Recent Blog Posts
-
Mapping Company Raises Millions
Nov 20 20094:09 pm EDT -
Facebook Valuations Are All Over the Map
Nov 20 200911:30 am EDT -
The Future of Tech, 2010 Edition
Nov 20 20099:13 am EDT -
Automatic Pancake-Making Machine Attracts $2 Million in Capital
Nov 19 20094:53 pm EDT -
Apple Talk of Microsoft's Annual Meeting
Nov 19 20091:27 pm EDT -
There Is Still Hope for the News Business
Nov 19 200911:50 am EDT -
The Google Phone May Be Near
Nov 18 20094:10 pm EDT -
Amazon Grocery Service Goes Mobile with iPhone
Nov 18 20099:13 am EDT -
How Microsoft Blew It in Mobile
Nov 17 20093:55 pm EDT -
Ten Reasons Why Startups Fail
Nov 17 20092:18 pm EDT
Links
- Engadget

- Pandora

- GigaOM

- USA TODAY Tech

- Todd Bishop's Microsoft Blog

- Somewhat Frank's tech conference list

- BuzzTracker Tech

- The Long Tail

- Tom Foremski

- Roger McGuinn's Folk Den

- John Battelle's SearchBlog

- Mark Cuban's blog

- SciTech Daily

- Romenesko

- Kevin Maney's site

- Steven Johnson

- Marc Andreessen

- TechCrunch

- Fred Wilson

- paidContent

- Spiedies, mmmm

Cameras That Understand What They're Shooting
As the latest issue of American Scientist points out, digital cameras are shifting from just recording the image they see, to knowing something about what they're shooting. That way the cameras can adjust light and color to take a better picture. A few months ago, Kodak CEO Antonio Perez talked about some of that when I interviewed him at Dartmouth. Photo software is getting smarter, too.
Here are excerpted comments from Perez about all that:
Q: What about cameras being able to recognize that it's grass in the picture, for instance?
A: There is no such thing as a digital camera yet. In my view, we're not selling digital cameras yet. The industry is selling filmless cameras. The first automobiles, they were called horseless carriages, because they took the carriage as it was, and they took the horse away. They put in a motor, and then that thing was moving. And they said, well what is this? Well, this is a horseless carriage.
This is what the digital cameras of today are. To make a camera, you basically need only three things. You need a lens; you need a shutter; and you need something that captures the light. In the past, what captured the light was the film. And the whole structure of the camera was around the film.
You look at the digital cameras today, they actually look like the old cameras. And they have no reason in the world. I mean, who said that you have to hold a camera? Why do you have to hold a camera? It's because in the other camera you had to hold them, right. You buy cameras -- we sell these too -- that they have a little thing in the side that is like the place where we used to put the roll. Actually people think when they have that, that it's a better camera! It's just a piece of plastic.
We have, for instance, we have flash. Most of the digital cameras have a flash. Well, a sensor can see much darker than film. So you don't need a flash with a well-designed sensor. Now we are designing those sensors. Kodak is designing those sensors now, so you will never need a flash, which will make a much better picture.
The idea that you have to change the lens -- well, you have to change the lens all the time because the sensor was the film. But there's nothing that will stop you now from putting five sensors in a camera. The sensors are very small. So you can have one that is wide angle and one is for distance.
Q: And the pictures themselves will have more information?
A: I have a collection of twenty thousand, or twenty-five thousand digital files with my family. I can go and say, this one is Emily in front of the Christmas tree. Certainty it's easy (for the photo software) to identify Emily, and identify the Christmas tree. So I'll have Emily from the moment she was born every year all the pictures that are taken, without having to organize anything that would come with that.
I could say, I want to see Emily's friends when she was five. This system will identify Emily, it obviously knows who Emily is, and will find Emily, and then it will find the five or six little girls or boys that tend to be with Emily most when she's five, and they will appear like that. The system will identify candles and in a white dress and a birthday cake, and a boat and the sea. So there's really no need for you to label your pictures.

Credit: Steve Marcus/Reuters/Landov






