DIGIC 4 provides a three-fold improvement in face detection speed and high-sensitivity processing speed, a ten-fold improvement in motion detection speed, and improved shadow correction function. How has all this been achieved?
- Ikeda
- Shadow correction involves making dark areas brighter, so if you do it in too basic a manner the noise problem gets worse. For example, let's say you're taking a photo of someone's face lit from behind so the face appears dark. If you take it without using a flash and then perform shadow correction, the face appears grainy. That's because there's too much noise. The naked eye can see the face clearly enough, but the camera can't. DIGIC 4 overcomes this problem and creates an image closer to what the human eye would see. What makes this possible is low-noise technology.
- Kawaguchi
- The shadow correction function is very useful if someone's face is dark.
- Ikeda
- The face detection function is extremely important in terms of improving image quality. If the camera is able to automatically detect that there is a face in the frame, then the face part of the image can be processed in a way that maintains the overall quality of the image.
Users of compact digital cameras in particular tend to take a lot of photos with people's faces in them, don't they? So the combination of a face detection function with shadow correction should be very useful for them. By the way, with face detection, how does the camera determine whether there is a face in the frame?
- Ikeda
- It's based on the contrast between light and dark areas on the face, or in other words the shadow pattern. With DIGIC 4, not only has the detection speed been improved, the degree of precision has also been enhanced, so faces can be recognized as such even if they are viewed in profile or obliquely.
Does detecting faces that are viewed in profile, or at an angle, present major technical challenges?
- Ikeda
- If you were to ask the specialist team responsible for developing face detection technology, they would tell you that because hair takes up so much space on a human face in profile, it's hard to identify areas of shadow, which makes detection very difficult. Face detection technology was originally developed for security applications, where the emphasis is on faces viewed from the front. In the past, applications that required the detection of faces in profile have been rare. The problem is that, when you're taking a photo of a baby or a young child, it's very difficult to get them to face the camera directly. So digital cameras do need to have the ability to detect faces in profile, and we have been working to develop this capability as quickly as possible. We just managed to complete the development of this technology in time to have it incorporated into DIGIC 4.
- Rengakuji
- Once we clarified the problems that had to be overcome with profile face detection, the process of developing the technology could proceed reasonably smoothly. That's true with any technology development challenge.
I see. And then after achieving a high level of precision in face detection, you were able to exploit that to improve the auto-focus function?
- Rengakuji
- That's correct. The camera can determine that the intended subject of the photo is a face in the central area of the frame that is moving, and then keep adjusting the focus accordingly.
- Kawaguchi
- The reason why we have been able to make this function work so well is that, thanks to improvements in design technology, the active detection speed of DIGIC 4 is ten times that of DIGIC III.
I can see from what you've said so far that improving processing speed has been the key factor in the development of DIGIC 4. Besides ensuring that the camera user doesn't get impatient because of slow image processing, it would seem that a high processing speed also helps improve image quality and overall camera performance. So improving processing speed was not just an end in itself, it was also a means to achieve a number of other development goals. Would I be right in thinking that enhanced processing speed was also the key factor that enabled you to achieve full HD (High Definition) video recording capability with DIGIC 4?
- Rengakuji
- The main factor there is the high performance of the image sensor, but it's certainly true that DIGIC 4 is capable of processing the large quantities of high-speed signals sent to it by the image sensor.
- Ikeda
- Another point is that, even with video recording, there is no signal loss. You can record video footage with almost the same high definition as still images. That's another advantage that high-speed processing provides. I think we can safely say that the good reviews the EOS 5D Mark II has received for its outstanding full-HD video recording quality are mainly attributable to this.
