DIYing A VR Headset For Cheap

VR has been developing rapidly over the past decade, but headsets and associated equipment remain expensive. Without a killer app, the technology has yet to become ubiquitous in homes around the world. Wanting to experiment without a huge investment, [jamesvdberg] whipped up a low-cost headset for under $100 USD.

The build relies on Google-Cardboard-style optics, which are typically designed to work with a smartphone as the display. Instead, an 800×480 display intended for use with the Raspberry Pi is installed, hooked up over HDMI. An MPU6050 IMU is then installed to monitor the headset’s movements, hooked up to an Arduino Micro that passes this information to the attached PC. The rest of the build simply consists of cable management and power supply to all the hardware. It’s important to get this right, so that one doesn’t get tangled up by the umbilical when playing.

While it won’t outperform a commercial unit, the device nevertheless offers stereoscopic VR at a low cost. For a very cheap and accessible VR experience that’s compatible with the PC, it’s hard to beat. Others have done similar work too. Video after the break.

18 thoughts on “DIYing A VR Headset For Cheap

  1. There are dual 1440×1440 IPS displays on ebay including the HDMI to dual mipi controller for around $100. There are also some higher res 2:1 aspect ratio panels meant for car rearview mirrors in that price range too.

    Unrelated, you can find 4:1 aspect panels too.

    1. Or you could just buy a used Galaxy S6/S7/S8/S9 or Note5/6/8/9/10 and supported GearVR from eBay or anywhere. Total cost won’t be much different. For software there’s an app called ALVR.

  2. If you have a google cardboard capable headset (or Oculus Quest) you can use Riftcat’s VRidge to stream SteamVR to your device. They also have a tracking controller emulator app. For a year or so until recently, I used this setup with 2 old phones emulating controllers to play around in VR until I was convinced the ecosystem was rich enough to justify a dedicated headset.

  3. I’ve been into various degrees of messing with similar lashups. I dunno if I’m trying to combine too many objectives into one thing though. Or it’s that hacker ADD thing, do this, no that, no this again. Anyway, one application is trying to be the display for a wearable computing system. One application is trying to be a night vision headset, and one application is trying to be a generic content consumption device (casual game, 3D video etc) Just got given a Fire 7 with dodgy USB port, which might get bodged into one of the cheapy phone holder sets to mess with.

  4. There’s a heck of a lot more to a VR HMD than a screen strapped to your head. This is akin a “DIY iPad” that consists of an Arduino with a resistive touchscreen attached. It has a sort of baseline resemblance, but not too the degree of anything useful. To wit:
    – Optical correction to maintain orthostereo
    – IMU sensor fusion with baseline tracking system for 6DiF tracking
    – low-persistence pixel driving
    – sufficiently low round-trip latency (motion-photons latency) between IMU sampling and pixel energisation to stay within the acceptable 20ms budget. NOT trivial, everyone abandoned the ‘just treat the screen as a monitor’ approach very quickly for good reason
    – software pre-warp
    – Software rotational latency compensation (AKA timewarp)

    Etc etc. These aren’t ‘nice to haves’ for VR, they’re hard minimum requirements for something that is on the minimum edge of acceptable.

        1. I don’t know, back in 2014 I built my own headset using a Galaxy S3 and some 50mm 5x DCX lenses. I played Quake II for 2 hours at a time without vomiting. Guess it just depends on your constitution. I found it quite enjoyable, I impressed a few DK1 owners as it was visually superior. Took it to a lot of Meetups and Maker Faires and Tech expos. No one got sick. Benchoff got to see it at one of the Meetups and said I should post it here, but I never got around to it. Didn’t really document the build.
          Sure, everybody wants a Holodeck where it’s indistinguishable from reality, but how do you know that your not already in one? kind of nice to stay grounded with a few minor imperfections. Same with audiophile stuff. I don’t need to hear the subtle nuances of Pavarotti’s tuxedo sleeve as it rubs his jacket, seemingly right next to me to enjoy good music.

    1. -Optical correction is something that needs to be tweaked, but with simple circular lenses that can be dialed in well enough.
      -The easiest thing to do would be to get a Vive tracker and attach it to the front or top of the headset, but I don’t know if there are any software paths to get that working with SteamVR, as at least with phone based DIY solutions you have Trinus or Riftcat. Though with how ridiculously overpriced Vive trackers and base stations are, I don’t know how viable that would be from a cost perspective over getting a used Vive.
      -(Combining both points about the display) Using one of those “Dual 1440p displays attached to a driverboard” solutions you can get from Aliexpress is good enough for something DIY. When you’re concerned with a mass market device where every single fraction of a nanosecond counts, that’s when you start to worry about things like “low-persistence pixel driving” and “latency between IMU and the photons of the subpixels”. DIY, where the end user likely has a far higher tolerance for the hardware being less than perfect, a “Treat the screen as a monitor” is well more than good enough.
      -(Combining both software points) Well that’s a software side issue, not a hardware one, isn’t it? Again, software exists for making the garbage phone ‘vr’ work for PC gaming, but that’s as close to being able to bridge non-standard hardware and SteamVR I’m aware of.
      If there was a hacky way of spoofing a headset being connected to a PC, have positional data come in from a Vive tracker acting as the headset tracking, and using the desktop mirror view with it set to the full headset output full-screened on the headsets displays as if it were just a monitor, that might just about work, bit I don’t know if the latencies involved would be even beyond those who have the heartiest of VR legs be able to cope with.

  5. Killer App? Flight simulators. Consider how much people spend on a multi-monitor system to make their flightsim more immersive and realistic. X-Plane supports VR, Microsoft is endeavoring to roll VR support into their latest and there is even a fair bit of VR support in FlightGear, though it doesn’t seem to be a mature feature at this point.

    1. half the ‘fun’ and immersion of sim pits is the ergonomic reaches. you might be able to put optical trackers on your controls and hands but that still increases your tactile separation from the controls; if a flight yoke’s motion is “wrong”, it’s easier to forgive if you can see it. Example being, pivot vs slide flight yokes. a pivot yoke can’t be AR’d into a slider and v/v. but if you have a slide yoke on your ‘pit and you’re flying a pivot yoke aircraft, it’s easier to translate your gestural habits

  6. This is a “version” of what we did at Ono-Sendai in the early 90’s. But we didn’t have arduinos or Rpi’s. We had to take apart camcorders for the LCD displays and drive them with a little Forth board with a Motorola chip. The optics were some plastic things you could put a leaf or bug into and look at them. They were sold through science shops. We put the LCDs at the focal plane.

    I am glad the tech has really improved and also come down in price. Plus the display drivers are much better than what we had – I had a 12 MHz PC- state of the art!

    It takes any ground-breaking tech 25..30 years to hit “maturity”. Holds true for cars, aircraft, lasers, cell phones, internet, and VR. Starting in 1990, that puts VR at 2015 to 2020. Look at when commercial VR helmets and inexpensive motion capture came out – right in that period.

    Keep in mind a VR helmet from VPL Research was Big Bucks – only the government and big universities could afford them. A decent headset is $400..$800. You can get a decent system for under $2K. A Cave was spendy because of the projectors; now you can get them at Costco or Best Buy.

    I had to write a little tank game as a demo. Very basic graphics. Maybe 15 FPS; just out of the barfagenic zone. Now you can get a good graphics card for under a K-buck, with a system to run it totaling under $3K.

    Short version; you can get the hardware (and freeware) for a kick-ass system and put it on a credit card.

    BTW, for those who are old, we were contracted by Sega for their VR game system for Christmas 1993 to do the 3DOF tracker.
    This is an interview I did on the the part with Sega. https://www.sega-16.com/2006/07/interview-bandit-sega-vr-helmet-engineer/
    The whole Ono-Sendai saga would take at least a 6-pack of beer to tell.

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.