Автор: scuba_d
Дата: 25-03-10 16:00
http://www.bythom.com/index.htm
Mar 22 (commentary)--The latest overhyped technology, InVisage Technologies' just announced "quantum dot" sensor seems to be provoking a lot of not very diligent reporting, as we have multiple sites stating that it will change cameras and imaging dramatically. It reminds me a lot of the original hype surrounding the Foveon sensor.
Consider that these venture-backed companies have to make noise. Otherwise no one notices them. The big sensor companies already have a lock on most product designs out through mid-2011 at this point, so not only do you have to make lots of noise if you're the newcomer, but you have to do it early. And early it is: the technology InVisage is talking about isn't even going into early production sampling until the end of the year. In other words, if you wanted to stick something into a development product today you couldn't. To put that in perspective, that means that their technology won't likely get into an iPhone two generations downstream. (For those that don't understand that, we've got a new generation of iPhone coming this summer, already committed to with a 5mp BSI sensor. Plus the generation after that, likely launched in the summer of 2011, would need to have a sensor decision lock down before InVisage can deliver production technology.) Worse still, what InVisage announced today will have to compete against next generation BSI sensors from the big fabs.
But it strikes me that most of those reporting this "breakthrough" are just regurgitating the marketing materials from InVisage and making some errors in conclusions while doing so. The biggest issue with small image sensors right now is the color filtration. This removes as much as 2/3 the light. If you note the InVisage sensel design, it still has a filtration layer on top of the quantum dots. Thus, their gains are coming post filtration (i.e. on the 1/3 of the light that's left). Moreover, at least one of the documents they've provided makes the claim that the current state-of-the-art is a sensor that converts 20% of the light getting through the filters. That's not what I hear. Some sensor companies are already claiming 50% efficiency with current sensors, which suddenly knocks a big chunk off that 4x claim. Put another way, it appears that InVisage is comparing a future technology against a past technology.
This has all the earmarks of Foveon all over again, with one slight difference: since the InVisage technology appears to be a coating layer that replaces the photo diode, they may be able to license it to existing sensor makers if they can convince those makers that it has enough impact. Indeed, if it were my US$30 million invested, I'd be pursuing that as Job One. Why? Because there's a lot of IP in the rest of the sensor, and it's moving fairly rapidly at the moment. To think that a venture-funded startup can catch up to the big sensor makers in the rest of the sensor design and fab techniques without stepping on IP at the same time as coming up with an entirely new process that layers on top is a little bit beyond expectations. (I should note that no one seems to know how InVisage is getting the electrons out of the quantum dots, so perhaps there is more going on in the overall sensor design than I'm guessing. Still, a sensor is not just about light capturing. It has to store and read, and communicate the data, too, so readout noise, cross-talk, and other things in the sensor need to be state-of-the-art, too, thus my comment.)
Indeed, if you read the press release and other materials carefully you don't see InVisage claiming that they've got a sensor. They mention QuantumFilm (the layer) extensively, but the language regarding sensors is much more opaque and vague ("the first QuantumFilm-enabled product, due out later this year..." my emphasis added).
Meanwhile, the reporting on this technology announcement is doing exactly what InVisage probably hoped for. "...could potentially ditch the Nikon and rely solely on your HTC" reports Gizmodo. Yeah, right. Also, how did they get "four times sharper...with double the dynamic range"? They appear to be guessing the 4x efficiency claim means that you can have 2x the pixels (no). Even if that were true, where does the double the dynamic range then come from, Gizmodo? Sloppy math, at best, but also a misunderstanding of sharpness, resolution, area, size, efficiency, and dynamic range, all in one nice compact misstatement.
I wish InVisage well (yes, I really do). If they can increase efficiency of small sensors in any meaningful way (it doesn't have to be 4x) then we have something worth pursuing. But having played the Silicon Valley startup game many times in my career, I'd suggest that coming out in total hype mode that you can't live up to and letting the blogosphere run wild with that is going to come back to hurt InVisage, not help them.
Basically, it boils down to this: QuantumFilm increases the cost of producing a sensor by some factor and it increases the efficiency of the post-filtration light detection by some factor. If cost is too high, the efficiency increase too modest compared to other technologies that are in development, or they have to trade off any other sensor metric (e.g. readout noise), the idea will get passed by like most tech innovations. Simple as that. Silicon Valley graveyards are filled with such ventures.
![[beer]](https://photo-forum.net/forum/smileys/smiley20.gif) ![[shtrak]](https://photo-forum.net/forum/smileys/shtrak.gif)
|
|
![[smilie23]](https://photo-forum.net/forum/smileys/smiley23.gif)
![[smilie24]](https://photo-forum.net/forum/smileys/smiley24.gif)
4/3 ![[thankyou]](https://photo-forum.net/forum/smileys/thankyou.gif)
![[smilie5]](https://photo-forum.net/forum/smileys/smilie5.gif)
![[smilie3]](https://photo-forum.net/forum/smileys/smilie3.gif)
Стани фен на Фото Форум във Facebook
Фото Форум в Instagram