 |
The Sigma SD9 was announced in February 2002, just days after Foveon had revealed the X3 technology for which it would become known. Not only was it the first X3 camera, it was also Sigma’s first digital camera, following on from a history of making film SLRs.
And, while we’ve seen alternatives and variations of the Bayer color filter pattern, the SD9 represented one of the most concerted attempts to offer a totally different way of capturing color in digital photography. As part of DPReview.com’s 25th anniversary, we look back at one of the most radical cameras we’ve yet tested.
3.4 x 3 = ?
The Foveon X3 works on a completely different principle to most imaging sensors. The majority of digital cameras place color filters in front of each photodiode so that each receives only specific colors of light, letting it measure how much of that color is present. This has two downsides: firstly it means that a lot of light is absorbed by the color filter and is not then available to the camera. More critically, though, it means that different colors are captured at different locations, and the chroma resolution of a sensor is much lower than the pixel count would imply. The colors of light that were lost to filtering need to be interpolated from neighboring pixels, which softens the image by sharing information across pixels and creates more opportunities for aliasing.
The X3 sensor works very differently: it doesn’t use any color filters and instead essentially stacks three photodiodes on top of one another. Different colors of light can penetrate silicon to different depths, so color information can be derived based on how many photons reach the different depths within the sensor. There are downsides to this approach, too: the sensors are complex to fabricate and it’s hard to match the pixel counts of the much simpler conventional designs. Also, you need a lot of light to ensure that enough penetrates into the lower layers of silicon, to correctly assess the color and keep noise under control.
The SD9 had 3.4 million pixels, with three photodiodes at each pixel, giving full color at every location from its approximately APS-C sensor (∼21x14mm). A direct contemporary, Canon’s D60, had 6 million pixels, but with more standard Bayer color filter meaning that although they had a higher nominal resolution, in practice they were softer and couldn’t match the Sigma’s pixel-level performance. In his original review, Phil concluded that the two cameras were capturing comparable levels of detail overall.
 |
 |
| The SD9 had two battery systems, one to power the systems you’d find on a film SLR, a second to power the digital side of the camera. | |
On the physical side of things, Sigma had clearly adapted one of its existing SLR designs, with separate batteries, with one set powering the camera’s metering system, autofocus drive and viewfinder displays, while a tray of AA batteries powered the digital side of the camera. 2003’s follow-up SD10 used only a single set of batteries, with a dedicated Li-Ion approach being adopted by 2006’s SD14.
Despite being Sigma’s first attempt at a digital camera it had some thoughtful touches such as an RGB histogram that updated to reflect the zoomed-in area you were reviewing, in playback mode.
The review also includes Foveon’s marketing illustrations that suggest the three different layers of silicon capture blue, green and red light at the different layers. This isn’t strictly true: only red photons will reach down to the lowest levels, but some get absorbed higher up. The middle layer therefore collects some red and most of the green light, with the upper layer collecting some red, some green and all the blue light. This means that the contribution of each color has to derived, with the noisiest ‘red’ channel factored into all the other color information.
The SD9 wasn’t able to do these calculations, so the camera output Raw files that could be processed using the company’s Sigma Photo Pro software. And even on its first outing, Phil recognized that lower light levels represented a significant challenge. But he also recognized the level of innovation that had gone into the camera.
Still promising
Twenty two years later, Sigma now owns Foveon and its X3 technology and says it’s is working on a full-frame variant of the design. We’re hoping that advances in sensor design and fabrication, along with a larger surface area to capture more light will deliver something more of the original promise of the X3 concept.
With pixel counts now far exceeding those required for most applications, the pressure for Sigma to match the latest Bayer sensors pixel-for-pixel is reduced. And if that results in a low-ISO camera that produces pixel-level perfect images, then perhaps there will still be a market for that.
Author:
This article comes from DP Review and can be read on the original site.
