This post of Thom Hogan‘s summarizes up the important elements one needs to consider when designing a new digital camera.
thebard37 wrote: > … can comfortably overlap, but it takes development. This is the Swiss Army Knife approach to design, and one I disagree with. The problem comes when you start compromising one target use to serve another. You can’t be all things to all people. I’m a firm believer in that you must design to a specific target. If others see potential in that and pick it up and live with the compromises, great, that means you have another nearby target you can produce a refinement for. But the minute you make one single compromise to your original target, you open up additional points of failure. > 1.) It looks out of place on the pudgy little G1 that appears > needlessly large (my hunch is that Panny are playing it safe and > giving consumers a look at a camera that feels and appears like a > DSLR when it is something potentially better for a lot of tasks. Oh I hate those words: “playing it safe.” Playing it safe doesn’t win. I can cite way too many examples of that. And it opens up too many ways that competitors can hit you. I’m a firm believer in the “do it right” approach to design. I’d rather have 1000 potential customers who think my product is spot on than 10,000 potential customers who think my product is almost there. No one can take those 1000 customers away from me very easily, but anyone can take those 10,000 customers away. > So > another camera (something like a L/M CL with the VF offset on the > right part of the camera?? Maybe something even smaller with hotshoe > protocols to add on a 35mm sized EVF when it is needed? The question is this: the person who would appreciate a very high quality, near shirt pocket size camera wants what? Well, a pocket Leica, basically. Not a smaller DSLR. So, yes, I think there’s still an open product niche that the G1 isn’t going to nail. > Instead it debuts as something like a superzoom with the most vanilla > lens imaginable Yep. That’s one design decision that has me scratching my head and thinking “they’ve just decided to downsize the consumer DSLR.” Again, there may be a market for a downsized DSLR, but it’s not the camera I personally want, and I don’t think it’s a defendable market long-term as there are no barriers to keep others from doing the same thing. (Before someone protests, the barrier for high-quality compact is this: market size. First there with the right product wins the majority of the users.) > It will sell, but it won’t find its way into the bag of many advanced > amateurs who want a discreet camera to operate beside their > d700/5DII/A900. I’m of a mixed bag on that. The foldable LCD makes it a discreet camera, and if I can turn off the inevitable beeping sounds, it can be very discreet. That’s one of the things that I like to the approach.
Thom also says that amazingly, no camera maker is taking the iPod / iPhone approach to designing camera, something that he feels they should start doing:
Actually, it’s been proven over and over that the more choices and differentiation you give the human brain in products, the more questions they ask and the harder it is for them to make a final choice. Study after study show that our brains do paradoxically the opposite of what we think. That’s why a blank slate product like the iPod or iPhone does well. Both products were designed and pitched as complete breaks from all the competing products. The choice then becomes: (a) sort through all the details of all those competing products: or (b) just buy the iPod/iPhone and be done. No one, repeat, no one is doing this in the camera space. Indeed, it appears that none of the makers even understand how to mount such a project.
Thom gives some tips on how to get into the mind of the camera-buying consumers:
Well, just in case a camera engineer is out there reading this, let me state what should be obvious but apparently isn’t. Consumer cameras are forcing users to define the problem 100% before the camera will do the job of “automatically” setting the camera. There’s nothing “automatic” about this, actually. You have to tell the camera if it’s day or night, whether you’re shooting motion or not, and a host of other things. Funny thing is, if the sensor is running all the time, you’ve got a data stream from which you should be able to figure out all those things. Nikon has played with a variant of this at the high end with the 3D Color Scene Matrix metering, which tries to figure out what it’s looking at before applying a matrix algorithm. But there’s so much more than can be done that just isn’t. Some of this is that it takes some computing horsepower, and that equals cost, but even simple things aren’t being done. I can list dozens.A true consumer—and I still think the G1 is targeted at such consumers, not me—doesn’t want to make 100 decisions or settings to take a picture. They want those decisions narrowed down to a few key ones, most notably the “when” to take the picture. Moreover, I’m really surprised by the fact that no camera really tries to analyze a picture after the fact and suggest how it can be improved. HP and Nikon have both done some work in this area, but imagine a camera that sees that the data is likely blurry and IMMEDIATELY and without prompting took a second putting in place decisions that should improve the chance of getting a better picture. THAT’S a consumer camera.
Just to add to what Thom referred to as 3D Color Scene Matrix metering, the first generation of this technology had been implemented in the Nikon D3, D300, D700 and D90.
Thom has frequently shared on the forums that he helped design the first digital camera but never directly revealed which company he worked at, or the name of the camera. From this post, I believe Thom was involved in the development of the Logitech QuickCam digital camera for the Macintosh platform, and the company was Connectix. MRD probably stands for Market Requirements Document:
When you design, you must have goals and specific targets. The quality level you wish needs to be defined. The size needs to be defined (“small” may be a shirt size, but it’s not a recognizable camera size). “Easy to use” means nothing unless you define what that means. And “versatility” is often just a code word for “it does lots of things, none great.” When I’ve done products and headed up product development, we’ve always had very specific wording that directed the product. I’ll tell you one of the key MRD points for the original QuickCam, for example: “the fewest number of parts that will get sensor data to the CPU.” That may not at first sound like a camera design goal, but it was a statement that we didn’t want to spend any time designing in-camera logic. From the beginning we made the assumption that if we had excellent raw data over at the computer, we had plenty of computer horsepower to do things with that data. That’s one of the reasons why the QuickCam was so darned inexpensive and created new markets.