3D Printed Helix Displays Graphics In 3D

It looks like [Michel David] and his team at volumetrics.co have really upped their game: the game being production of a 3D volumetric video display.

We’ve covered an earlier version of the same technique, and still the best technical explanation of what they’re up to is to be found at their old website. But it’s a simple enough idea, and we expect that all of the difficulty is in making the details work out. But if you look at their latest video (just below the jump), we think that you’ll agree that they’ve ironed out most of the wrinkles.

The newest version uses a standard overhead projector (“beamer” in “German”) to project images onto a rotating helix. By changing the image rapidly in sync with the rotor, you can choose how high up the corkscrew the light beam hits. Spin the rotor around (on a Dremel!) fast enough and your POV takes care of the rest. We really like the simplicity of the current setup, and the results speak for themselves.

Doing the math to figure out exactly where and when each point of light should be projected is no mean feat, but if you’ve got a very accurate model of the helix, it should be possible. And that makes this a perfect excuse for a big 3D printer — you can ideally make a helix model in your computer and then print out the exact same shape.

We’d like to see a lot more detail about the newer version, and maybe a link to some of the software. And we’ve still got one nagging question: how do they get the sync right between the spinning helix and the video display? We can’t see any rotation sensors or feedback. Is it just tuned?

38 thoughts on “3D Printed Helix Displays Graphics In 3D

    1. Any brace they put at the top would block the light from the mirror/projector. The image started out stable then became more unstable over time. Maybe a balance issue or the structure coming apart over time? Maybe a machined aluminum helix?

  1. “we’ve still got one nagging question: how do they get the sync right between the spinning helix and the video display?”

    Answer: They don’t. Not yet, anyway. You can tell because everything they are displaying has no variation in the vertical axis which is all that would need that sort of synchronization.

    To make it synchronized they will have to rotate the helix at exactly 60 times the frame rate of the projector which they might make work with feedback that compares an rpm pulse to the vertical blanking interval and drives the motor with the error signal, just like a PLL.

  2. It looks like they have a photocell or something mounted next to the Dremel for sync. Not sure if they’re using a laser or just the projected light for a source though. This will be really impressive with a purpose built rig using a proper spindle motor and a little better balance on the rotor. I wonder if you could mount the projector below the helix and use a magnetic bearing or something for a really seamless presentation…

  3. The math for calculating “real” 3D objects (as opposed to the extruded 2D shapes on the video) could be very tricky and you’d probably need a much higher refresh rate on ze beamer.
    Also that fast-rotating helix probably generates quite an airflow that probably adds to the wobblyness of the structure.

    1. The math isn’t so complicated. But a video refresh rate in the 10s of kHz would be needed for high quality 3D.

      Airflow I expect could be reduced by a flat plate at the bottom of the helix. Wrapping the helix in a cylindrical window would then cut airflow to almost nothing.

      1. “Airflow I expect could be reduced by a flat plate at the bottom of the helix. Wrapping the helix in a cylindrical window would then cut airflow to almost nothing.”

        Physics 101 fail. If you obstruct the airflow you increase the air resistance of the helix and effectively reducing the maximum speed you can achieve (for the same motor power).

    2. I would have thought you need a full frame refresh for each line of z-axis resolution, and synchronised projection (beaming) with projector change.

      Call it 18Hz (beginnings of POV) into 120Hz (high-end 3d projector refresh rate) gives you 6 lines.

      And then you’d lose xy clarity and size because of not spending as frequent visits of the screen to each voxel. Or you could project from inside the helix, back-project, and have an easier time with less depth.

      Now, make a much larger zoetrope, put a greater number of fins on it and sort the timing, and you’ll get there. Only just safer than lasers and exploding air.

    1. I could be way off, but if using a laser projector, that wouldn’t be an issue. I remember seeing a story here about a guy that did a “3d display” with a spherical light bulb protector “dome” that used a laser projector to compensate for the variable distances…

    2. Using lenses. You use a wide-angle lens to expand the image to the correct size, then immediately have it go through a magnification lens which directs the correctly-sized image into a consistently sized beam.

  4. Thanks for the great feedbacks, we will try some of your ideas…
    As for the synchronization of the Helix to the projector, in this prototype there is no sync. We wanted to test the 3D printed helices quality, and most importantly make it simple for anyone to make.

    1. Michel, please fix your “Get news and updates” sign up button–nothing happens when I enter my email address then click on it.

      Also, “We want you to build your own volumetric display by either printing a helix using our downloadable file…” Could you please post a link to the download? I don’t see it on your website.

  5. I’m guessing that it looks better in person, without camera frame rates screwing up the image. So, my only choices are to visit Germany or make one. But one major thing not mentioned in this article, which you won’t discover unless you click through the link to the Volumetrics website, is that they’re ???????????????????????????????????????????? you to make one.

    Gotta book some 3D printer time at TechShop…

  6. Spare projector? Check.
    Access to a decent 3D printer? Check.
    Spare mirror? Check.
    Spare motors? Plenty.

    I’m going to try to replicate this. I’ll report back with my results.

  7. I worked on the US Navy’s experimental 3D volumetric display.

    We used NEOS accousto-optical devices to steer the laser in the X and Y into the volume occupied by a two-bladed helix machined from a solid cylinder of aluminum 13 inches in diameter.

    Because of a limit in the time it took the Bragg diffraction grating to settle down to a new value of deflection, and the fact that we were going for a 20 Hz refresh rate, we were limited in the number of voxels we could display per frame to about 40,000.

    The software was written in C++ in 1989, using the Zortech C++ compiler, and the code ran on a 40MHz 386, talking to an ISA board designed in the lab that held the buffers for driving the X,Y D/A converters that drove the RF modulator that drove the crystals in the AO devices.

    The math was absurdly simple for this device: The Z axis was slices of time, and we got an interrupt when blade 1 crossed through 0 degrees.

Leave a Reply

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