harry@harlqn.co.uk (Harry Fearnhamm) (04/09/91)
(1) GoggleVox My apologies to all and sundry interested in the GoggleVox headset mentioned a couple of weeks ago; I gave the wrong address and have only now been able to post owing to severe mailer problems. The current situation seems to be that the inventor is clinching major deals for the device AS A PERSONAL TEEVEE SYSTEM; his interest in VR will be limited for the time being simply because there is not a mass market for it yet, and consequently he has not produced a spec relevant to VR applications. I managed to talk to him, but for now he would prefer to go through his agent, Ms Samantha Turner, who can be contacted on +44 (0)81 202 0976. I have been promised some kind of blurb but it is "in the post", so I will send it on if and when it arrives. (2) A comment on ongoing data visualisation. I'm surprised that noone seems to have mentioned PARC's latest stuff, (unless it got lost in the post ;-) as feartured in a recent Byte (Dec/Jan?) - they seem to be making good use of the notional third dimension within convetional displays, with depth and shadow cues to improve the user's understanding of what they are viewing. I think they are also trying to use sound to further augment this. (3) Object update accountability. We've been through the concepts of what ought to be happening at the object level, and there seem to be two obvious camps - firstly when you render some raw data (video, sound, remote sensors, uninterpreted database...) into a virtual space - and we must be prepared to do this - and secondly, when everyone behaves themselves and communicates via whatever standard(s) are decided upon in the coming years. Now in the former case, you may need considerable processing power locally to achieve your ends, but that may be acceptable if you aren't involved in other sessions (or maybe the processing power can be assumed to be available), but the other one still seems to present some problems. Let us suppose we have access to high speed networks, and we are all very careful about how we announce object change info. We all have arbitrarily large databases to hand giving us the ability to have symbolic descriptions in the objects which can be looked up in the database and used in the rendering, thus saving on the amount of information transmitted (tho' you must be able to (a) override it and (b) learn new desciptions as painlessly as possible). But there will always be occasions when you have no choice - you *have* to have a huge update, such as when you enter a VR session for the first time. What happens? You might chose to wait for all the zillions of objects to be described to you, or it might be described to you with a level of detail proportional to a combination of your point of focus and the distance you are from the objects. It seems to me that there must be a well defined accountability, but I'm not sure where that should lie - if several individuals enter a VR, that VR must exist on some machine, which is responsible for telling the users of its changes, just as you should tell that machine of yours; but should it be this VR machine's duty to tell you about other individuals present, or should any changes from these individuals go directly to you (and other users present) in some form of selective broadcasting? I'm inclined to think that the former would be more efficient and more manageable. Just some idle ramblings - ignore them at your leisure. -- Harry Fearnhamm, ,---.'\ EMAIL: loki@harlqn.co.uk Harlequin Ltd, (, /@ )/ ...!ukc!cam-cl!harlqn!loki Barrington Hall, /( _/ ') VOX: +44 (0)223 872522 Barrington, \,`---' FAX: +44 (0)223 872519 Cambridgeshire, DISCLAIMER: Nothing is True. ENGLAND. Everything is Permitted.
piggy@gargoyle.uchicago.edu (La Monte Yarroll) (04/11/91)
In <1991Apr9.203748.7430@milton.u.washington.edu> harry@harlqn.co.uk (Harry Fear nhamm) >(3) Object update accountability. ... >(b) learn new desciptions as painlessly as possible). But there will >always be occasions when you have no choice - you *have* to have a >huge update, such as when you enter a VR session for the first time. >What happens? You might chose to wait for all the zillions of objects >to be described to you, or it might be described to you with a level >of detail proportional to a combination of your point of focus and the >distance you are from the objects. Usage note: I use quotes (") to mark terms which I'm intensionally using as vague technical terms. This is a problem I've been considering lately. The model I've been thinking about comes from a graphics transmission technique I read about some years ago (in DDJ?). The idea of the technique is that you can quickly get a low resolution image on your screen. The image resolution improves over time as you receive more data. This way you can decide early on whether or not you are interested in the particular image, and about quickly if you decide you don't want it. The algorithm for generating this data stream is approximately as follows: Divide your image into a small number of symetric regions (say, 4 squares). Decide on an "average" colour for each region. Transmit these data. Recurse breadthwise. Subsequent colours are coded as changes from the "average" colour already placed in the given region. I've been thinking about 3D generalizations of this algorithm, which seems to be what we want for VR. The obvious extension of dividing space into cubes and picking an "average" colour, seems inadequate to me--it seems that you'll get stuck with a bunch of opaque blobs. Is it perhaps possible to come up with an "average" shape for any spacial region--something like approximating a sphere with an icosahedron? >It seems to me that there must be a well defined accountability, but >I'm not sure where that should lie - if several individuals enter a >VR, that VR must exist on some machine, which is responsible for >telling the users of its changes, just as you should tell that machine >of yours; but should it be this VR machine's duty to tell you about >other individuals present, or should any changes from these >individuals go directly to you (and other users present) in some form >of selective broadcasting? I'm inclined to think that the former >would be more efficient and more manageable. If you do indeed have a "VR manager" in the middle, would it not make sense for this manager to cache info on it's participants? >Just some idle ramblings - ignore them at your leisure. Just some more idle ramblings - ignore them at your leisure. :-) >-- > Harry Fearnhamm, ,---.'\ EMAIL: loki@harlqn.co.uk > Harlequin Ltd, (, /@ )/ ...!ukc!cam-cl!harlqn!loki > Barrington Hall, /( _/ ') VOX: +44 (0)223 872522 > Barrington, \,`---' FAX: +44 (0)223 872519 > Cambridgeshire, DISCLAIMER: Nothing is True. > ENGLAND. Everything is Permitted. La Monte H. Yarroll Preferred: piggy@gargoyle.uchicago.edu Work: piggy@mwc.com Home: piggy@baqaqi.chi.il.us often: postmaster@clout.chi.il.us