Fotowoosh and Photosynth: 2D photos and 3D models

Over on my Computerworld blog, I've taken a brief look at two software tools which can model 3D environments based on ordinary photographs. Judging by the demos that I viewed, the more visually impressive tool is Fotowoosh. Microsoft's Photosynth seems to be an interesting concept, but the end product seems more aimed at creating a geospatial model that contains 2D imagery, rather than true 3D visuals.

When such technologies become available to developers and/or are integrated with other 3D modeling tools, watch out! Here are some possibilities that I outlined on my Computerworld blog:

... If this technology (or one like it) can be integrated with other 3D modelling tools, the impact on a number of industries will be profound. Imagine a FPS game based on actual street scenes, a la Escape From New York. Or a virtual world based on real photographs from a specific time or place. Or a Web-accessible 3D real-estate demo created from standard digital photographs and a floor plan.

Techcrunch reports additional details about Fotowoosh, including news about creator Derek Hoiem and the VRML technology which is used to create the 3D models.

New media projects at Computerworld: Internet video news and Tech Dispenser

One of the best aspects of my current job at Computerworld is that I am often involved in many new media projects for the Computerworld website. I was hired in early 2005 to get Computerworld's blogging area and editorial webcasts off the ground, but since then I have also helped develop Computerworld's podcasts as well as user-generated/community content offerings such as Shark Bait. Currently, I am involved in two new editorial initiatives: Web video programming and Tech Dispenser.

The event video is a sponsored daily recap and interview show from one of Computerworld's largest events, Storage Networking World. The event ended this morning, and we created three episodes in all. The second episode will give you a good idea of what the program is about in terms of the format and content:

SNW In Focus, day 2: Mini SANs and the UK's fastest supercomputer

This is not the first time I've worked with video -- my first job in journalism, back in the mid-1990s, was working as a newswriter and narrator for an English-language TV program at the China Television Company (CTV, 中國電視公司 or 中視 in Chinese). Then, as now, producing professional-quality video programming was an extremely labor-intensive effort, requiring a great deal of teamwork and coordination. I know there is a big movement toward vlogging and user-submitted video content a la YouTube, but anyone interested in sponsoring or creating professional, TV-quality content should know that it's not as easy as it looks. My colleague Lucas Mearian and I were the faces of Computerworld Events: SNW In Focus, but behind the scenes there was a video production crew working full time, not to mention additional help from editors and online production staff back at Computerworld HQ in Framingham. Despite all of the work required, we consider the program a success, and look forward to doing similar editorial video programs in the future.

The second new initiative that I am involved with at Computerworld is Tech Dispenser. This is an editor-driven blogs aggregator that I think could be a useful alternative to traditional, algorithm-driven aggregators like Megite.

Wait a second, you say: Aren't you the same guy whose research interests include advanced data-mining technologies and computer content analysis? What have you got against algorithms and existing aggregators that highlight interesting news and blog content in the giant, semi-structured database that is the Web?

My answer to that is best summed up on a recent post on my Computerworld blog:

The aggregators are extremely efficient in finding and highlighting news or topics of discussion, but there is a flaw that all share: An inability to identify quality content. Computers are good at counting the number of links pointing to a specific blog post, or measuring the number of topical keywords in a news article. But they are incapable of spotting a scoop, let alone an elegant analysis of a technology trend. Hence, we see lots of highlighted articles and blog posts on the aggregator sites that are simply repeating what someone else has already said, or weak writing samples that are a waste of readers' time. A few sites use deceptive SEO techniques and other weaknesses in the algorithms to manipulate the aggregators to get their articles or posts to the top positions, and on several occasions I have seen suspected astroturfing campaigns highlighted on the blog aggregators.

Tech Dispenser hasn't launched yet, but I am interested in seeing how the tech blog community reacts to the idea. Besides driving traffic to blogs that participate in the Tech Dispenser blog network, the site also includes a revenue share model that should appeal to many tech blog writers.

3D Scanners and Starch Replicas

One of my childhood memories is watching a scene from the first season of the Japanese anime TV series Star Blazers, a dubbed version of which was shown on Boston-area UHF television in the mid-1980s. In it, the ship's science officer shows off a device which can replicate any small- or medium-sized object -- I think the example used was a scooter.

It seemed fantastic at the time, but a preliminary version of the same technology exists now. The New York Times reports that relatively low-cost 3D scanners and printers are available to individuals and small businesses. The scanners can scan a small object in three dimensions, using lasers and rotating turntable. The "printer" can take the data from the scan, and recreate the object, using starches or resins. The NYT article, by Peter Wayner, explains:

The world is just beginning to grapple with the implications of this relatively low-cost duplicating method, often called rapid prototyping. Hearing aid companies, for instance, are producing some custom-fitted ear pieces from scanned molds of patients. Custom car companies produce new parts for classic cars or modified parts for hot rods. Consumer product makers create fully functional designs before committing themselves to big production runs.

The scanners cost between $2,500 and $40,000, and the printers cost $40,000 and up, says the article. Individual copies of an object using printers made by Z Corporation cost $70, ranging upwards of $700 for larger copies. Printers made by Dimension Printing use plastic in their replicas, costing about $50 per cubic inch, says the article.

While prototype manufacturing appears to be the dominant market for these types of services, there is a potential market for people's own belongings. Imagine being able to scan that rare Hummel figurine you've treasured for all of those years, and make a nearly exact copy of it to keep in your office.

Unfortunately, there may be a hitch with such plans, notes the article:

The legal landscape, though, may not be ready for the Napsterization of three-dimensional things. Most of the cute, small tchotchkes in my house that fit on the turntable of the NextEngine scanner I tested are copyrighted. Zapping up a new version might run afoul of the same laws being used to fight the piracy of songs.

Scans also require a powerful system for the home user who want to use the 3D scanners. The article says the $2,500 NextEngine scanner requires a fast PC with a recent video card and 2 GB of RAM and lots of storage -- an individual scan typically takes up 100 MB of space.

Also, the scanners can make near-perfect 3D models of objects, the functionality of those objects -- if there is any -- cannot be replicated. So while it would be possible to scan my son's toy robot, the resulting replica would be a statue, without any wind-up movements. This is unlike the Star Blazers example, in which the replica scooter was able to work as soon as it left the chamber.