Jaime’s notes

it doesn’t really seem like the problem is that difficult….everything exists, it’s just all in different projects

there’s a bit of a hardware handicap….hardware to do it the right way is really expensive….so you either have to skimp on features or carry around an 80-pound backpack

ok, the main hurdles are: 0 video/still capture: how to deal with motion blur, focus, making sure you get the shot you want, etc. 0 GPS integration 0 facial recognition 0 audio capture and storage: also matching a voice to a person 0 integration and data mining: being able to store and retrieve a single word said by a specific person at a specific place at a specific time

after the main technology hurdles: 0 get hardware size and costs down 0 need a usable UI; possible integration with cell phones, earpieces, heads-up displays in glasses for playback, etc. 0 deal with social issues: privacy concerns, selling people on the idea, etc.

Chris’ response

OK, bluetooth. Whew. It’s a neat standard even now, and in 10 years it (or its descendants) will be amazing. I’m imagining that all the LifeCam interconnections (cam, storage, control pad, playback device) could be bluetooth, but it also opens up locational connections, too. Since a standard bluetooth device has a 10m range (I think),

For instance:

• Share multi-angle photos or video. If more than one LifeCam user is in the same location, the units share photos and video between themselves. If the guy in the front of the gondola has a better view of the bridge, you can intercut your video with his (even much later) to improve the quality of your “memory”. Even better, the guy on the bridge watching you as your gondola passes underneath.
• Pick up “official” audio tracks for locations. That same bridge has a bluetooth station broadcasting interesting facts about it.

Think about what a standard system was in 1994:

Pentium 100MHz or PPC 603 80MHz 16MB RAM 500MB hard disk 2x CD-ROM 28.8Kbps modem 15″ CRT (1024×768) DOS 6.2 or System 7.5 or Linux 1.0

…and for the same price now:

P4 3.2!GHz or G5 2!GHz 2GB RAM 500Gb hard disk(s) 48x CD-RW / 8x DVD-RW 1Gbps ethernet + 50Mbps wireless 17″ LCD (1280×1024) Windows XP or Mac OS X 10.3 or Linux 2.6

So, 1000x disk space, 30x clock, 100x RAM, 30000x network (or 2000x with wireless).

OK, now think of it the other way around. I have a 100 MHz computer with a 490×240 display, 528MB disk, 1.5Mbps network. I keep it in my pocket. I take photos with it. It cost about $500, but that was almost 3 years ago.

Apple had a digital camera in 1994, the QuickTake 100:

http://www.epi-centre.com/reports/9403cdi.html

It cost over $1000 and took 16 VGA photos at a time. It looked like a pair of binoculars. My work had one, it was pretty awful. People looked like they were made of lumpy cheese.

What I’m getting at here is the way we can predict what 2014 might bring. Given the QuickTake and that Pentium 100, could we have developed iPhoto? Given a DiscMan and that same Pentium, could we have come up with the iPod, or at least a crude, hulking version of the same thing?

Also, what *didn’t* happen in that decade? What did we expect in 1994 that didn’t materialize?

A bit about face and object recognition software:

http://answers.google.com/answers/threadview?id=2174

Jaime’s Response

I’ve been inspired by your talk of past/present/future technology. I started thinking fifteen years out, and realized that we’re still thinking small. How about 3D imaging technology?

We’ll have to have some 3D technology now…visual for recognizing people/buildings and auditory for recognizing speaker locations, etc. However, I hadn’t been thinking about 3D camera technology, and it’s probably something that will be fully fleshed-out in another fifteen years.

Here are some first thoughts:

• There’s already (early) technology to search 3D objects based on 2D sketches: [3D http://news.yahoo.com/news?tmpl=story&u=/ap/20040416/ap_on_hi_te/search_dimensions_6 Searches]
• Intel (and others) are attempting to standardize on a universal 3D still image format, [U3D http://www.theregister.co.uk/2004/04/21/intel_u3d/]
• 3D viewing technology is already lacking, but there have been some attempts like this “Time for Space” site that use basic ideas to produce a surprisingly realistic 3D effect: [Time For http://www.well.com/user/jimg/stereo/stereo_list.html Space]

Thoughts are brewing, but I must run…

Links

So far, this is just a place to dump random links relevant to a LifeCam project.

[Face recognition software http://vasc.ri.cmu.edu/demos/faceindex/05062004/users/2236.html]

Note the photo collage and the successful face matches. It doesn’t determine *who* the face is, but it would still be handy to tell if there’s a face in a given image so it can be tagged for more complex recognition algorithms later. This could also help an autofocus system determine what “interesting” objects are in a scene.

[2 GB MMC flash card http://www.dpreview.com/news/0405/04053101pretecmmc4.asp]

Flash media is getting bigger and bigger, so the buffer on that gumstick computer will be able to hold 30 minutes of MPEG2 video (or 4 hours of MPEG4).

Related: WikiBub