h a l f b a k e r ynon-lame halfbakery tagline
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
Before you is a glass box: 2m wide, 3m tall, .5m deep. This box has most of the air removed by means of a vacuum pump at the base. Inside, affixed to a hinge at the top of the box is a white sheet, slightly less wide and a bit taller than the box. Attached to the bottom of the sheet is a machine that
rapidly vibrates forward and backward (traverses the .5m dimension). The result is a sheet that is flapping very quickly.
Behind you is a projector. This projector is similar to modern projectors that plug in to computers, but with a faster refresh rate. The computer supplying image data to the projector is also controlling the vibration machine, and is programmed with a sheet flapping model that allows it to know precisely where every point of the sheet is at any given moment. The computer uses this information to know when to direct the projector to color a point on the sheet at any given time. For instance, if trying to model a 1 pixel diameter red rod in the center of the sheet that is perfectly aligned in the .5m dimension, it will tell the projector to always turn that pixel red. If modeling a point closest to you in the .5m dimension, it will instruct the projector to light that pixel once per flap.
The result is a 3-D display that can be viewed from any angle.
I hope I've done a competent job explaining this highly visual idea. Let me know your questions and I'll add to this description.
A visualization
http://bz.pair.com/fun/flapSheet.html
24 Feb 03 | My shot at it. I have two volumetric lights* projected on the flapping sheet showing that, depending on where the sheet is in its flapping cycle, the disc of light (ostensibly the "pixel") would be shown at different actual depths along the axis of the flap. [21Kb image]
*Volumetric only to help show where the light is in space. A normal projection as described wouldn't show up as a laser beam as I have it here. [bristolz, Oct 04 2004, last modified Oct 21 2004]
(?) Conceptually sort of kind of similar
http://www.felix3d.com/paper_pw_02.pdf
Helical version (PDF file) [half, Oct 04 2004, last modified Oct 21 2004]
DLP
http://www.dlp.com/...imate_modulator.asp
One way to get the refresh rate we need. [Worldgineer, Oct 04 2004, last modified Oct 21 2004]
(?) Actuality Systems
http://www.actuality-systems.com/
Their 3D viewer sounds similar to this. [st3f, Oct 04 2004, last modified Oct 21 2004]
[link]
|
| |
Hey autoboner, please let me know what problem you see with this. |
|
| |
//a 3-D display that can be viewed from any angle//
Wouldn't you lose the 3-D effect unless looking at it straight-on (in line with the projector)? Or at least it would develop a skew distortion until you're viewing it from the side, when you'd see no image.
Minor complaint, though. I like it! Have a rapidly vibrating croissant. |
|
| |
On the contrary, every point of light would exist in three dimensions so it wouldn't be skewed no matter what your perspective. I guess if you were standing exactly beside it the length of sheet between you and the lit point would block your view. |
|
| |
That's not true, if you viewed it from the, say, the right side, the pixel illuminated from the front would only be visible from the time the sheet was in front of you to the end of its excursion to the right (the deepest point on the .5m axis). You would only see roughly half of the line as it was drawn in space, provided the sheet was opaque and I am assuming it is opaque from its "white" description. |
|
| |
I think I sort of see how this might work. The effect of the "flapping" would be to set up a sinusoidal wave in the sheet, moving vertically. The peak and trough of the wave would be at a known position (the sheet flapping model). So, when a given point on a 2 dimensional view of the sheet is supposed to be "close" to you then that point will be illuminated when the peak of the wave is at that location. If the point is supposed to be far from you then that point would be illuminated when the trough of the wave is aligned vertically with that point. |
|
| |
(Please, pardon my rusty physics) Never having studied the physics of a sheet flapping in a vacuum, I am left with a couple of questions. How will the energy be absorbed once the wave hits the upper attachment point of the sheet? In order for the wave to travel upward, as opposed to being a standing wave, I think there would be a problem forcing the top edge to be a node or zero amplitude point. Seem like the energy would get reflected back down the sheet and mess with the waves trying to travel up the sheet. |
|
| |
Seems the period of the wave would have to be fairly short to get any kind of decent scan rate out of this. Short period = high frequency. What sort of mechanism do you envision being able to traverse the .5m peak-to-peak distance of the wave? Seems like it would have to be extremely low mass to move that distance at high frequency. |
|
| |
My art skills being what they are(n't), I'm wondering if the viewing angle might have more of an effect than you think. Some artist needs to help me with this one: If I'm looking straight on, an illuminated pixel in the center of the screen will appear to be in the same x,y location regardless of the z location (depth/nearness/farness). But, If I'm viewing from an angle, I tend to think that the pixel would appear to be in two different horizontal positions depending on whether it is being projected on the peak or the trough of the wave. |
|
| |
I've clearly spent too much time trying to figure out if this idea will work. |
|
| |
I understand that it's a sort of stroboscopic effect, but I can't see how you could represent anything in three dimensions other than a horizontal bar. |
|
| |
It's a bit tough for me to build a complete working model in my head, but as I see it, you wouldn't be "representing" it in 3-D. It would actually be in 3-D as the pixels would be physically at different distances from the viewer. |
|
| |
Indeed, it would. The position of the dot impinging the sheet surface would be controlled both by when the pixel was illuminated, effectively providing the depth, and where in x,y space the pixel is projected. |
|
| |
Maybe you'd do better to just use some oscillating corrugated rigid surface. |
|
| |
It seems like being limited to representing depths equal to the wave height is a drawback if you're trying to compete with conventional 3-D (which may not be the case). |
|
| |
Another random thought: If the projector isn't a laser, and isn't a fair distance from the sheet, the projected pixel would take up less physical space on the sheet in the "near" mode than the "far" mode. Not exactly sure what that would mean in terms of perceiving the depth differences. |
|
| |
As [half] said, that's a pretty difficult physical implementation (looks like you're around a thousand g's acceleration on a 30 hz sheet rate), but, hey - we're only half-baking it, right? The scan rate on the projector is not just higher than normal, it's *incredible*... Gotta calculate the depth resolution you want at the central plane of the oscillating sheet where it's moving fastest. At the cycle ends, it gets better resolution - but the image is compressed unless you are adjusting the frame rate constantly. |
|
| |
half, my use of 'represent' was in the sense that this device would display the image of a horizontal bar, which a representation of it, not the actual bar. |
|
| |
Still I don't see how it would be possible to display anything other than said bar. There's only x axis (and z, of course), but you can't have any corresponding y axis, not all at once. The y axis moves with the flap. I doubt that persistence of vision would be enough to compensate for the 'flap' rate. |
|
| |
I think I can see where you're coming from, [waugs]. That's what would happen with a commercially available projector. Imagine the projector has a frame rate that is many, many times faster than the rate at which the sheet is flapping. (i.e., "Sheet rate" times "Large number" times "WIBNI" kind of speed) |
|
| |
Love that graphic, [bris]!
If you project the pixel on the peak of the wave, and "surf" that wave, you've drawn a vertical line -- assuming the sheet flaps vertically. To then draw a horizontal line, project a pixel incrementally to the left or right, each time a peak passes. Now surf the trough, or any height in between, for your 3-D effect. Surf's up. Radical.
Can you do the same thing with a flapping flexible matrix of LEDs and eliminate the projector? Or even better, I bet there’s still a roll of flexible flat-screen TV cloth lying around the HB. |
|
| |
Ah, re-reading (paying more attention to the vertical dimesion) and seeing [bristolz]' drawing makes me realize that my "wave" concept may not be quite what you were thinking. I was visualizing the surface to look more like [Monkfish]'s "oscillating corrugated rigid surface" only the series of corrugations would be moving in one direction instead of oscillating. All that stuff I said...never mind. ('cept in my scheme you might be able to get the .5m amplitude without actually moving the mechanism that far) |
|
| |
It sounds like it'd work, all except for the speed of the sheet. It'd be pretty hard to move it that fast. Maybe a metallic sheet moved by electromagnets? |
|
| |
My holographic display idea tries to eliminate as many moving parts as possible, but then it's got that pesky laser business. Oh well. |
|
| |
This is baked. The military has several prototypes in use for holographic radar displays. The only difference is they rotate the projection surface inside, which is more predictable. |
|
| |
Wow, you sleep one night around here and you miss a lot. Great comments, everyone. Bris, you are awesome and I bow down before you. Refering to your comment long ago //provided the sheet was opaque and I am assuming it is opaque from its "white" description//, I was picturing a bed sheet, or something else translucent so you could see it from both sides. On second thought it had better be something very light if we're going to flap it that fast - maybe a white plastic film. As far as getting it to flap at a sufficient rate, I'd imagine the arms in a hard drive swing at an even faster rate. Of course, they don't swing .5m and have a lighter mass than our sheet probably would. Perhaps we'd have to scale it down a bit to deal with this issue. |
|
| |
Refresh rate required: Multiply the refresh rate you'd like by the pixel depth (.5m dimension) you'd like. If you'd like 60 hz and 100 pixels deep, that's a 6khz refresh rate. Luckily, with DLP technology we can get to 10khz. |
|
| |
[half] //I think there would be a problem forcing the top edge to be a node or zero amplitude point// True, I was imagining that the top few inches would be unusable. Either that or you set up another flapping machine up top. |
|
| |
Nice idea. This would be a genuine 3D image. Best viewed in a darkened room so that the only thing illuminating the sheet is the projector. |
|
| |
It sounds a little similar to the viewer made by Actuality Systems (link), but since they have vitually no detail on their device, coudn't say for sure. |
|
| |
st3f and froggy, looks like you're talking about the same technology - click on Product then FAQ and it talks about a spinning disk. Very cool stuff. I'm afraid that they don't support motion - their .mov files just have the camera walk around the device. Don't know if this was just a bad demonstration. Also, although they say they use a projector, it really looks like they are using lasers. |
|
| |
Fantastic idea, though. How long has that one been cooking in your brain? |
|
| |
~3 years, but never seriously. Only recently thought up the projector idea. Originally I was thinking of an analog device - replace the projector with a bright light shining on an object, then some lenses to project this image onto the sheet. However, the uses for that would be very limited. I guess it really came from Mr. Wizard (a tv show when I was growing up) and his demonstration using a vibrating string and a projector to make a screen out of a string - I'm just adding a dimension. |
|
| |
This vibration-based idea seems to require more engineering than the benefit yielded. In other words, you're leaning on awful lot of heretofore non-existent technology developments just to get 3D without using that fancy modern 'laser' technology.
For non-laser, I can imagine different approaches more likely to succeed. The whole vacuum tube idea seems so...last century. How about developing addressable translucent substrates extending existing LCD technology? |
|
| |
If you're designing without lasers and want a large-scale 3D projection system inside a vacuum tube, why not simplify this and just fix the translucent sheet on an axis and spin it? Rotation speed can easily be synchronized with refresh rates of two or three projectors, and you'd have simpler mechanics, easier projection calculations, and more available pixels per cycle to illuminate for a more realistic image. |
|
| |
//awful lot of heretofore non-existent technology just to get 3D without using that fancy modern 'laser' technology// Other than the sheet flapper, what's non-existent? And the point isn't really to avoid laser technology because it's fancy, but because it looks terrible. Compare the Actuality Systems pictures to film and you'll see what I'm talking about. |
|
| |
//why not simplify this and just fix the translucent sheet on an axis and spin it// I don't think this simplifies it at all. If you use 3 fixed projectors, you'll only have 3 slices of a 3-d object. I believe how the spinning disk system works is by mounting the laser on the rotor. The disk is then fixed in relation to the laser. Replacing this laser with a high quality projector may be quite difficult due to size and communication issues. |
|
| |
(quoting from idea subheadline)//3D Display without using fancy laser technology//
You called it fancy...
My suggestion employs the same projectors as your idea calls for. Say two, at perpendicular angles. Each does as your idea specifies, shoots a beam at a particular XY coordinate at the instant the controller calculates the substrate will be there to reflect (Z coordinate determined by timing). And since the substrate sheet is sweeping (spinning--center axis) around a 2 meter diameter 360 degree field as fast as your mechanism in a vacuum tube can handle it, your substrate 'plastic sheet' will occur in an infinite number of coordinates, allowing multiple projectors to paint even more detailed 3D imagery.
Beats "sheet flapper" and attendant limitations on imageable coordinates. |
|
| |
Great link from st3f--wish I'd checked if first.
(from the Actuality Systems FAQ:
"Basically, the 3D information is put into a format that is similar to thin slices of an apple around its core. Next, a high-speed digital projector shines these "slices" onto a special screen that spins at approximately 730 rpm. Therefore, you can place imagery at locations which have true (x,y,z) coordinates. The result is a convincing, interactive, colorful 3D image that can be seen from any angle in the room. "
It's clear from their spec sheets and color photos that they are not using lasers. They use RGB projection so some pixel depth is required for full range color. Sounds like at a minimum they start with 3 bit (8 colors). Resolution increases in the projection (or substrate limitations?) would be necessary to produce 8 bit or 16 bit, for thousands to millions of colors.
They have a readable white paper on the link as well. Sounds like they've baked a lot of the optics issues discussed here. Maybe we're all baked now. |
|
| |
[roby] After reading the white paper I agree. This thing's baked, and Actuality's system is very well thought out. If I had a spare $50k lying around I'd get me one of these. Don't know what I'd do with it, but it looks darn cool. |
|
| |
I think there's only one question left. Are there any unbaked great ideas left in the world? |
|
| |
How do you provide output for such a device. It is a good idea, but how do you input the picture/object that it will display? |
|
| |
A problem with the projector is focus, the sheet will be moving, with only one focal point somewhere in its path, most of the image projected will be blurry. "Lens Parralax" will also make the "pixels" further away be slightly bigger. |
|
| |
Trodden - I agree, and that's an issue. That would also be an issue for roby's spinning version. The only way I can think of at the moment is either by locating the projector far away (which would diminish but not remove the problem) or having a very light, small lens that is moved forward and backward very rapidly. Thus, the back plane would be in focus and an image page is projected onto all sheet points that lie on that plane, then the next layer is in focus, etc. |
|
| |
The input data for 3D projection is limited to digital data for all methods. I don't think there are any feasible analog-to-3D conversions in the pipeline. (The 'holodeck' is still in the realm of magic WIBNI) But there are a number of applications for holographic viewing of 3D digital models. Approaches to focus issues include using broad focal length projection (possible), pixel focus correction on the substrate (theoretical, remember we'd be projecting RGB bits, not analog images), or the Actuality systems method, a focusing mirror that rotates with the spinning sheet. |
|
| |
Does anyone know of a digital video filter that bluescreens everything that's not in sharp focus? If so, I have an idea for a 3-d videocamera. |
|
| |
Sounds like a very good idea. But it seems to me that the flapping might be a little unpredictable and quickly wear down the screen that is flapping. A lot like those little flag that people put on their cars, the constant flapping really seems to wear them down, but I am unsure of how this would work in a vacuum. Plus, there always will be slight variations in the screens used. Which would alter the flapping rate. Causing every single device to need to be tested individually if mass-produced. I think that the rapidly osculating rigid surface would work better. Like [galukalock] suggested using electromagnets, I think it wouldn't be to far off to use a super light opaque plastic, mount it on a metal frame and use a set of coils to create a magnetic field to move the thing back and forth. If the whole thing were riding on magnetic fields inside a vacuum it would be virtually frictionless. Although there still would be some, just an extremely small amount. Not to mention it would be a lot more quite than a flapping sheet of plastic, ever heard one of those; WHOOP-WHOP-WHOP-WHOP-WHOOP! Just thought I'd combine a few ideas into a new one. |
|
| |
I am suprised to see that I forgot to vote + for this the first time I read it. Oversight corrected. |
|
| |
//a super light opaque plastic, mount it on a metal frame and use a set of coils to create a magnetic field to move the thing back and forth// |
|
| |
I actually like that idea. There's much more mass to jiggle around, but it's not inconcievable that you can move it fast enough. Also, in the middle it will bow out and in, giving more of a round shape for 3-D projections. |
|
| |
I think our sytem is better than Actuality's though 5000 frames a second is nice. They don't return my e-mail. That hokey air force system ain't so hot. |
|
| |
There will come a day when we fear perfect replication. |
|
| |
The sheet could be designed to resonate at the required frequency, simplifying the control hardware needed. |
|
| |
Pretty much everything pertinant has been said here. Except, [bris] what happened to the in-line illustrations? I was starting to like them and now your piccy's have to be hidden behind a text link :-( |
|
| |
Almost all of my illustrations are as external links. Inlines are a new feature and I've been scribbling for a couple years with most pre-dating the inline feature. |
|
| |
couldn't you mount a sine wave shaped sheet in between two reels. Then just spin them.
very similar to those "never ending" rock climbing walls. that way the wave shape would be easier to control (perhaps discrete steps would be used instead of a wave shape) |
|
| |
I'm not quite sure I have the right image in my head
of what you mean. Sort of like a waterfall of rippling
sheet coming from above (and some way of returning
that sheet to the top in an endless loop? I like it, if
you could control it well enough. |
|
| |