PhaseOne Sensor+, pixel binning, and the future of digital photography.

editorial May 15, 2009

With the recent announcement of the P40+ MF digital back with Sensor+ technology, Phase One has brought to the market two digital backs with a feature called pixel binning (the top-of-the-line P65+ is the other Sensor+ enabled unit). Pixel binning is a method the combines the input from several pixels on a digital sensor into one “super-pixel”. This has the effect of reducing the overall pixel count of the image, but increase the speed and sensitivity of the sensor without adding noise.

USA.NM.VeryLargeArray.03.jpg

Pixel binning is not new. Astronomers have been using this technique for decades. The NRAO’s Very Large Array Telescope is this exact technology on a massive scale. By combining the inputs of 27 25-meter radio telescopes, the VLA behaves like one giant radio telescope.

Doug Peterson of Capture Integration talked with me at length about the advantages of this method and the unique (and patented) way Phase One has implemented it on their digital backs.

Let’s say we have a digital sensor and each pixel gathers 10 electrons worth of signal (10e). As the data is passed through the required amplification process, another 10e of noise is added. Thus each pixel has 10e of signal and 10e of noise resulting in a signal to noise ratio (SNR) of 1:1. In traditional pixel binning, the signal from multiple pixels is averaged to improve SNR. The amount of improvement is based on the square root of the number of samples used in the average. If, for example, we averaged the signal from 4 pixels, the total signal for this “super-pixel” is 40e and the total noise is 40e divided by the square root of 4 (which is 2) = 20e giving us a SNR of 40e:20e or 2:1. While this is great for radio telescopes, it’s not quite good enough for digital photos.

Another issue is related to image quality. All camera sensors (except Foveon based ones) use a Bayer pattern (RGGB) to approximate color values. In a 4×4 pixel array, your colors are laid out as follows:

R1 G1 R2 G2
g1 B1 g2 B2
R3 G3 R4 G4
g3 B3 g4 B4

By averaging the colors together you get a 2×2 super-pixel with the colors averaged as follows:

R1,R2,
R3,R4
G1,G2,
G3,G4
g1,g2,
g3,g4
B1,B2,
B3,B4

This looks all well and good, however, this method drastically reduces the resolving power of the sensor. If you look at the optical centers of the original 4×4 matrix concentrating on the green pixels, there are some gaping holes in the array and portions where they do not overlap. This has the effect of reducing the resolving power of the sensor by a factor of 16!

G1 G2
g1 g2
G3 G4
g3 g4

The Phase One Approach

Phase One tackles the noise issue by moving the signal combination step into the sensor itself, prior to the amplification step that injects noise. In our previous example, 4 pixels are sampled and combined in the sensor to produce 40e of signal. This is sent to the amplifier and 10e of noise is added. The SNR of this method is 40e:10e or 4:1. This means that the super-pixel noise level is now 4x lower than the full resolution pixel at the same ISO. Doug equates this to 4 people speaking loudly and discordantly vs. 4 people singing in chorus (“resistance is futile”).

To overcome the quality issues, Phase One takes the data from the green pixels and turns them 45 degrees relative to the red and blue pixels. This results in a pattern like this:

R1 R2
G1
B1
G2 B2
G3
R3 g1 R4 G4
g2 B3 g3 B4
g4

The resulting green pixel pattern fills in all of the gaps. The practical benefits are improved ISO sensitivity (clean images up to ISO 3200), faster image capture (up to 1.8 fps on a P40+) all the while maintaining 16-bit/pixel image capture. Another benefit of this approach is that even in Sensor+ mode, the entire digital sensor is used to capture the image without any cropping or aspect ratio changes.

Here is a sample image at ISO 3200:

Sensor+ Sample

So now with Sensor+, a photographer does not need to own two systems – one for high-resolution images and one for jobs that don’t require MFD resolution. Phase One’s solution means its cameras and backs can be used efficiently for giant landscapes, architectural spreads, and high-end studio work and now for Facebook profiles, senior portraits, and web based imagery. The smaller, high-quality files produced in Sensor+ mode also speed up workflow and cut down storage requirements drastically.

Since Phase One has patented this system it is possible that they will license it to other camera makers. Both Doug and I feel that this technology is too big to be kept in the medium format realm and will probably make its way into mainstream pro DSLRs within two generations.

I would like to thank Phase One, Capture Intergration and especially Doug Peterson for helping me gather information for this article.

Great! You've successfully subscribed.
Great! Next, complete checkout for full access.
Welcome back! You've successfully signed in.
Success! Your account is fully activated, you now have access to all content.