Here’s a snapshot of me (in England) and Nisakiman (in Greece) and Grandad (in Ireland) in Emily’s appear-in (Boston?) bar yesterday:
It worked fine with two people. But it wasn’t very successful with three, because Nisakiman sounded like he was speaking from the bottom of a swimming pool, with his words occasionally bubbling up to the surface a minute or two after he’d spoken them. But it was a bit of a laugh all the same.
I think that Nisakiman’s problem was that he has an old, rather slow computer. He and Grandad don’t agree with me about this. But in the peer-to-peer free version of appear.in we were all sending video to each other and receiving video from each other. So with N people each computer has to process 2.(N – 1) video streams, I reckon. While in a hosted server system, the central hub would have to process 2.N video streams, and each peer would have to process only 2 video streams. So I reckon that if I get a paid room with central host computer, Nisakiman’s slow computer will most likely be able to handle two video streams. In fact, we know that it can. The limiting factor with a central server is probably the speed of the server, and appear.in probably aren’t using a supercomputer.
That set me thinking about a few other possibilities.
Emily’s bar has rather disappeared in the image above. I wondered whether it would be possible to get rid of the home backgrounds behind us, and put us all in a proper bar setting. In any video stream only a few pixels are changing at any one point in time, as people move. The backgrounds shouldn’t be changing too much. I wondered if you might get rid of the backgrounds simply by only sending only the pixels that had changed colour. These would then be stitched together and a background and foreground added. So me and Nisakiman and Grandad could all be put in a bar context, sitting at a bar with beers and ashtrays in front of us, and stuff going on in the background, a bit like this:
A further thought was that we might all be cartoonized:
And the background and foreground might have people wandering around. Perhaps you could have a background hum of conversation. And music playing on the juke box.
I realised this morning that I almost know how to do all this. My orbital simulation model does 3D in a fairly simple sort of way. So I could create a simple bar scene (the above is the bartender’s point of view), and populate it with cut-out cartoon people. Some of these people would be cartoonized real people like above. Others would be invented people who would be sitting in the background drinking beer and smoking cigarettes. The only thing I don’t know how to do is all the internet stuff needed for the servers. In principle, a powerful host computer could be taking the live video streams from the participants, scaling and cartoonizing them, and stripping out the backgrounds, and placing them all inside a busy virtual pub.
If the whole thing was taken a step further, all the real people could have 3D replicas of themselves rather than 2D cut-out cartoons. There would be multiple points of views. In fact everybody would have their own point of view.
The possibilities are endless. Appear-in is just a simple first stage, which bundles together a set of unprocessed video streams.
I’m beginning to think about how to strip out backgrounds (and cartoonize people). This might become my next computing project.