Sigma releases SD1 (46Mp)

Is this sensor really so special or just a marketing?

"46 megapixel 24×16mm APS-C X3 Full-colour image sensor
The 46 megapixel (4,800×3,200×3 layers) 24×16mm APS-C X3 direct image sensor featured in the Sigma SD1 captures all primary RGB colours at each and every pixel location, ensuring the capture of full and complete colour. Using three silicon-embedded layers of photo detectors, stacked vertically to take advantage of silicon’s ability to absorb red, green and blue light at different respective depths, it efficiently reproduces colour more accurately, and offers sharper resolution, pixel for pixel, than any conventional image sensor. Since colour moiré is not generated, the use of a low-pass filter is not required, meaning light and colour, generated by the 46 megapixel APS-C X3 direct image sensor, is captured with a three-dimensional feel."

It is different. As for special, it all depends how you look at it. Right now, all Sigma cameras use a lower res. version of that sensor, and the results are not like sensors with a Bayer grid. I wouldn't say it's better or worse, just different.

They're being a bit silly calling it 46MP: it isn't, it's ~15MP.

On normal sensors, there is what's called a Bayer pattern filter. To greatly simplify, the sensor only 'sees' the light level. There is then a filter which 'marks' each pixel as being of a particular colour. So for example one pixel is red, the one next to it is green, the one next to that is blue and so on.

On the Foveon, each pixel captures colour 'properly' - every pixel records every colour (something to do with different colours of light penetrating the surface to different depths). So a 15MP Foveon sensor in a Sigma body is, theoretically, 3x better at capturing colour (I'm not sure about sharpness and other qualities) than a 15MP Sony sensor in a Nikon or Pentax body. That's where the stupid 46MP claim comes from.

AMD did something similar with their computer processors a few years back. In the Pentium 4 era when Intel based everything on clock speed, AMD started giving their processors model numbers to reflect what AMD claimed was their performance-equivalent in Intel terms. So for example one was called the "AMD Athlon X2 3800+" because AMD claimed it offered performance equivalent to a 3.8Ghz Pentium even though its clock speed was 2Ghz. In actual fact it didn't (short story: it was an early dual core processor and software was written for single-core high-clockspeed processors) but I'm aware that the relevance of this is tenuous at best so I'll keep it to a minimum

The Fovean sensor is logical in its construction in layers - very similar to the construction of a colour film.

Sadly, performance has not so far been noteworthy and for whatever reeason only Sigma, a small player, have taken it up.

It's worth continuing to watch.

Sigma likes the sensor so much, they bought the manufacturer.

I remember reading about this sensor in New Scientist or one of those when it was just a gleam in a scientist's eye. It is one to watch.

I personally love the results from Foveon sensors, I'm just not so enamoured with the cameras they come in. I'd like to see Pentax offer one of their bodies with a Foveon option - I'd buy it.

Bayer filter contributes to softnes, so lack of it in theory may help with sharpness.

Would you remember those Optik sets that Magnus Pyke endorsed in the 80's - you could actually construct a plastic film camera and telescopes with the kit - which worked to a degree....

The SA/SD cameras just remind me of that...

I do remember - that's probably what's putting me off!

Thanks for the info.

I have seen a Sigma SD14 converted professionally to a Nikon F mount....

The Price

http://www.dpreview.com/news/1105/11052010sigmasd1.asp

The camera looks well specified but if I had that kind of money, it'd go towards a 645D.

Absolutely pointless IMHO to spend thousands of pounds on a camera and only be able to use Sigma lenses on it.

Fovean sensors look different in some ways, but whether its better is a matter of taste. Bayer filters may in theory reduce sharpness, but in practice it doesn't seem to matter.

Now photo diodes that "see" in colour would be something else!