|
|
The goal: Quick mapping for mobile live stage setup Hi folks, I'm coming to you because I don't even know where to start. I am realizing that I need to start hanging out with people much smarter than me because I know what I want to do is possible, but I don't know anyone that can tell me where to begin. Long story short, I lead a Zimbabwean style marimba band that mixes that music with Electronica using DIY midi triggers and live video projection. I build all the instruments myself. I really want to take things to the next level with much more seamless video mapping integration. Seeing the work of Trentemoeller's crew , Skrillix's crew, and this: http://www.botndolly.com/box and all the other amazing stuff is very inspiring. What I want to do: Have several shapes built from cardboard or plywood, etc. I'd love for those shapes to not be only square or rectangle. The shapes would always generally be in the same place on the stage every day, but there will be small shifts from day to day as the band is in different venues. My thought is that we'd have sensors on the corners of each shape and have those sensors be able to be read by the computer in a way that the sensors have a fixed number-IE one shape has the corners where each sensor is always 1, 2,, 3, and 4, etc.... Then have a method of scanning the sensors in space to see where they are on the stage (IR sensors?). Then you could have subtle or even massive scenery changes and just automatically map to the changes. The above video shows this obviously already exists. But I'm a guy that know some PHP, and has build some guitar amps and synths, etc....I don't know how to make the leap from my current knowledge to what I want to do. So I need to either team up with some awesome people that already know how to do this and are willing to share, or learn myself, or a combination. Any advice on where to start? |
Have a look, also, at WiiProject and Johnny Lee Youtube channel (https://www.youtube.com/watch?v=nhSR_6-Y5Kg). But easier would be image recognition, a) if all the shapes are different, the program can recognize that, or more simple, b) you just adapt the projection area to their actual siluetes (if it has not to be a specific mapping) = easier said: a) any shape can have its unique ID merely by recognizing its shape, so you can avoid number, and b) if for example you have faces on every shape, it would be a specific mapping, but if you have textures flowing, you can just isolate their siluette and tell the program to project only on that siluete (you can easily separe a sliuete of something of its back projection infrared light on the back, the camera would see that but the eye not). I guess just projecting the shapes won't be enough, you want them tridimensional in space, isn't it?