Wow! I mean, really?!? Words cannot express this amazing (new-ish) technology. A picture is worth 1,000 words, and this technology brings together thousands of pictures in new and innovative ways. Perhaps we’re getting close to a proper expression now. I’m not going to attempt to rationalize or dissect this concept because I really don’t know a lot about what goes into it. You’ve just got to go see it for yourself.
Microsoft Labs: Photosynth Preview
Just in case you were too amazed to properly put together what you just saw, here’s my summary. SeaDragon developed an imaging solution (algorithm, I suppose) that regards bit depth and pixel count as only contextually relative. To an extent, your computer does this today. It’s just not as efficient as what SeaDragon was able to do. If your computer is asked to display 100 images in a grid and your computer’s monitor is capable of displaying 786,432 pixels (1024×768), then the SeaDragon algorithms resize the images to 7,800 pixels each (100×78) – don’t argue, just follow the logic of my simple example. Graphics programs such as Photoshop do this with a choice as to how you would like to interpolate the images (Lanczos, B-spline, Triangle/tent, nearest neighbor). I haven’t heard what math is used to do it in SeaDragon’s technology, but it must be something fast (not necessarily accurate). The implementation allows you to seamlessly zoom in and out of an entire image collection without delay. Microsoft demonstrated the technology as a new way to media market. The example involved showing a newspaper with vector text on it and images flowed – just as you would expect. However, the interesting bit was where they zoomed into an area of the paper to bring a micro-fiche like add to the foreground with distinct detail in the images and text (of course). It really merges the world of raster and vector imaging solutions. So here’s where Microsoft steps in and, correct me if my timing is off, bridges the technology with some Microsoft magic to create an imaging technique with spatial recognition. The example you see at the Photosynth link is a collection of Flickr photos of a specific cathedral. They’ve spatially mapped all the pictures into a model of the cathedral. This combined 2D/3D world now allows you to move around within the 3D structure while using a variety of images to give you the detail. As you move the mouse around, it highlights spatially plotted images that were recognized to fit. Now that you know what’s going on, go back and try it again!