The Joy of Pixel Density

Started Jul 13, 2008 | Discussions thread
ejmartin Veteran Member • Posts: 6,274
Re: The Joy of Pixel Density

John Sheehy wrote:

ejmartin wrote:

Are you sure it's the higher pixel density?

If the QE doesn't drop, why not. Resolution is definitely better,
and read noise in real or virtual quadrature drops with current
technology. Even if there were no read noise at all, and we were
just counting photons, the higher density would be beneficial for

I agree that gain/unit area seems remarkably uniform across a wide variety of photosite sizes, from the FZ50 to the D3. I am rather doubtful about the claim regarding read noise. Read on...

Or is it that the
smaller DR of the smaller pixels puts less demands on the DR of the
VGA/ADC, so that the small pixel camera is limited by the pixel DR
while the large pixel DSLR sensor is limited by the VGA/ADC and can't
demonstrate its full potential?

Well, the two seem to be married, until someone invents a photon
counter that only works on huge pixels. Resolution will still suffer.

There are any number of tricks to get around the limited DR of the VGA/ADC combination; for instance the dual amplification scheme I put forth, or alternatively, at a cost in frame rate, since sensor reads are nondestructive on CMOS, one could read the sensor twice using the same VGA at different amplifications. Either way, you achieve the lower read noise for photosites with low photoelectron count.

You are making a claim about what is achievable with the photosite technology; if the photosites themselves have higher DR than the camera is delivering, I don't see why one should use the VGA/ADC limited figures in determining what could be achieved. Rather, one should use the high ISO read noise which is not limited by these components.

It seems inconsistent to argue to Barry Fitzgerald about what could be done but isn't being implemented wrt small photosites on large sensors, but then hide behind what is currently implemented vs what could be done when it comes to extracting all the DR that the photosites are capable of, but not delivering in the data output due to the limitations of other components of current cameras.

BTW, I found a current article by Eric Fossum that suggests that the ISO amplification is done off the photosite in a typical design:

It's not obvious to me that you've
proven that the small pixel sensor is better per unit area; rather
you have shown that the sensor/VGA/ADC combination favors the small
pixel sensor. One still has to disentangle which component of that
combination is responsible.

Well, the low area-based read noise at base ISO isn't being achieved
with low pixel density with current technology. More pixels per unit
of area allows lower read noise.

That's only true if you take the low ISO read noise of current implementations, which I have argued is limited by off-photosite electronics. I've outlined an alternative to achieving 14 stop DR at base ISO with big pixels.

Those 14 stops come from lowering the base ISO read noise to 4-5 electrons, as it is at high ISO, using the dual amplification trick. Let's now run the numbers:

Canon 1D3: 4 electrons read noise, 7.2µ pixels

Panasonic FZ50: 3.5 electrons read noise (your figure, roughly, IIRC), 1.97µ pixels

Taking the ratio of read noise to pixel pitch to make a level comparison, the big pixel DSLR wins by a factor of more than three.

Personally, I'd take the higher resolution in many cases even if the
noise were slightly higher, but it isn't.

I'd like to see choice available for the consumer -- high resolution for people who need it, large photosites (and actual delivery of the full photosite DR) for people who need wide DR and are willing to give up the resolution.

-- hide signature --
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow