Archive

visualization

A color-coded map of four dimensions of the semantic space. The branches show "is-a" relationships.

A color-coded map of four dimensions of the semantic space. The branches show “is-a” relationships.

This is one of those moments in science that makes me think I’m so lucky to be alive right now. A team of scientists at UC Berkeley have found a way to map the brain’s representations of objects into a shared semantic space– a multidimensional space in which related things are nearer than unrelated ones. And there’s reason to believe this might be not just a semantic space, but the semantic space: they ran their test on five different people, and found that the first four dimensions of this semantic space were the same for all five subjects–dimensions easily labeled with ideas like moving/stationary, man-made/natural, animate/inanimate, and so on. In other words, the brain’s way of relating different objects might be something we all share at much more than a superficial level. This alone is pretty mind-blowing to me.

As if that weren’t enough, they’ve also created a very cool interactive visualization that shows how all of this plays out on the surface of an actual brain. (That page requires WebGL and a lot of memory, so if you’re reading this on an older device, you might want to just watch the video instead. Actually, you should watch the video anyway, because it’s really well done!)

Go science!

The third eye is for seeing the fnords.

If you look closely at my friend Drew Olbrich, you might notice something strange about him. He doesn’t like to talk about it, but: the part of Drew you can see is just an infinitesimal slice of a much higher-dimensional being.

To give the rest of us a sense of what it’s like to be him, Drew’s written a nifty little app (for your iPhone or iPad) called The Fourth Dimension. Give it a spin, and see if it doesn’t thicken your mind up just a little bit.

Flickr photo courtesy of purplemattfish

Flickr photo courtesy of purplemattfish

There’s been some discussion brewing among certain filmmakers about the impact of making movies that play faster than the current standard of 24 frames per second. Peter Jackson is shooting The Hobbit at 48fps, and others are reportedly experimenting with rates like 60 or even 120.

Mixed into the discussion are some really deep misconceptions about how vision and perception actually work. People are saying things like “the human eye perceives at 60fps”. This is simply not true. You can’t quantify the “frame rate” of the human eye, because perception doesn’t work that way. It’s not discrete, it’s continuous. There is, literally, no frame rate that is fast enough to match our experience of reality. All we can do, in frame-by-frame media, is to try to get closer.

The problem is that our eyes, unlike cameras, don’t stay put. They’re active, not passive. They move around in direct response to what they are seeing. If you watch an object moving past you, your eyes will rotate smoothly to match the speed of the thing you’re looking at. If what you’re looking at is real, you will perceive a sharp, clear image of that thing. But if it’s a movie made of a series of discrete frames, you will perceive a stuttering, ghosted mess. This is because, while your eyes move smoothly, each frame of what you’re watching is standing still, leaving a blurry streak across your retina as your eyes move past it, which is then replaced by another blurry streak in a slightly different spot, and so on. This vibrating effect is known as “strobing” or “judder”.

Applying camera-based effects like motion blur only makes the mess look worse. Now, your stuttering ghosted multiple image becomes a stuttering, ghosted blurry multiple image. (The emphasis on motion blur is particularly bad in VFX-heavy action movies, which is why I try to sit near the back.)


Click the image to see a demonstration of the "judder" effect. This is what your eyes actually see when you watch an object moving back and forth on a movie screen. Even with motion blur, you can see that there's a distracting sawtooth vibration to the ball that can be reduced, but not eliminated, by increasing the frame rate.

Filmmakers tend to work around this problem by using the camera itself as a surrogate for our wandering eye: tracking what’s important so that it effectively stays put (and therefore sharp) in screen space. But you can’t track everything at once, and a movie where nothing ever moves would be very dull indeed.

I am pretty sure there is no frame rate fast enough to completely solve this problem. However, the faster the frame rate, the less blurring and strobing you’ll experience as your eyes track moving objects across the screen. So I am extremely curious to see what Jackson’s Hobbit looks like at 48fps.

There’s a second problem here, which is cultural. My entire generation was raised on high quality feature films at 24 frames, and poorer-quality television (soap operas, local news) at 60 fields per second. As a result, we tend to associate the slower frame rate with quality. Commercial and TV directors caught on to this decades ago, and started shooting at 24fps to try to get the “film look”. How will we perceive a movie that’s shot at 48fps? Will it still feel “cheap” to us? And what about the next generation, raised on video games that play at much higher frame rates? What cultural baggage will they bring into the theater with them?

test9

I’m fascinated by the way our viral video has been spreading around the web. It’s a very chaotic process. One day it’ll get a few hundred views, and the next day 60,000. Blip.tv gives some very basic statistics, but it doesn’t really tell the story of how the clip’s popularity spreads from one community to another. So I’ve started using Processing to try to visualize it. Processing is really easy to learn to use, but certain commands, like bezierVertex(), have slightly less-than-intuitive arguments. The image above was the result of one of several failed attempts to understand that particular function.