Dick Lyon dispels FUD on Sigma’s New Sensor

The biggest news in the photographic world this week is Sigma’s announcement of the DP Quattro series of cameras with a new take on its (in)famous Foveon sensor.
Product shot of Sigma DP2 Quattro

There are no cameras out there in the wild to be tested yet and Sigma has provided no RAW samples so the photographic community has been indulging in a great deal of speculation with, for example, multiple threads with hundreds of posts in the Sigma forum on DPreview. The reason these cameras generate so much interest is their use of Foveon sensors rather than Bayer sensors. Foveon’s Wikipedia entry provides an introduction to these sensors, there are a series of technology papers at Foveon’s site and this is a good simulation of how these different sensors’ imaging works.

Apart from debates about the radical design change to the body and how ergonomically efficient it is likely to prove, the discussion has generated a great deal of FUD about its new sensor. On all previous Foveon sensors the three layers have been of equal size (on the most recent Merrill sensors 3 * 14.6 megapixels) but on the Quattro sensor the top layer is 19.6 megapixels and the lower layers 4.9 megapixels; in other words, these lower layers have photosites 4 times the size of those on the top layer. The sensor, it would appear, falls somewhere in the middle of being best described in RGB or LAB terms. The top layer records luminance + blue(ish) data and the lower layers record chrominance data on the second green layer and third red layer.

The larger photosites on the lower layers mean that there is a loss of data compared to the previous Merrill sensors but also that there is less noise, the single biggest problem with Foveon sensors. The loss of data, however, is not as bad as might be expected because there is diffraction/dispersion on these lower layers anyway, especially on the lowest (red) layer.

All this results in a camera that WILL perform better in low light with many people expecting (hoping) for usable results at ISO 800. Time will tell whether the images that it produces are better or worse or indistinguishable from previous generations of the Foveon sensor.

That doesn’t stop the arguments and speculation continuing apace of course but, interestingly, legendary US inventor Dick Lyon, cofounder of Foveon (sold to Sigma), has joined the debate on DPReview a few hours ago. Dick has confirmed that when they wore working on the Foveon sensor back before he left in 2006 they had extensively tested this new design configuration and been happy with the results despite lacking components of the quality Sigma now has access to (“low-noise fully-depleted top-layer photodiodes”). His view is that whilst it is possible it is very hard for most people’s eyes to determine that chrominance detail is less sharp than luminance detail.

In advance of seeing any sample images it would appear then that Sigma may be onto a winner here, manufacturing cameras that produce images at least as good as their predecessors, and gain a couple of stops of low light usability. Of course, these cameras are likely to remain niche cameras, as in use they will never have the flexibility of other cameras. Indeed, disappointment has been registered by those who were hoping that Sigma’s next move would be a mirrorless interchangeable lens camera rather than another three fixed lens cameras.

I would not be surprised if they launch at around £1000 or more in the UK. However, this does mean that the DP Merrills are now available at £339 in the UK, down from £799. Not at all bad for a camera with an acuity that has been compared favourably to a D800e, a camera that costs 10 times as much or more with a lens with comparable quality.