Creating Multi-touch, Multi-user Games on Microsoft Surface
rm301 (sponsored) (GDC Summary)
Eric Havir (MS Surface) , Dyala Kattan-Wright (Carnegie Mellon University – D&D), Gretchen Richter De Medeiros (MS Surface), Joe Engalan (Vectorform)
My Summary:
Pretty damned useful sales pitch, actually. All my questions were answered, and I was left pretty excited by the possibilities. If this catches on. Later I talked to some industry mocap people and my excitement was reeled back in a bit (because they can track more dots with single $500 camera than this entire $10,000 5 camera system can handle).
Raw Notes:
… got here early. long line outside. perhaps i’m in the wrong session. or perhaps people are just retarded and confused.
– “playTable” was a 2001 MS experiment in putting board games and card games on a video table. NUI (natural user interface).
A system of 5 IR cameras is underneath. finger has shadow and a sharp edge on one side, which shows us direction.
– He is comparing how games started in “commercial businesses” (arcades), and that’s why they’re here at GDC: with a commercial focused product.
– talking about collaborative learning in schools. (mentions “multimouse” for multiple kids around one PC…MS isn’t involved in this are they?). …Q: is this in any schools?
SurfaceScapes (video example)
– they’re letting characters mess with their stats on a mobile device. Dungeon Master is on a laptop. Ties into MS’s “cross platform” push.
* “Everything on our laptop is done with “WPF” since we don’t need graphics.” (eh? wha? what is this “windows presentation format”?)
been working on this for 2 semesters at entertainment technology center at carnegie mellon. no sponsorship from WizardsOfTheCoast or MS? (w: really?)
Microsoft Project History
Gretchen was hired for PlayTable, to design games for children. (going after the children space). will talk about 4 foci they have. natural interaction, massively multitouch, multiplayer, object recognition.
? – my hp tx2z touchSmart supports 4 points of input. iPhone supports 5 points. How many does surface support?
More than 50 unique inputs !!!! (mentions up to 72 inputs). 40 fingers in her game example (while mentioning hungry hungry hippos). They have gotten up to 7 players crowded around a surface. 4 is ideal though.
cooooool. They should person painting with a regular paintbrush.
– they show cool critter “toy” with big eye, which is an input to table. Eye is magnifying lens for table? (… can’t find this via google)
It uses off the shelf hardware (DLP projectors, core duo processors). windows based. familiar developer tools (C#, .net, … WPF – Port Expression Blend to Surface. uses XNA for high end graphics.)
showing a 4 player tower defense. uses orientation of your finger to determine whether you can interact with other people’s pieces.
Vectorform – a company making games for Surface.
letter form – recognizes tagged tiles. forms words on side of board.
(W: could redo scrabble, where as you form words, you create the grids that others have to fit into. no need for turns. You all are creating the board each time, by placing component tiles. bending and such to make grid blocks of certain size …maybe you have to have a least a certain distance between other’s words).
(Maybe the scrabble tile scrambling/picking could also be reimagined – with tiles down you can’t see, but computer can. Maybe it could make a game out of choosing where you can pick from, to handicap)
? – How/why did Vectorform first get involved. … http://www.vectorform.com/surface/
get SDK – It has an emulator. Talk to Jim about making a hardware example. http://partner.surface.com/EN/Pages/GDC2010.aspx
visit http://www.surface.com/GDC2010 to get started with dev. (Can you access blobs? or is i just final process data?)
See about getting OSU to ask for developer program? (no, it appears dev program is closed…) MS has public sector team which reached out to certain colleges and universities and seeded. not really seeding any more. sigh.
see who to contact anyway? Maybe OSU is already involved?
You really have to fine tune it to detect contact plane gestures – or gestures up in the air above surface.
it’s all vision based, not capacitive. should be able to handle 52 touches (from their testing). (I have this random note: “had a bar you could take with you” – wha?)
5 cameras. combine the image. specialized hardware pre-processes image. 8 minute video goes into the processor and how they send the images across. ($15K for hardware)
They evaded question of camera resolution by focusing on whether surface could track real dice. it is not at the resolution where it could read a bar code. 6 sided dice is ok. 20 sided die gets nuts (tiny faces. not always arranged the same. different shape and sizes). could probably read numbers on a credit card.
… find myself wondering if you could mount 4 extendable antennae along each edge. so they could extend into the middle and move pieces. or create tactile divides.
