fabbed into a larger
> wafer with parallel readout is more imminently possible and realistic.
>
> --

John, if you look at any circuit diagram for 4T CMOS active pixel
sensors, and compare it to any circuit diagram for a low-noise
amplifier, you will see that there is no amplification going on in the
pixel. Eric Fossum pretty much dismissed your notion that there is
special circuitry at the photosite that is ISO dependent, in the
thread he started at DPR. There is no ISO dependent structure at the
photosite. Therefore, in its contribution to read noise, an *upper*
bound on the photosite read noise is the lowest read noise (in
electron equivalents) as a function of ISO. This occurs at high ISO.
The advantage of FZ50-like pixels at base ISO has little to do with
pixel properties. Since there is nothing going on at the pixel that
is ISO dependent, one can infer that the pixel read noise is *at most*
the read noise at high ISO (in electron equivalents). That is 3-4
electrons for the best pixels, across the spectrum from small to large
pixels and relatively insensitive to pixel pitch.
From: dj_nme on
John Sheehy wrote:
> rfischer(a)sonic.net (Ray Fischer) wrote in news:48827f34$0$17178
> $742ec2ed(a)news.sonic.net:
>
>> Who cares? People don't care about "pixel densities". People care
>> abut noise per pixel, not noise per sensor area.
>
> Then they are stupid. It's the total image that counts, not the individual
> pixel quality, and the total image is the product of per-area quality and
> total area.

What you have just stated should be comparing whole sensors (ie: total
image) to whole sensors (ie: total image): ie the whole image on one
sensor compared to the whole image on another sensor.
So, please explain again why you are taking a crop out a larger sensor
and comparing it to the whole smaller sensor and not just comparing the
performance of the whole sensors against each other?
You don't seem to have actually done this explanation.
From: Ray Fischer on
John Sheehy <JPS(a)no.komm> wrote:
>rfischer(a)sonic.net (Ray Fischer) wrote in news:48827f34$0$17178
>
>> Who cares? People don't care about "pixel densities". People care
>> abut noise per pixel, not noise per sensor area.
>
>Then they are stupid.

Not as stupid as are you. Noise per sesor area has no effect on one's
photographs. Noise per pixel does.

> It's the total image that counts,

Which is affected by noise per pixel.

> not the individual
>pixel quality, and the total image is the product of per-area quality and
>total area.

Bullshit. "Per-area quality" is wholly irrelevant. Want proof? Find
a photograph in the web and, without knowing what camera produced it,
tell us how big the sensor was and what the noise per area was.

You can't. Why? Because noise-per-area doesn't affect the final
result.

--
Ray Fischer
rfischer(a)sonic.net

From: ejmartin on
On Jul 19, 8:59 pm, dj_nme <dj_...(a)iinet.net.au> wrote:
> John Sheehy wrote:
> > rfisc...(a)sonic.net (Ray Fischer) wrote in news:48827f34$0$17178
> > $742ec...(a)news.sonic.net:
>
> >> Who cares?  People don't care about "pixel densities".  People care
> >> abut noise per pixel, not noise per sensor area.
>
> > Then they are stupid.  It's the total image that counts, not the individual
> > pixel quality, and the total image is the product of per-area quality and
> > total area.
>
> What you have just stated should be comparing whole sensors (ie: total
> image) to whole sensors (ie: total image): ie the whole image on one
> sensor compared to the whole image on another sensor.
> So, please explain again why you are taking a crop out a larger sensor
> and comparing it to the whole smaller sensor and not just comparing the
> performance of the whole sensors against each other?
> You don't seem to have actually done this explanation.

The point is that John wants to compare the qualities of a large pixel
sensor such as are currently found in DSLR's, to a same size sensor
tiled with small pixels. The problem is that no such sensor of the
latter type is currently being produced. So the only option is to
compare a sensor with small pixels, such as the FZ50, to a crop of a
DSLR sensor comprising the same area at the focal plane (ie on the
sensor), both shot at the same exposure and at the same focal length
(real focal length, not the artificial "effective" focal length
obtained by multiplying true focal length by crop factor). So while
one cannot compare whole sensor to whole sensors, one can at least
compare crops of sensors of the same physical size, which is
equivalent to comparing whole sensors to whole sensors of DSLR size by
a common rescaling of the results for each sensor. The important
point is to compare like to like, and not apples to oranges.
From: David J Taylor on
Ray Fischer wrote:
[]
> Who cares? People don't care about "pixel densities". People care
> abut noise per pixel, not noise per sensor area.

So why buy a 12MP APS-C DSLR over a 6MP APS-C DSLR if people don't care
about pixel densities?

David