Canon global

Next, I would like to ask about HD CMOS PRO. What is the greatest characteristic of this CMOS sensor?

Koizumi

The ability to shoot beautiful video even in dark settings. We made the perfect sensor for parents who want to capture video of their children in a play or recital. It more than doubles the sensitivity of earlier sensors when shooting video; it's a sensor that I wanted for myself (laughs).

High sensitivity delivers bright, noise-free video even in dark settings

How did you achieve this in terms of technology?

Koizumi

It's not something we achieved by incorporating a huge technological feat. We simply improved the sensor technologies that we had accumulated up to that point.
First, we reduced the number of effective pixels to 2.07 million, which is the same amount as 1080p high-definition video, greatly increasing the size of each pixel. I think that reducing the number of pixels was a very bold decision in the planning of the product. The previous generation of CMOS sensors offers over 8.5-million pixels and can capture stunning still images, but sensitivity is somewhat lacking. If the overall surface area of the sensor is the same, reducing the number of pixels increases the size of each pixel. The efficiency at which a pixel collects light improves as the area of the pixel increases, so decreasing the number of pixels inversely increases sensitivity.

Comparison of CMOS sensor pixel sizes

On top of this, we increased the curvature of the on-chip micro lens (a convex lens on each pixel that collects light on the photo diode) placed on each pixel to increase the area of the photo diode receiving light. We also further increased sensitivity by 20% through the placement of wiring in each pixel as a result of being able to efficiently collect light based on optical simulations.

I see. So you can capture beautiful video even in the dark thanks to improved sensitivity?

Koizumi

No, I'm afraid not. While the sensitivity of the sensor has been greatly improved, that's not enough to shoot beautiful video. The light that we went through all the trouble to collect can't be extracted as an electric signal unless the charge stored in each pixel is increased. To achieve this, we not only increased the area of the photo diode, but also the depth of the electrons stored. If the amount of charge that can be stored is increased, the dynamic range increases, significantly improving performance.

By increasing the area of the photo diode and the depth of the electrons stored,
the amount of charge that can be stored in each pixel was increased

HD CMOS PRO doesn't consume much power, does it? Was it difficult to achieve improved sensitivity with reduced power consumption?

The DIGIC DV III image processor plays the role of bringing out the best performance of the lens and CMOS sensor. What are the main points that contribute to high-quality video?

Nakayama

It's hard to pick out a single point because it broadly moves across both image processing and control, but the excellent performance offered by the DIGIC DV III are its automatic control of such aspects as focus and aperture, noise reduction, dynamic range control, and adjustment of color and sharpness based on shooting settings.

How does it work with dynamic range control, for example?

Nakayama

If it determines a high-contrast scene, with a large difference between light and dark areas, it automatically expands the dynamic range to limit overexposure and underexposure as much as possible. Dark areas are made brighter and the amount of brightness in bright areas is decreased. When doing this, the problem is noise in dark areas, since the noise is also amplified and needs to be eliminated. Methods such as these are used to create high-quality video based on the high sensitivity of the CMOS sensor.