dbs@speech2.cs.cmu.edu (David Sanner) (11/07/90)
hi again, ( maybe this belongs on sci.electronics, but i thought i'd post it here first) some time ago, a couple folks posted some messages on their steps towards 'homebrew' virtual reality systems. unfortunately it seems to have stalled, for the moment. i'm interested in starting a thread in this group for people who either 1) have developed their own systems, or 2) are interested in developing them. i fall into the latter category. i have a lot of ideas, and i'm currently working on the software end... i need help, though, putting together the hardware end: sure it won't be fast. i don't care. sure it will have lousy resolution. i don't care. sure it will be buggy. i don't care. i think many issues in VR can be explored with a "cheap", do-it-yourself "virtstation" (or, if you prefer, "verminals"!) as i mentioned, i'm primarily interested in what people are doing, or are thinking of doing, with the hardware end of things.. ( i.e. all those great input/feedback devices). ultimately, the system should allow a user to be able to walk, sit turn head, etc. with the appropriate feedback. the system should also be able to be built by anybody with a soldering iron, determination, and a meager amount of cash. that is, perhaps, way too ambitious at the moment... but it's something to shoot for (remember, it doesn't have to be perfect!) as for the software 'visualizations', i'm keeping my pie-in-the-sky ideas in check, at the moment. just some 3d polygons or lines, updated in (mabe) real-time response. then i'll go from there. let me start it off: [to get the greatest possible use of available equipment, lets assume i have a room which can be more or less used for VR alone. this might allow us to consider less flexible systems (like, wall mounted tracking systems, etc.)] i have an amiga 1000. ok, it's anchient, but it still works, and i've hopped it up to 2.5megs and a 20MB harddrive. i'm just pounding out the code for 3D at the moment. the code should be done soon (and the check is in the mail...) and i'm looking toward the next step. for visual feedback, i'd like to do the following: take 2 2.7 inch b&w sony watchmans (or similar device), and rip them apart. mount the ntsc circutry on a backpack/chestpack/tether. mount the screens on an old motorcycle helmet at the appropriate angles. for sound, insert some thin earpieces into the helmet, and connect to the amiga. for a locating device, well... some people have mentioned an electronic compass (like the 'shack has) - hopefully, using "flux-gate" technology, the influence of stray magnetic fields (like those in a normal house, or around elec. circuts) can be decreased (any one care to mention how this works?). this can be mounted on the helmet, and the data fed into the computer. when (if!) the data-glove rs232 version comes out, mabe it can be used in a similar manner (the sonar method the power-glove uses seems so limited...) now, several things rule this choice: black and white watchmans are cheaper than color. my amiga outputs a TV signal. the helmet is no longer useable on the road. the amiga has stereo sound outputs. the electronic compass is cheap, and with schematics, you might be able to build your own. provided, of course, that it can produce stable output in the presence of normal magnetic fields produced in a normal house (and circuts). notice that all these things spell 'cheap', 'cause i'm poor. as i mentioned, i'd like to keep the approach as cheap as possible (and that would be an accomplishment in itself !!) major flaws are apparent in this system. for instance, i'd need 2 amigas, or switching/buffering circuts to use the ntsc outputs, to get 2 seperate pictures on the watchmans. but i think the visual part is doable (since some folks out there have already done similar things). the elec. compass may work ok in a level plane (parallel with the ground), but what happens when you incline your head forward/backward... and does it have a settle time before it's output is valid? how do you detect someone lowering or raising themselves in altitude? etc... naturally, there are lots of alternatives to the design. since i've assumed i have a whole room at my disposal (well, a basement), i could mount sensors on the wall, and have some kind of receptor on the helmet. or, since the helmet encases the whole head, some kind of mechanical/electronic marriage may provide adequate response (e.g. a linkage assembly attached to the back of the helmet, and also to your back. motions of the helmet are translated as relative motion to your (assumed straight) back). anyway, i'm hoping the rest of you will pick out problems or suggest solutions of your own, and hence a new thread of discussion... thanks, ... dave sanner dbs@speech2.cs.cmu.edu
warren@debra.doc.ca (Warren Baird) (11/07/90)
In article <10709@milton.u.washington.edu> dbs@speech2.cs.cmu.edu (David Sanner) writes: > > for visual feedback, i'd like to do the following: take 2 2.7 inch b&w > sony watchmans (or similar device), and rip them apart. mount the ntsc > circutry on a backpack/chestpack/tether. mount the screens on an old > motorcycle helmet at the appropriate angles. for sound, insert some > thin earpieces into the helmet, and connect to the amiga. for a locating > device ... I'm interested in knowing how important this helmet idea is. Is it necessary to have full 360 degree visual effects for a virtual reality? Would a system based around a large monitor not allow you basically the same effect? Say with a three dimensional controlling device in one hand and a powerglove on the other, to control both movement and interaction with the environment. You can generate a fairly good three dimensional image on a flat screen, although it isn't an encompassing as I assume a helmet image would be. > > ... dave sanner > dbs@speech2.cs.cmu.edu -- Warren Baird | warren@dgbt.doc.ca ...utzoo!dciem!nrcaer!dgbt!warren Doing a Co-op term at Communications Canada, Ottawa
frerichs@ux1.cso.uiuc.edu (11/08/90)
your basic ideas are sound... but I think you are missing some basic improvements that would make your life easier, and make your VR much more realistic... One is the dual B&W watchman setup... do you really want 2 CRTs that close to your eyes for that long...? Radio Shack or any store that sells casio products can provide you with 2" color LCD TVs ripe for tearing into and directly piping in NTSC... at about the same price or less than the watchmen... ... another is making the VPL mistake of depending on the host computer microprocessor to do the graphics... this drives the cost of the VR setup exponentially... build an external interface that can accept object input from the host, process the object into two views, and then output to the LCDs... ... there are more, but I don't want to give away all of my secrets... good luck... -dfRERICHS
pepke@SCRI1.SCRI.FSU.EDU (Eric Pepke) (11/09/90)
It's probably not a good idea to use the Sony B&W Watchmans, because (1) you don't want to have KV in front of your face, (2) most of the tube would have to hang down below the eyes or stick up above them, and (3) binocularity would be dependent on the amount and linearity of scan, which would be dependent on a voltage level that might fluctuate. I looked at a few 2" and 2.5" and 2.7" color LCD displays, but I don't think they will be sufficient either. They don't seem to be big enough to provide enough peripheral vision with a reasonable power lens. Instead, I think one of the 4" color LCD's would be good. I haven't been able to find a source for the screens themselves. Everybody I ask just says, "they're Japanese." (Sigh. Doesn't anybody in this country love the smell of burning phenolic any more?) 4" color LCD TV's are prohibitively expensive, but there is a portable video game system by Atari called the Lynx system which sells for about $170 and contains a 4" color LCD. I'm still working on the PowerGlove so I haven't had time to follow up on it and/or rip one apart, but if it were possible to find out more about the device, one might even enlist the use of the game processor. From the demos I've seen, it seemed that it could do some pretty zippy scaled bitblts. Maybe this would be useful, maybe not. I'm currently thinking about two possible setups. One would be made from a Craftsman set of welders' goggles. They have 2" lenses held in by screw caps which could be replaced by 2" magnifying glass lenses, and the rest could be mounted on the front. The other would be made of a couple of 2" by 4" rectangular Acrylic magnifying glass lenses which would be abutted and cemented with a nose notch cut out and worn directly on the face. In this configuration the center of the lens would be outside the eyes, which is actually desirable, as it makes any image in front of each lens appear more toward the center. The rest of the box could be made by cementing 1/8" sheet acrylic, formed and cemented into a rectangular frustrum, to the lenses. A second magnifying stage, perhaps a Fresnel lens, would go between the eye lenses and the screens. I think this would give just about the right amount of magnification. It would probably be pretty lightweight, and maybe a single heat-formed strip of acrylic over the head and a strap would be enough to keep it on. Eric Pepke INTERNET: pepke@gw.scri.fsu.edu Supercomputer Computations Research Institute MFENET: pepke@fsu Florida State University SPAN: scri::pepke Tallahassee, FL 32306-4052 BITNET: pepke@fsu Disclaimer: My employers seldom even LISTEN to my opinions. Meta-disclaimer: Any society that needs disclaimers has too many lawyers.
rnm@uunet.UU.NET (Robert Marsanyi) (11/14/90)
regarding magnetic compass v. ultrasonics: given the information that has been showing up from that intrepid person reverse engineering the Power Glove regarding the resolution of the ultrasonics circuitry included with the glove, how come the preference in your posting for magnetic position sensing? Just curious. And, yay, I too am poking around with using an Amiga for breadboarding low-level approaches to this technology - it seems like we could all save ourselves some development effort here by using the same platform.