Volumetric Display Projects 200 Million Voxels Per Second

Over the last four years, [Will] and [Gav] have spent their time creating a huge, high-resolution 3D display. The’re just about done with their build, so they decided to offer it up to the Internet in the hopes of people creating new 3D content for their display. They call their project the HoloDome, and it’s the highest resolution volumetric display we’ve ever seen.

The HoloDome operates by spinning a translucent helix around its vertical axis at 20 rotations per second. A pico projector above the helix capable of projecting 1440 frames per second (an amazing device by itself) displays 72 ‘z-axis’ frames for each of the 60 ‘x and y frames’ per second. The result is a 3D display with a 480 * 320 * 72 voxel resolution capable of displaying 20 frames per second.

This isn’t the first time we’ve seen a swept helix used as a volumetric display, but it is by far the highest resolution display of its type in recent memory. [Gav] and [Will] have put their HoloDome up on the Australian crowd-funded site Pozible if you’d like to buy your own, but thankfully the guys have included enough detail on the main site to reconstruct this project.

Check out the video after the break to see the HoloDome in action.

[youtube=http://www.youtube.com/watch?v=5Q1CMNvfCNI&w=470]

42 thoughts on “Volumetric Display Projects 200 Million Voxels Per Second

  1. Every time I see a mechanical volumetric display like this I try to remind myself that some of the first televisions were mechanical as well.

    Looks pretty snazzy but I looked for an explanation for why they chose only green for their examples, as the projector seems to have more than one color LED for display. Is there some reason that multicolor won’t work correctly?

    1. Just guessing, but colour on DLP displays is time-multiplexed, displaying R/G/B one after the other. At least some, dunno how many, of the projectors use a spinning colour wheel in front of the DLP chip.

      So RGB colour drops the frame rate to 1/3. If they can only get 20fps now, colour wouldn’t work, be too slow for persistence of vision, and would flicker and look awful.

      1. Nah, blue’s harder to see. The human eye’s most sensitive to green (well, yellow really), and least to blue.

        I’m sick of blue anyway. Since blue LEDs got cheap at the end of the ’90s, every cheap bit of electronic crap is blue-LEDded up. Looked smoov in the 80s, looks cheap now.

      2. @Greenaum:

        I agree. I remember electronics used to have red, green or yellow. Now they stick blue on everything. It doesn’t make it “cool”.

        And you are right about the frame rate. That is the reason they mention in their web site.

    2. Another of their videos does mention they have colour, but doesn’t show it. I suppose you could do it easiest with 3 projectors, one for each colour, placed around the helix at wherever’s best, and synchronised together. Would need 3x the data rate, but that’s just a matter of tripling the width of the output buffer.

      They seem to have hit the limit of their hardware at the moment, but nothing stops them triplicating it, then rigging up some sync.

      1. Hi, we have not tried color, but know how it will affect the display. Currently, we are rendering 1440 slices per second in monochrome. If we use a pallette of 3 primary colors, we will take a hit on the slices per second and it will drop to 480, and thus our vertical resoluion will drop from 72 slices to 24.

        Color will come in time. But we feel the joy of seeing a floating hologram, feels magical enought without it.

        cheers
        Gav

      2. If you use 3 projectors, it wouldn’t lower the frame rate at all, tho as I said, you’d need to sort the sync out. But with 3 projectors you’d essentially be projecting 3 “holograms” over each other at once, so the frame rate / resolution would stay the same.

        You’d need to process the image to compensate for having the projectors spaced around the helix. Perhaps if they were next to each other, then you could get away with just shifting the image a bit laterally.

        Getting the extra colour channels wouldn’t be too hard, since video is output with 3 channels anyway. If you can get another 2 projectors, give it a try! I’d love to see the results. It’d be the only thing better than a monochrome hologram!

  2. Nice demo!

    The problem with doing full color is that this DLP projector outputs 1440 binary (only “on” or “off”) frames per second. For normal display applications, DLPs combine many binary images using a variation of pulse width modulation to vary the intensity of each color channel.

    This particular projector outputs 1440 binary fps. If they cut their Z-resolution by 3 (480 * 320 * 24 voxels) and enabled the projector’s red,green, and blue LEDs, they would still only have 3-bit color (e.g. red, green, blue, r+g, r+b, g+b, “white”, and black). Similarly, combining 3 projectors (1 for each color channel) would allow for keeping the original resolution (480 * 320 * 72 voxels) with 3 bit color, with the added cost and complexity of aligning 3 projectors.

    Essentially, for a given projector’s binary frame rate, you can choose to allocate frames to either depth layers or bits of color. The product of these two parameters must be less than or equal to the binary frame rate.

    TI makes higher-end DLP projector chips that run much faster – 32kHz. At that speed, it is possible to make a volumetric display with 8-bit color, or even 24-bit color if you have a separate chip for each color channel. However, the development board for this is around $10k.

    In theory, you could hack a $500 off-the-shelf DLP projector to do this, but TI does not release the data sheets for their DLP chips, so it would require some reverse engineering. You would also probably require an FPGA to convert incoming DVI signals to the appropriate binary frame pattern and drive the DLP chip appropriately.

    1. ” You would also probably require an FPGA to convert incoming DVI signals to the appropriate binary frame pattern and drive the DLP chip appropriately.”

      That is what I thought when I read their description. This is a task better suited for hardware, not software.

    2. Nice follow up Jon.

      We did spend about a year trying to ‘Hack’ DLP, and ended up hitting a brick wall. We removed a color wheel and use the slices to increse from 60 fps to 180, but thats still very poor for volumetric.

      We had enormous problems trying to get the projectors turn on without globes in them, which we did to test using a laser rather than a conventional HID lamp.

      One moment of pure genius however came when we figure that plugging a resistor into the lamp socket might work. The first resistors blew up in a puff of smoke, however, we persevered, and in a Eureka moment, pluged an electric kettle into the lamp socket. It worked, and the DLP became active, and putting several thousand volts though the kettle did not seem to bother it!

  3. I would use a laser mems projector. Frame rates are very high and full color. Video rates per 8 bit pixel are 300Mpix/sec. Can modify each pixels intensity using Laser driver DAC. I think it’s maxims new 3601 chip. Or something like that. Lasers are so much cooler anyway.

  4. Thanks for all the positive comments. To answer some of your questions.

    The questions about color, are all explained in detail on the openvolumetric.org site, but its basically a trade off bewteen color fidelity and frames per second. We need loads of fps to get good vertical resolution whilst avoiding flicker. 20hz is about the slowest you can go before flicker gets annoying.

    We can go faster and have color, but need raise the funds via our pozible.com/holodome campaign to buy some more hardware for prototyping.

    Regarding Laser MEMs, we’ve not explored this avenue, but might if we raise our target on pozible.com

    We can curretnly choose RED, GREEN or BLUE, but GREEN is the color that the human eye is most sensitive too. However, BLUE looks awesome so stay tuned for some of that.

    If you like our HoloDome facebook page, you will get notified when we have some more content.

    Next up is the famouse Princes Leah scene from starwars, so stay tuned!

    After that we are open to suggestions?

    Holographic sneezing panda?
    Dubstep dancing baby?

    Drop some comments on our facebook page, and if you’re are a character animator, let us know if you want to knock up a 10 second showreel in 3ds max for us to render in the HoloDome.

    p.s For those of you wondering why we went with Pozible. We did try Kickstarter, but its USA only at the moment. This might change soon.

    p.s.s I loved the reference to ‘Another World’. That game gave me the chills when I first saw it on the Atari ST. You can actaully play it on Ipad now!

  5. Matias…..we aggree with you

    Imagine the posibilities! Holographic Soccer, Pacman with layers of pills so you can go up and down too! Asteroids, and of course the Starwars tench run….’Look at the size of that thing’

    We believe the medium will all for the rebirth of old school vector based games, with a futuristic and mesmerizing twist.

  6. The airports should be looking into this. to be able to see the sky around the airport in 3D?!?!?!
    I think these guys should RAISE the price (or keep their hands in) if the airports want to consider it.

    1. There’s also a lot of money in medical imaging. Dunno how patentable your gadget is, so you’d probably need some decent legal protection to avoid getting ripped off, but you could approach the companies that do medical imaging.

      Dunno if the people that make MRIs etc subcontract or do their own imaging hardware. I do know that it was an early market for cutting-edge 3D rendering (to a 2D screen, I mean). MRI and the like work with a series of 2D slices anyway, so the data’s in a format you could work with.

  7. Interesting, but like a lot of ideas a homebuilt volumetric display is nothing new. Over thirty years ago, the article ‘Graphics in Depth’ in BYTE magazine May 1978 (69Mb size download, available here ftp://helpedia.com/pub/archive/temp/Byte/Uploads%202012/BYTE-1978-05.pdf) used a rotating oblique mirror instead of a helix. The problem I see with a helix is, it takes a lot of energy to move as it basically a vertical fan, thus limiting rotational speed unless in a vacuum. The oblique mirror pushes a lot less air around when rotating.

    1. Hi Mr Name ,
      Regarding your comments. We are not proclaiming to doing something new. In fact swept surface volumetric displays have been available before , targeted at medical imagining and design. They were about 30k to buy , too much for a consumer product by a long way. We have experimented with mirrors, but the problem with them, is that mirrors are usually only useful in creating an illusion of volumetric space, and not creating a space that can be be viewed from 360 degrees by a group of people simultaneously. There are many ways of moving a swept surface , and the helix was chosen for our prototype due to the ease of driving it. The air movement is only an issue when the helix is ‘naked’ , our next protype will have it housed in an acrylic cylinder, and the images will be projected from below onto a translucent helix. This should generate very little air movement indeed.

      Concave mirrors can be used to enhance our display further by projecting the image into space using the mirage trick, but the cost of this is prohibitively high for a consumer product and it also limits the viewing angle.

      Our number one aim, is to make this technology accessible to the open source community and creating a framework for a set of standards for volumetric projection.

      There are lots of youtube videos proclaiming to having holographic projection, but when you dig deep, you will find that the vast majority of them, are using mirrors and fixed viewpoints to trick a static audience into thinking they are seeing a hologram.

      Our prototype is fully volumetric in that you can look at it from any angle you choose, and we’d like to see the technology used bring friends around a shared space to interact with each other, much like with traditional board games, and that could entail playing ‘Holo-chess’ or watching volumetric captures of sporting events. Theres really no limit to what is possible. Only the imagination of the people who will create content.

      cheers

      Gav

      1. Gav, thanks for your detailed reply. Encasing the helix in a lucite block or something similar sounds like a good idea, I hadn’t thought of that! (slaps forehead). However I’m still a little puzzled why you reckon the rotating oblique mirror doesn’t provide a 360 degree view, if projection is done from overhead or below? It seems you’ve really done your homework on this so all the best with it, it really looks very good.

    2. Mr Name Required –

      I have never seen that BYTE magazine article before. How interesting! It also seems available here.

      I was the founder of a volumetric display company called Actuality Systems (Perspecta was a 100 million voxel display); seen a lot of articles over the years, but I can’t say I’ve ever seen the paper you cite. It seems like prior art for a number of systems people have built, indeed.

      -Gregg

  8. Hi Philippe,
    Translucency certainly comes with all volumetric displays. Its can be seen as a feature or problem depending on your mindset. There are displays which claim to offer voxel occulution, but the devil is in teh detail here, as those displays are not actually volumetric, rather they are refecting a pre-rendered view of an object in a particular direction. So, in that case, you might argue that there is no voxel occusion, as there are no voxels to occude, rather the ‘pixels’ that make up the object at the other side are not rendered in that particular view.

    Of course, it is still possible to occulde voxels during rendering if you so wish. You just need to decide which viewing direction you wish to affect. For example, if you were rendering a human body , you could selectively turn off voxels inside, or enclude half of the entire image. I cannot think of a method whereby you could have a true volumetric display with solidity. But who knows what the future might hold.

    Thanks

Leave a Reply to MatiasCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.