VoLumen — The Most Advanced Persistence Of Vision Display Yet

Whoa. We’re just blown away by this new project by [Maximilian Mali] and [Sebastian Haushofer]. It’s a stacked Persistance of Vision display, with 9 layers — effectively creating a Volumetric 3D POV Display.

We recently shared one of [Maximilian’s] other projects, The Ripper CNC Machine. As it turns out, the reason he built The Ripper was to aid in the manufacture of his VoLumen project. He’s been designing these Volumetric 3D displays for about 3 years now, with the first iteration called the viSio, capable of 40 fps 3D video without the need for any 3D glasses.

The new and improved VoLumen features 34 micro-controllers, each with 512MB flash memory for storing animation data. In total there are 1024 high power RGB LEDs, which draw a whopping 200W at full load, making it bright, crisp and visible even in direct sunlight. It’s an incredible project that [Maximilian] started when he was only 16 years old.

You have to see the video of this thing in action.

Today, [Maximillian] and [Sebastian] are finishing up a mechatronic engineering degree in Vienna, and are hoping to receive a grant to continue their studies abroad. So if there are any university coordinators in our midst — let’s get these guys some support!

47 thoughts on “VoLumen — The Most Advanced Persistence Of Vision Display Yet

  1. This is a very old idea, the rotary POV displays were very popular advertisement displays in the 80s and 90s. 3D versions have been discussed, but until recently small volume production of such a thing would have been expensive. But enough history, let’s talk drawbacks:

    Fixed video memory: the onboard flash memory suggests that it will not be able to play back live generated video such as games.

    Resolution: The resolution seems to be 32x32x1024 pixels, you can not really display anything complex with only 32 pixels in two of the dimensions.

    Curved space: These displays were usually made with the LEDs mounted on a PCB strip oriented the same as the spinning axle. The ones that weren’t would display distorted graphics since the inner edge of the display would be shorter than the outer edge. Being 3D there is no way this display could avoid this problem. It could theoretically be possible to build models that take this into consideration, but not with only 32 pixels of resolution.

    Noise: This type of display is going to be very noisy unless you vacuum pump the dome, but then you probably won’t be able to get rid of the heat. Another way of limiting noise would be to use a very thick dome, but that would make it heavy and expensive, and would also limit cooling.

    1. Why talk drawbacks. It’s good, well engineered, good software, fantastic concept. Everyone knows it’s not for anything and has no uses other than to gawp at but gawping is good. No one can say where it may go in the future but for sure if people didn’t do things like this we would still be pulling carts with horses. :)

        1. Well then let the disagreements begin! You claim that 32 pixels won’t be enough depth. 32 pixels would be good enough for indicating layers: a car could drive in circles around a building, and although the car wouldn’t be rendered in HD quality, you’d still see the ideas. Motion blur will help trick your eyes into ignoring the lack of detail.

          You claim that “curved space” will be some kind of a problem with insurmountable distortion issues. Polar mathematical transformations solved that problem long before the advent of digital computers – it’s just a matter of applying some well understood math to the timing. If the timing isn’t good enough for what he’s trying to display today, he can solve that easily by using faster processors. It’s already good enough to prove his ideas are sound.

          You also claim noise will be a problem, but somehow conveniently neglect to say which environments that will be a problem in. Here’s the short list: you probably won’t want it running in a sleep clinic, a church, or during a violin recital; none of which are on the short list of clients with business needs for purchasing volumetric 3D displays. Plus, there are a lot of potential ways to deal with noise issues: airfoils to wrap the blades, for example, or a glass case. You’re really stretching to invent this as a problem.

          Instead of calling them drawbacks, I’d say this project is incredibly cool already, and that I can’t wait to see what else this guy will come up with!

    2. man, if you’re going to dog the project that much, at least compliment it A LITTLE or propose fixes or adjustments that would remedy the issues. Anyone can come up with a thousand reasons why something is a bad idea. it’s not a special trait or anything that makes one look smarter than the rest of us, just makes one look a lot unhappier and lazier than the rest of us.

      1. Did you even notice what i said about every hackaday article comment section being full of drawbacks, and me just attempting to list them right away? Also, if you read between the lines you can easily see suggestions for improvements, i don’t feel i have to spell them out in detail.

        Here is a suggestion for improvement: If the radial dimension is extended all the way to the center, the display volume will be one solid cylinder instead of a toroid, making it possible to display much larger objects undistorted. Using about double the number of LEDs and extending them all the way to the center would make a lot of sense.

        1. Sorry about that other comment, I kind of failed typing the information in the right boxes…
          Actually we wanted to do that at first, but as that would have required twice as many leds and drivers and our budget was limited this was not possible.
          Furthermore the objects displayed in the center on the other side of the device you are standing wouldn’t always be visible due to the overlap of the arms in some regions.

        2. It’s OK, Sven. I agree with many of your estimations. And recognize that nothing you said is news to the inventor/builder/developer/etc.

          It’s a project. And by definition it is reasonable to expect progress in the form of improvement on the thing, whatever it’s called.

          One area not on ‘the shortlist’ is my living room. One reason for having it that is not on ‘the shortlist’ is a 3D experience, with a(suggestion) an interactive element to it i.e. – mouse is laser/ir pointer? touch/motion? some other sensor?

          Noise/Vibration/Size/Weight – all these are market acceptance drivers.

          Anyhoo, my two cents…

    3. Your 32x32x1024 isn’t accurate. Remember, the blade is MOVING along the rotational axis. As long as the LED can switch very quickly, you can have an effective resolution of something much much higher (persistence of vision relies on seeing something that isn’t there any more – the LED doesn’t have to and normally doesn’t display a pixel for very long. I mean, you can clearly see IN THE VIDEO, multiple words all over the inner and outer track. Effectively, you have a 32 high x 1024×1024 cylinder.

      Why did you bring in a problem that has already been fixed – use a spinning axle and problem solved? I’m not sure what else you could use (3D grid of LEDs? Not sure how it would be distorted since there’s even less movement). You can also change the density or speed of light flickering to compensate for any odd rotating shapes, though I may be misunderstanding what distortion you’re talking about.

      Noisy motor? Have you heard any hybrid car or electric car driving in electric only mode? Electric motors make almost no sound driving around 30 tons at hundreds / thousands of RPM. Keep the RPMs low (40 fps should mean 40 or 80 RPM) and you’ll remain extremely quiet or silent. Most Computer fans only make a lot of noise because of the air blowing through, not because of the motor itself.

      Fixed 512MB RAM… How does having video ram exclude the possibility of a passthrough? This I don’t even get. Just because my computer has a 500GB hard drive suddenly makes me stop being able to download a youtube video from a remote source (Internet)? The beginning of that video even shows him plugging in a usb cable into it for a “passthrough mode”, so as long as the laptop can translate the display into whatever format fast enough, the system should be fine displaying it.

      So I just destroyed all your arguments with real world examples or common sense…

      1. Effectively, you have a 32 high x 1024×1024 cylinder.
        You were so excited about debunking his claims that you overreached.
        It’s a 32×32 pixel grid, spread among 32 PCBs with 32 pixels each. So while the 1024 angles can increase, the number of voxels is still fundamentally still going to be a 32x32xN torus.

      2. 32x32x1024 is very close to the proclaimed 1 million pixels, note that i said “seems to be”.
        Also you are completely wrong, there is no way you would get a 32x1024x1024 cylinder using this technology.

        The distortion is the space curving around the central axis, the path of the LEDs nearest the axis is shorter than the path of the LEDs farthest from it. The radial resolution is only 32 pixels, way too little to be able to display anything undistorted.

        The motor might be completely silent, it doesn’t matter, the boards will make a LOT of noise just moving through the air. Also 40fps means 40 rotations per second, not 40 rotations per minute.

        I am not speaking in general terms, i am speaking of this specific display. The large onboard FLASH (not RAM) of this specific display means it will most likely only display stored frames making the possible uses very narrow.

        1. I think the use of flash was just to let them get the optical / mechanical bits working, without having to worry about routing megabytes of data through 32 spinning PCBs. That’s a separate problem, and a more boring one.

          So they started on the bit that actually shows something, the rewarding bit. The rest comes later. Indeed the work they’ve done so far is the vital part, you have to have the visible bit working before you can pipe data to it.

    4. You could do both… have a dome full of air for cooling, within another bigger dome that’s evacuated. Noise problem solved. If there really is a big noise problem with it. Perhaps a bit of aerodynamics and some good bearings could solve a lot of it anyway.

    5. Another way to get rid of the noise is to use magnetic bearings. Rare earth magnets would reduce the sound of bearings and the only real noise would be from the motor running the thing.

  2. I so wish that high brightness RGB led’s would get a lot smaller, it would be epically cool if the resolution of these things would double.

    and as a “home made project” this thing is better than most corporation prototypes.

    1. one could stagger the LEDs and get any resolution they wanted. all compensation could be done in software. they don’t have to be in a perfect line, they can be arranged however you need to get the resolution you want, you just have to compensate in software.

  3. Forget university coordinators – I’m in higher ed and have yet to see a university handle a bright idea well. Get them to Festo or some similar group who’ll actually run with it.

  4. With only 9 layers it’s pretty low resolution vertically. An easy way to double that would be to put LEDs on top and bottom pf the blades, with the bottom ones at the edges so their light is visible.

    9 layers with 18 vertical pixels.

Leave a 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.