When you start soliciting coverage from the Wall Street Journal and New York Times, I guess you're looking for attention. And that's just what Lytro got. The question is this: why?
First, let me dismiss some of the "can't be done" or "impossible technology" comments that have been made by some on various Web forums. Light Field capture has been a hot research topic for some time now, and is moving from labs to reality. If you want some nitty gritty, read Ng's PhD dissertation (Ng is the founder of Lytro).
As a Silicon Valley native (born and raised) and practitioner (~20 years), I'm happy to see the Stanford-to-startup connector still working. Lytro claims to have raised US$50 million for their venture.
That plus the early publicity engine has me scratching my head, though. Lytro expects to launch a camera with the technology later this year. I expect said camera to be relatively low in pixel count compared to other cameras. If we're to believe the hints from what they've posted to date, it will also be a square image format, so it will be decidedly at odds with current compact cameras just in those two aspects already, without the Light Field aspects even considered.
It's that Light Field bit that is intriguing, and it may be a Trojan Horse. Is the aim really just to refocus pictures after the fact and then treat the image normally from there, or to come up with a new after-the-fact photo browsing experience? I strongly suspect we're headed towards the latter, especially considering all the "play with me" images on the site. How much do you want to bet that there's going to be a Web Gallery aspect to Lytro's offering, and that the real money will be made there? I'll take that bet.
The problem with pure hardware startups is that you need a lot of sales volume to recover initial capital outlays. At US$500 and a gross profit margin of 40%, you need to sell 250,000 units to just pay back the initial US$50m in VC. And VCs want 20x and 30x recovery. So we're talking 5m units to satisfy the investors already in place (to put that in context, that's more than 5% of the current compact camera marketplace per year and a bigger market share than some very established brands). But I'll bet the GPM isn't 40% and the sales price can't be US$500. Halve those numbers and the payback bar becomes 20m units. It's just tough to make hardware from scratch work. I've tried it three times in my career and succeeded once, got undermined by other factors in the company once, and failed once.
But here's what I really wanted to say: Silicon Valley gets this "tease early" strategy wrong far more often than they get it right. I believe in sine waves. Everything happens over and under the logical "norm." When you first announce something that is incredibly new and innovative, you get over-hyped, especially by the press, who are constantly looking for positive stories. When you deliver a 1.0 product, you get over-criticized. And these swings can be incredibly large. Remember the Segway and what was said about it before it was announced? ("Might hit US$1 billion in sales as fast as any company in history." We're still waiting 10 years later.)
If you're going to play the hype game, this has to come incredibly close to launch, what I call Hype-and-Sell. The longer the hype is out there without a product to support it, the larger the hype tends to build (re-tweeting isn't going to help that problem, and I've already gotten many emails today asking "have you seen this?"; yes, I have ;~), meaning that the actual product will almost always be a let-down. So you get panned by reviewers right at the time you're trying to sell to your first customers. All it takes is for some customers to reinforce that pan and you're toast.
You might ask, "but wouldn't the customers be let down if you hyped and immediately sold them an under-performing product to those expectations?" Perhaps, but you should already have a plan for that, too. That plan consists of fast and constant iteration, each addressing known issues. That, of course, means spending more money on R&D, which makes the hardware gamble all that much bigger. Big, existing companies can make this bet easier, as they aren't betting the company, but they don't tend to because they get complacent.
So what's the hype here? After-the-fact focusing. What's the downside? Well, the prototype required a 16mp sensor array to produce a 90kp image. Some similar relationship is expected for a production camera. The good news is that using so many photosites per pixel (256 if I read the papers right) also means that a number of noise-related issues go away, The implications, however, are that we need one heck of a lot of megapixels to get large sized images. Their samples appear to be no more than 360kp images.
Don't get me wrong, the Hype Early strategy can work. But it's a big gamble for a startup. It makes everyone look very hard at what you do eventually release. Are you still going to be enthused if the camera produces 600x600 pixel images?.