Large format digital sensors?

[welly]

Maybe in the not too distant future!

http://www.canon.com/news/2010/aug31e.html

[Joanna Carter]

But they don't actually state the number of pixels, and are only talking about video, which doesn't require the same resolution as still photography.

[Emmanuel BIGLER]

Maybe in the not too distant future!

To me the problem is not time (waiting for it to reach the market) it's ... money ;)

[paulr]

This chart (stumbled onto on the entry for "Moore's law" on wikipedia) charts pixels-per-dollar over time.

It stops in 2005, and is based solely on the price of Kodak cameras in Australia, so take with a grain of salt ... still I think the linearity is interesting. If that continues, dirtbags like me may one day have a nice digital camera.

http://upload.wikimedia.org/wikipedia/commons/9/94/Hendys_Law.jpg

The caveat, explained to me recently in another thread, is that it's much more expensive to scale the size of the sensor than the pixel density. And I suspect we've already started to hit diminishing returns with the helpfulness of pixel density. For one thing, lens technology is not improving with any great speed. And we're be up against diffraction and other limitations imposed by physics.

[SamReeves]

Can you imagine cleaning the dust off those in a view camera? :D

[williamtheis]

another problem is my old Masterview may not have enough precision to do this sensor justice! Using a Better Light scan back and, man, does it have problems with low light... resolution is not an issue. give me fat fast pixels!!!!

[Kirk Gittings]

But they don't actually state the number of pixels, and are only talking about video, which doesn't require the same resolution as still photography.

That is really not true anymore, as FF chips are now doing double duty to capture HD video and also look at the impact of the RED double duty cameras.

[Juergen Sattler]

Canon just announced their latest development - a 202x205mm sensor with incredible sensitivity to light - I am sure this is years off before it becomes commercially available, but it is a step in the right direction:

http://gizmodo.com/5626133/with-expanded-size-comes-greater-light+gathering-capability-in-canons-worlds-largest-cmos-sensor

[paulr]

Can you imagine cleaning the dust off those in a view camera? :D

I've wondered about that. someone would need to develop a simple and effective way to keep the thing clean without risk of trashing it..

[Kirk Gittings]

I've wondered about that. someone would need to develop a simple and effective way to keep the thing clean without risk of trashing it..

I'm not saying it wouldn't be an issue but the larger the sensor the less impact small dust particles would be-like getting dust on your film in a camera-35 vs. 4x5.

[Jay DeFehr]

Any thoughts on very fast lenses that cover that format? A new, ultra fast, Canon LF lens might be interesting, and certainly very expensive. I'm trying to imagine what a complete system built around that sensor might look like. Fascinating!

[ic-racer]

Like most things in the digital marketplace it makes no sense to me. The longer lenses for that sensor will be slower than those for the smaller sensor at the same glass size.

[Mike Anderson]

Any thoughts on very fast lenses that cover that format? A new, ultra fast, Canon LF lens might be interesting, and certainly very expensive. I'm trying to imagine what a complete system built around that sensor might look like. Fascinating!

But if it's really 100 times more sensitive, you don't need a fast lens. The electronics might end up being economical by playing well with cheap (slow) lenses. And you save on lighting equipment too.

...Mike

[Ben Syverson]

Mike is right... With a larger sensor, you can make each photoreceptor site geometrically larger. More area means more light, more light means more sensitivity.

Still, I'll take the same sensitivity and resolution as my DSLR but scaled up to 8x10. That would be a pretty nice 1 Gigapixel sensor.

[Nathan Potter]

Scale the increased 100X sensitivity to the area increase in pixel size compared to the best 35 mm full frame sensor. Since we can assume about the same CMOS technology for both, the pixel size of the 8 X 8 inch mentioned is probably about 50 um compared to about 5 um +/- for the best 35 mm.

Production cost for a 12 inch CMOS wafer in prototype maybe 250 K$. For a mature process maybe as low as 5 K$ depending on number of levels and sophistication of the wiring and especially the wafer yield.

Dust would be managed by a semi hermetic glass window well outside the image plane; probably coupled to a processing algorithm for analyzing and correcting for any intensity variations due to dust on the remote window.

Nate Potter, Austin TX.

[Ben Syverson]

Nathan, interesting... I understand what this sensor is for (highly classified military and astronomy), so 50 micron pixels make sense, but no photographer in their right mind is going to buy an 8x8" sensor with only 16 megapixels. The sensitivity of existing 35mm full frame sensors is fine -- imagine being able to shoot digital LF at 1600 ISO that looks better than 400 ISO film! I say just give me more pixels.

If we really need to increase sensitivity, 100X is too much. How about 10X? That would mean 15.8 micron pixels, for a 165 megapixel 8x8" sensor. Not too shabby. But I'd still rather take a 1.65 Gigapixel sensor with 5 micron pixels!

[Struan Gray]

Big pixels on a large sensor are the only way to get high-resolution infra-red images.

Sensitivity, dynamic range and noise are controllable through materials science and process engineering. Diffraction isn't, and making aberration-free fast lenses is probably the biggest challenge facing high-resolution small formats.

So, I see a use for, even a need for, large area sensors. Whether any of them are engineered to a price point that is relevant to hobby photographers is another matter.