By the way, when you start work on a new project, you always have certain specific targets that need to be met, don't you? For example, with DIGIC 4 you wanted to, among other things, achieve a three-fold improvement in the high-sensitivity image processing speed and make sure that the shadow correction function provided a certain level of performance. How were these individual objectives decided?
- Ikeda
- The key factor here is the discussion between the different team leaders on how the available space on the LSI chip should be allocated. Once you've decided that, then you've more or less decided on the target figures for everything else.
- Kawaguchi
- Actually, Ikeda kept asking us, "Please try and do this, too. And can you also do this?" (laughs)
- Ikeda
- Hey, don't give them the wrong idea! I just wanted to make sure that we tried to include the most advanced technology possible during the development period. I sometimes hear people saying that Canon only introduces new technology in dribs and drabs, but that's not true at all. As someone involved in technology development, I can say that with my hand on my heart.
- Rengakuji
- All the same, given that you have a deadline by which the development process must be completed, you have to draw the line somewhere.
There must be a conflict between the development team's desire to perfect the technology and the need to quickly commercialize and make products available to the end user.
- Rengakuji
- Yes, but the team leaders get together and discuss these things. There is very close coordination between the teams.
- Kawaguchi
- It's an ongoing series of exchanges along the lines of: "When can you get this item done?" "No, that's no good. Can't you get it done earlier than that?" "Well, we'll try our best."
Do you ever get into arguments with one another (laughs)?
- Ikeda
- Because I'm so close to the actual design process, I don't make unreasonable demands (laughs). What particularly impresses me about the circuit design team is that, even if the deadline has already been reached, if I ask them "Do you think you could incorporate this element as well?" they usually manage to do it!
So they've really got backbone (laughs)! From your point of view, Mr. Sekiguchi, would you say that the development objectives are always well defined?
- Sekiguchi
- Yes. The key tasks for us are to arrange the routing so that there is no interference between the high-speed analog and digital signals, and to make sure that the electrical characteristics match the specification. We also try to design things to be as small and lightweight as possible. These two objectives have remained unchanged with every new version of DIGIC.
I see. What was the most difficult challenge you faced during the development of DIGIC 4?
- Rengakuji
- I would say that the biggest problem was predicting [future trends].
- Ikeda
- That's right. At the point in time when you start on development work, you have to think about what kinds of camera algorithm will be in use several years from now, and what level of image quality and functionality the market will be demanding by then. Making those kinds of predication is very difficult. There are so many possible algorithms that you can't hope to include them all in the same device. You have to try to narrow them down based on the technology predictions. If you make a mistake in the predicting, it can have serious consequences.
- Rengakuji
- Making predictions is also a problem in the work that I do. When you are designing the interface that will receive the signals from the sensor, you have to try to predict what level of performance the sensors used in products equipped with DIGIC 4 will have in several years time. You have to make sure that the interface will be able to support the performance requirements of the sensor.
It must be very difficulty trying to predict how technology will develop in the future.
- Kawaguchi
- By comparison, the difficulties when designing circuits are more concrete. The challenge is to allocate the limited space on the LSI chip in the most efficient manner possible. And because you can't really make any major changes to the number and position of the pins on the LSI, you don't have much flexibility in terms of the exchange of data between the LSI and memory, for example. The difficult thing for us was designing it so that as much of the processing work as possible was carried out within the LSI chip. Given the need to achieve higher performance than DIGIC III, it was vitally important to ensure both efficient space allocation and effective processing within the LSI.
How about you, Mr. Sekiguchi?
- Sekiguchi
- Each new version of DIGIC has provided improved image quality and enhanced processing speed. So with each new version, designing the circuit board wiring has become more difficult. Particularly with regard to high-end camera models that have multiple Flash and DRAM memory chips, we had to really rack our brains to make sure that the circuit board layout would not compromise DIGIC's performance, and to ensure that the wiring would maximize data transmission speed.
Were there any particularly memorable events during the development of DIGIC 4?
- Ikeda
- Well, there was Rengakuji's "last-minute bug fixing." (laughs)
What happened?
- Rengakuji
- Just when we thought we had finished everything, we found a bug in one of the smaller circuits. We discovered it just before the start of the summer holiday, so of course nobody volunteered to try to fix it. In the end, I fixed the problem during the first week back at work after the end of the summer holiday, but I'd spent the whole holiday racking my brains as to how I could fix the bug. It was the least relaxing summer holiday I've ever had (laughs)!
