I’ve written a script for the WebTV class that is meant for distribution over the Internet (hence the WebTV label). It will be encoded in a streaming format and in all likelihood will end up at a quarter the resolution it was shot at and compressed to the edge of watchability.
To get quality streaming video requires some serious broadband access. Using the latest encoding schemes from Microsoft or DivX, you can squeeze a DV film into about one megabit per second (mbps). Dial up access to the Internet is about 0.05 mbps so there ain’t no way that’s happening short of dark magics… and the last time I tried that I barely got away with my soul (and there’s a certain demon that still owes me a hundred bucks- you listening Aeshma?).
Back on topic, where was I… Aeshma.. black magic.. dial up.. oh yeah DV quality video at one mbps. If you have a good broadband DSL, cable or satellite connection to the Internet you can actually use this. There are two problems here. One is that pretty much every connection to the Internet is a bit flaky and you’re lucky if you can sustain this high throughput for the duration of a one hour show. Two, there are still a lot of people in the .25 to 1 mbps range who would be out of luck. And thir- okay, there are three problems here. Thirdly is the fact that DV is the bottom rung of the quality ladder.
HDTVs two most used resolutions are 1280×720 which needs at least 2 mbps, and 1920×1080 which needs at least 3.5 mbps. It is going to be a few years yet before many people can get reliable 3.5 mbps Internet access.
This got me thinking on ways to maximize the quality of streaming video for the huge number of people who have at least .25 mbps access. Since I’m also looking into things like compositing, 3D animation, game modding and machinima, I hit on the idea of “greenstreaming”. This is an adaptation of chromakeying/greenscreening for the optimization of video delivery.
The upcoming release of Doom III shows what kind of quality we can expect from realtime animation on the computer. What I think should be developed off of such a game engine is a movie creation and viewing platform. Not just machinima, but a combination of live action composited into machinima. If you were to build a virtual set and have the actors work in front of a green screen, you could erase all the data from the video except for the actors themselves and cut down on the bandwidth needed by roughly 75% (back of the napkin calculator used here).
The viewer would have the entire set on their computer, so the streaming signal would consist of the actors, any unique 3D model elements, lighting information such as time of day and if the lights are turned on – and camera motion.
Camera motion would be a tricky element. You would need to track the motion of the camera on the real set and send that data to the viewing program so that the virtual camera can move through the virtual set in sync with the live action camera.
A DV quality stream could actually be sent at around .25 mbps while the two HDTV resolutions might get by with .5 mbps and .875 mbps respectively. This all fits into an access envelope that today encompasses more than 40 million people in North America alone. That seems like a good sized market to me.
There are drawbacks of course. The viewer would have to have the program and the virtual set on their computer and that is a barrier to entry. For it to be a rich environment, you’d probably end up with at least 500 MB of data to download. The highest quality game engine to base the viewer program on is probably the one that Id Software has developed for Doom III and the license for that would probably be around half a million dollars so the viewing software would have to be sold unless it was subsidized.
Along with the wicked game engine comes wicked hardware requirements… but that is only by todays standards. A year from now, todays fast game rigs will be middle of the road systems and the year after that they will be entry level stuff. As well, the game engines degrade gracefully depending on your hardware. If you have a high end VoodooPC, you are going to get the most out of the video but if you are using a yeoman Dell system you will get less detail, simpler lighting and more jagged graphics… but you’d still get the picture.
(Hey Carmack, how about tweaking the next engine for a more cinematic work? An optimized setting that works at 24 fps with a modicum of motion blur, depth of field blur and more control over the “camera” would lead to some sweet machinima even if my greenstreaming idea isn’t embraced.)
The benefits for all parties involved are clear though. The end viewer gets an image of much higher quality than they could otherwise get. The makers of the game engine could have another revenue stream. The companies that host and stream the video could deliver four times as much content over their existing infrastructure. The content providers wouldn’t have to put up with the severe degradation of their product just to get it out there.
And is it my imagination or are my blog entries looking more like articles? I think that may be the way I’m going with this but I should probably put a few more interstitial postings that are short and sweet so that the blog isn’t quite so… heavy.