Slashdot Mirror
Photosynth Demo
A couple of days ago Microsoft labs released a demo of their new Photosynth software on the web. Photosynth allows the aggregation of social picture networks (a la Flickr) into a completed image in addition to allowing a level of depth to image browsing previously unavailable. There is also a very impressive video of the demo available.
This lets you take all sorts of pictures of your room, and will automatically assemble them into a 3D environment. It will assemble your photos to look like an RPG, instead of a slideshow.
Using the example in the video...there are hundreds of online collections of people's photos of Notre Dame cathedral. Each photo is of a different part of it, from a slightly different angle.
This software takes all those different photos and assembles them into a 3D representation of Notre Dame cathedral, where you can look at any of the individual photos.
In addition, if someone identifies one of the saints in a statue on the cathedral, when you take a photo of it and your photo is added to the collection with the software, your photo will also have that saint identified--thereby enhancing the data contained in your photo.
- RG>
Hey pal, this isn't a pleasantforest, so don't waste my time with pleasantries!
This system was demoed a while ago, I think at siggraph. There are some videos on the original university of washington PhotoTourism page.. Also here's a repost of the video on youTube.
Also there's microsoft's page, which has the demo (I don't think that's new either). It seems to have some longer videos
Non-newness and marketing hype aside, this software is frickin' awesome. It lets you view and tag photos organized in a 3D environment that reflects where the photos were taken. It should be particularly useful once cameras have GPS built in.
I imagine the reason the software is still in the demo phase is because it's very difficult to take a large number of photos and reliably figure out where they were all taken from. For the demo purposes, Microsoft probably hand corrected a lot of the placements. Even so, everyone I've shown this too thinks its often (even non-slashdot readers!)
Unlike the first set of posters I managed to get over my self importance and watched a couple of seconds of BMW ads to see the actual video.
I liked the initial viewing of large quantity of hi-res images and the smooth zoom. The aggregation of many thousand flickr images of the Notre Dame (including one of a poster on a wall) into a 3-D image was fantastic.
C
This zoom-ability of the first part has a lot in common with the ideas behind Jef Raskin's The Humane Environment http://en.wikipedia.org/wiki/Archy/.
The second part, however, shows marvellous stuff. Especially if what I think he did, was search for patterns in images, and compare those for unique objects to collect a library of images of a single object.
This guy and supposedly his group shouldn't work for Microsoft in my opinion, but would perhaps feel more at home in a fundamental science laboratory. But I think my opinion on this is slightly partial.
B.
Every experiment which ends in a big bang is a good experiment.
This software is absolutely amazing, especially when you consider the programmatic side of this. People bashing this without actually watching the video AND playing with the operating demo are really missing out. You don't have to like it but at least have a reason that shows some form of intelligence. Not just "the intro was poorly done".
What I'm listening to now on Pandora...
I was one of thee engineers that worked on the first release Photosynth. It's a great team, and it was a super fun project.
I can tell you that we did not tweak any camera positions by hand. The only real "editing" we did was to eliminate pictures that just didn't correlate well, generally because they didn't have enough feature points in common with the rest of the photos. We didn't tweak any camera positions, but the camera positions (i.e. the locations of the orange camera frusta when you have frusta turned on) are a best estimate, which is subject to some error. Same goes for the projection planes.
What's great about Photosynth is that from the perspective of anyone outside the computer vision community, it appears to be magic. Enough so that lots of the blogosphere was convinced that we somehow "authored" the 3D point clouds. Nope. It's more or less an automatic (albeit somewhat prolonged) process. The hard work is done as a big preproceess, then the client consumes largely precomputed data.
It'll be cool to see Photosynth in action in BBC's upcoming How We Built Britain piece that was announced on Live Labs today.
I did a video interview about Photosynth a while back which is targeted at a non-technical audience but still might be of interest. (And I wrote the music for the original video at Live Labs.)
Microsoft didn't buy Photosynth. It bought Seadragon. The Photosynth client is indeed built on Seadragon's client, but the idea behind Photosynth (which was a joint University of Washington/Microsoft Research project called PhotoTourism) significantly predated the Seadragon acquisition, and there was a working client. When Microsoft decided to reimplement the client as a technology preview, that's when the Seadragon team and client came into the picture.
That said, Seadragon's technology is great. It's a fantastically smooth way to browse arbitrarily large images or collections of images, and it was a good acquisition indeed.
(I was on engineer on the Photosynth team.)