New Foveon: Another milestone in image sensor technology?

[semilog]

Yet development has shown us that we are continuously fitting larger sensors into smaller (or same size) bodies.

Same size, maybe. Smaller, not so frequently. At least for FF.

The problem with big sensors is that they are expensive to produce, and they are expensive to produce because bigger chips have high defect rates.

The biggest reason that we've seen exponential price drops with computers is that higher circuit density allows smaller devices: more devices per wafer, and better yields (a smaller fraction of defective parts).

None of those economies is possible with FF or any other large sensors. Yes, prices will drop, but it is unlikely that prices will drop precipitously.

For the vast majority of consumers, and indeed for many working photographers, APS-C sensors are more than adequate already. Very, very few casual photographers actually ever make prints bigger than 14 x 20, or require better than ISO 3200.

Add to that the fact that bigger sensors require more power (bigger heavier batteries) and bigger heavier lenses, and I don't think you'll see APS-C sensors going away any time soon.

Bigger sensors also have to dissipate more heat. If this heat is not dissipated, the noise advantage is nullified. They are also heavier, a consideration for designers who wish to incorporate sensor-based antishake tech.

Note further that mirrorless cameras with APS-C or micro 4/3 sensors already own 40% of the market in Japan, and that fraction is growing.

FF DSLRs are not going away, but they will be high-end niche products for the foreseeable future.

 

[scottgee1]

Why only Sigma?

Why only Sigma?

Am I the only one who would be more interested in Foveon technology if it was employed by someone other than Sigma?

😕

I dunno. Maybe they're trying to win the mucho-megapixels war?

 

[sevo]

It is fairly safe to predict that it will revolutionize photography as much as the previous Foveon generations...

 

[scottgee1]

It is fairly safe to predict that it will revolutionize photography as much as the previous Foveon generations...

😀

"Legends in their own minds . . ."

 

[Paddy C]

Am I the only one who would be more interested in Foveon technology if it was employed by someone other than Sigma?

😕

I dunno. Maybe they're trying to win the mucho-megapixels war?

Well they own the patent. So no one else, to the best of my knowledge, can use it unless they were able to make a deal with Sigma.

And, probably, every other company has so much invested in their own Bayer-related R&D that, even if Sigma were to open up, it would be rather difficult.

 

[scottgee1]

Well they own the patent. So no one else, to the best of my knowledge, can use it unless they were able to make a deal with Sigma.

And, probably, every other company has so much invested in their own Bayer-related R&D that, even if Sigma were to open up, it would be rather difficult.

Ooops! I missed that chapter of the story. Just looked it up and see that Sigma bought Foveon back in November of 2008; I gotta pay more attention to announcements . . . 😱

 

[Pavel+]

Full Frame? You mean 8x10?
Please ... free us from the tyrany of a snap decision made some 90 years ago when film stock was conveniently grabbed and gelled into a film standard for small format.

We can move forward now ... really! 🙂 APC gives more dof at the 35 equivalent field of view ... and that in a cam like this is nicer than larger frame, larger body, larger lens and larger cost.

 

[kxl]

Am I the only one who would be more interested in Foveon technology if it was employed by someone other than Sigma? 😕

I agree with this. Foveon, great. Sigma, not so great.

 

[antiquark]

Lame, they're still trying to fool us with that "1 pixel = 3 pixels" shtick.

 

[kdemas]

This sensor actually should be pretty amazing, too bad it is going to be stuck, for the time being, on a Sigma SLR with proprietary mount. I you've never shot a Foveon they really are quite different in their rendering (no AA filter helps!).

If Foveon, now owned by Sigma, was a little more shrewd they'd work to get one of the major players to take a flier on it (they're always looking for a competitive advantage). Imagine that new little Juji Hybrid Rangefinder with this sensor on board, would be sweet 🙂

Kent

 

[Olsen]

Sorry, but I don't believe that 45 million pixels is a correct way of describing the resolution power of this Foveon sensor. Which should mean that it makes larger files than, say, Hasselblad with a 39 mil. pixels digital back. Which it does not. A correct figure would be something like 15,3 million.

 

[PatrickONeill]

This would be perfect in the DP1/2 series cameras.

15mp is more than enough for me anyways.

 

[j j]

Whatever the numbers, this new sensor records enough to make a big picture.

 

[user237428934]

Sorry, but I don't believe that 45 million pixels is a correct way of describing the resolution power of this Foveon sensor. Which should mean that it makes larger files than, say, Hasselblad with a 39 mil. pixels digital back. Which it does not. A correct figure would be something like 15,3 million.

That's how they calculated the MPs all the time. The SD14 had around 4.7 MP times 3 and they advertised 14MP.

 

[wolves3012]

That's how they calculated the MPs all the time. The SD14 had around 4.7 MP times 3 and they advertised 14MP.

A quick read shows the Foveon to be a 3-layer stacked device. Pixel=picture element, therefore it has a pixel-count equal to 1/3 that claimed. In terms of equivalent performance, it may be better but in terms of actual numbers of pixel-count, they're just lying!

 

[Sonnar Brian]

Pixel is a Picture Element. Most sensors regard each photo-site as a pixel, whether it is under a red, green, or blue filter. So the Foveon sensor has 46 Million Photo-sites. Most will call that a Picture Element, so 46MPixels works. If you consider Red, Green, and Blue as separate elements. If you consider the combined red-green-blue co-located photo-sites as one picture element then the Foveon has ~15.6 RGB MPixels and everyone else fakes it with interpolation.

I saw a Bell&Howell "12MPixel" IR camera on Ebay for $90. It's sensor had 5Million Photo-sites, and 12Million Pixels via interpolation. So these days, the term "PIXEL" has lost a lot of meaning.

 

[Neare]

Yeah that is right, there are a total of 46megapixels in that sensor. The difference is that the final image dimensions will be smaller than that of an equal 46mp single layered sensor.

It is just different technology - 46mp is huge detail-wise and the dimensions of a 15mp sensor are large enough for most work.

 

[john_van_v]

I don't think that's necessarily true. .

You can think whatever you want, this is a free country (assuming you are in this country and you have not yet changed this feature of it).

Your argument is about something else that I cannot take the time to decipher at the moment.

 

[semilog]

A quick read shows the Foveon to be a 3-layer stacked device. Pixel=picture element, therefore it has a pixel-count equal to 1/3 that claimed. In terms of equivalent performance, it may be better but in terms of actual numbers of pixel-count, they're just lying!

They are not lying perceptibly more than the companies using a Bayer array are lying about their pixel counts. Unless you're dumb enough to think that spatial information is not lost during demosaicing.

 

[pixelatedscraps]

New Foveon: Another milestone in image sensor technology?

*Yawn

I'm pretty sure this isn't the first time Sigma have announced this.