Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!hellgate.utah.edu!caen!sdd.hp.com!spool.mu.edu!think.com!mintaka!mit-eddie!uw-beaver!milton!pete_leaback@cix.compulink.co.uk From: pete_leaback@cix.compulink.co.uk (Peter Leaback) Newsgroups: sci.virtual-worlds Subject: Re: We need a new language Message-ID: <15594@milton.u.washington.edu> Date: 31 Jan 91 02:34:00 GMT Sender: hlab@milton.u.washington.edu Lines: 51 Approved: cyberoid@milton.u.washington.edu I agree with many of your points. A standard language that deals with object communication and interaction is essential. But.. You say VR is here, it is only *JUST* here. At the moment, the programming of VR's are coupled very tightly to the hardware because of the crippling restrictions of the hardware. A VR system has to display a frame at 50 or 60 Hz or else one tends to get sick. If one standardises the object destription and rendering, each machine that has implemented the language is required to render images at roughly the same *speed*. The examples you gave of functions that have been pulled away from the hardware has only come about because the hardware has *allowed* it. Machines are fast/large enough for the OS to be portable. Laser printers have become standard because the resolution is similar and speed is not essential. VR is not at the point where it can peal off completely from the hardware. I suggest that a standard language should be layered in its definition. The highest would be such that it can be realistically implemented on most hardware.As our hardware progresses, subsequent levels would be defined. If a *complete* VR language is defined today, it would either be too restrictive or impossible to implement. An area of computing has a critical mass before it is useful to start standardising. In a few years, the non trivial layers of a VR can start to be standardised. E.G. a scene is defined with trees, a path, some rocks and a light source. A low end VR machine would implement the scene with flat faced polygons, no light source, no shadows and at low resolution. A high end machine would plot a much more realistic scene, but basicly the same as the low end machine. My point is that, at the moment, many machines won't be able to render that many objects using *ANY* method! So what would the program do ? Miss out 50 trees ? Regards, Peter Leaback.