OpenVR: Building an Oculus Rift for only $150

Home Made Oculus Rift The Oculus Rift is a really cool piece of kit, but with its future held in the grasp of Facebook, who knows what it’ll become now. So why not just build your own? When the Oculus first came out [Ahmet] was instantly intrigued — he began researching virtual reality and the experience offered by the Oculus — but curiosity alone wasn’t enough for the $300 price tag. He held off until he had a useful purpose for it, and as it turns out he did — he builds and flies multicopters, for which an FPV setup would be super handy!

Other FPV setups cost close to $300 as well, so getting a device with more features just makes sense. Promptly after realizing this, he faced the Maker’s Dilemma: Buy it, or build it? To test the waters, he decided to order some aspheric lenses to do some quick tests with a smart phone and a ghetto cardboard box setup — the results were surprisingly good. No turning back now!
The hardware consists of:

  • 5.6″ 1280×800 LCD
  • 3D printed enclosure
  • 12V power adapter
  • USB to TTL
  • 50mm aspheric lenses (5X zoom)
  • Arduino Mini Pro
  • GY-85 9DOF IMU
  • Various wires, foam padding, and glue

The majority of the cost here is in the LCD, with everything else being pretty inexpensive. Once it’s all built (details on his blog), it is time to get the software, downloaded right off of GitHub. The rest is pretty self-explanatory — just take a look at the results! We’re even tempted to build one now. Videos below.

We’ve seen other DIY Oculus setups before, but they just seem to keep getting better!

Comments

  1. rasz_pl says:

    FPV is inherently mono, so why two lenses and stereo headset? Its just lost resolution at this point.

    Are stereo optics easier?

    • It doesn’t have to be mono. There are 3D FPV projects.

      • Robot says:

        There are a couple of problems with this. . . well, not so much problems but issues:

        1) Most video downlinks are low resolution; thereby providing less information for your visual cortex to draw depth data from. (There is the DJI 2.4 GHz HD system which seems promising but keep in mind that two cameras are potentially required.)

        2) Depth data through stereoscopic imaging is moot when converging at infinity.

        3) It is very hard for one’s visual cortex to make use of “unnatural” image data. One’s brain has spent years calibrating to one’s eyes. To suddenly be presented with image data through cameras that are poor quality in comparison to the eye makes it hard for your brain to fuse the image data. This problem is endemic with VR systems, so much so that many users of VR don’t experience image fusion at all without some amount of training.

        4) This leads me to a thought: and ideal FPV setup would utilize one high quality camera and a laser range finder (or similar) and downlink both data to a base station which would process the data and present a user with a simulated stereoscopic image. This would be useful for negotiating situations in which depth sensitivity needed to be adjusted. For example; flying a multi-rotor through a building would require a much different depth sensitivity than, say, soring a fixed-wing at 5,000 ft.

        tl:dr
        Neat stuff.

        • alfie275 says:

          Regarding “unnatural” image data, I thought the opposite was true? For example they do that experiment whereby someone wears glasses that flip everything upside down, and it takes on the order of days for the visual cortex to adjust. Another example is with artificial sight systems such as implanted electrode arrays or those tongue-pad things, where the arrays are very low resolution.

  2. Simon says:

    There’s also another project to use the $15 STM32F3 Discovery board as the tracker portion of the Oculus Rift (compatible with official SDK)

    http://yetifrisstlama.blogspot.ca/2014/03/the-foculus-rift-part-2-reverse.html

  3. Telepath says:

    NIce job rebuilding the Oculus Rift like this :).

    But 1280×800 is the original resolution of the Oculus Rift – which made me sick even before moving around. Staring at blocks might be enjoyable for minecraft players, but for anything serious you need considerably higher resolution.

    • Sam says:

      Something like an iPad retina display could be great. The iPad mini 2 has a 7.9″ display at 2048 x 1536 with 326 ppi, which might be worth playing with.

      • Michael Buffington says:

        It’s not the resolution (or lack of) that makes you sick, it’s the sensor lag.

        I’ve built two DIY head mounted displays with motion sensing and I also own the first Oculus Rift dev kit.

        The second of my two HMD’s was built as an attempt to improve the first. I used an iPad Mini screen (not the Retina display) with a simple Chinese 9DOF board (a combo of gyro, accelerometer, and magnetometer).

        Like you, I thought the increased resolution would reduce the motion sickness (which I’ve never noticed, but others who’ve tried my rigs have). Apparently the lag was still enough to bother people.

        So I upgraded the motion sensor. I swapped the 9DOF board for a 3D Robotics APM 2.6. The APM 2.6 is a very capable multi rotor controller. Totally overkill for what I needed, but the sensors are top notch.

        Apparently the updated sensors have drastically reduced or eliminated the lag for those who had it the worst before.

        Out of curiosity, I hooked the APM 2.6 up to my original lower res screen, and the lag was gone as well.

        There are indeed issues with resolution, as well as screen technology. Higher resolution reduces the “pixel grid” effect. Screens with tighter spacing between pixels reduces the pixel grid effect as well, even with identical resolutions and similar sized displays. It’s my opinion that the resolution has very little impact on the feeling of “realness” – it’s the motion sensing input.

        If it’s wrong, the contract our brain keeps with all its inputs is breached and the brain lets it be known.

        Lag is what the Oculus Rift team have been putting a lot of effort into eliminating. If you’ve seen their latest prototypes they’re using motion tracking systems that use cameras and IR reflector dots in addition to sensors.

        Problem is that even if motion tracking get’s to be 100% accurate with zero lag it will always be a certain amount of milliseconds behind the rendering/physics engine of whatever simulation these headsets will be used for.

        • Telepath says:

          I’m sorry, I was a little unclear. The motion did make me sick after a short while, but that wouldn’t have stopped me from using it. But looking at that pixel grid close up is a long way from fun for me.
          I still had high hopes for the consumer version, until Facebook bought them.

          So, I really appreciate any work on alternatives :)

        • Dodo says:

          How did you connect an iPad mini display to a video source? I cannot find anything on internet about it, only for the bigger retina display which has eDP.

      • Whatnot says:

        Samsung just released two new tablets named galaxy s, one 8.4″ and one 10″, both have 2.560 × 1.600 superamoled screens. 359PPI and 288PPI.

        Retina is so yesterday – it seems.

        8.4 is a bit big for this and it would be hard to get such a spare screen at this point though obviously, but it shows retina’s resolution is becoming slowly mainstay.

  4. pcf11 says:

    Only $150

  5. Eirinn says:

    If hardware was the only problem then yes you could maybe build an oculus “clone” for $150. This is not the case however. It’s probably a very nice piece of kit, no doubt! But at this point, I don’t really think anything can compare to what they’ve got in store. Eagerly awaiting my DK2 (which I got sponsored through work) :)

  6. Chris Miller says:

    I’m not sure where $150 came from, considering the displays are $113.50 each. Unless the ebay seller jacked up the price because of this article…

  7. wicked says:

    Made one a few years ago while waiting for my original rift kickstarter to arrive.. here’s all the plans for it.

    https://www.thingiverse.com/thing:31758

  8. efahrenholz says:

    Correction, the screen isn’t the most expensive part of the project–its the 3D printer you need to print the enclosure.

  9. ERROR_user_unknown says:

    Ok so here is my crazy idea for a better 3d experience. Using a wide formant curved “3D” capable screen. with the correct polarization on the left / right lenses. would give a much wider view than current screens almost to the point of having real peripheral vision. as you could see across the nose line where they currently have a divider.

  10. r3 says:

    2×5″720p ? Y U no use ipad screen instead ?

  11. nusuns says:

    So does the monitor go to blue screen if you pick up interference or start dropping signal? If so it will make a poor FPV setup. Hobbyking put out a DIY kit with a monitor that doesn’t go blue screen and it sold out in a day for that reason. They had to have the monitor made special just for that purpose.

    • Barry Carter says:

      In not sure I understand what you mean. Why is the blue significant here? So the screen goes blue when the signal drops. What was different about the hobby king screen?

      • Mojoe says:

        I believe it’s more like an analog screen. Instead of dropping the image all together and going blue, the picture quality degrades. So if your flying with it as an FPV rig and you see the picture degrading you can still see well enough to turn around and start flying back until the signal gets stronger.

  12. AnthonyD says:

    I made one very similar to the post.
    Here is the early https://www.youtube.com/watch?v=Si9E8-GmubI&feature=youtu.be

    The only issue I had was the yaw sensor (compass) would spike to a high negative value when passing due north. I was able to fix it in FreePie by canceling any delta yaw readings over 100 units.

  13. will1384 says:

    I made a DIY Rift using a lower resolution 7″ 1024×600 LCD monitor, seems to work well on older games like Quake and Unreal, but its way to low resolution for some of the newer games, have a look over at

    http://www.mtbs3d.com/phpBB/viewtopic.php?f=26&t=19607

    Here is a parts list

    Two 5X Pocket Loupe Magnifier lenses from eBay
    7″ 1024×600 LCD monitor, mini hdmi, model RX-S702 or TKH702S, the one I got has a aa0700023001 LCD
    Some 3D printed parts from http://www.mtbs3d.com/phpbb/viewtopic.php?f=26&t=18035
    Welding Helmet Headband and Mounting Hardware
    Aluminum Carpet Edging Trim
    Plexiglass
    Foam
    Hot glue
    Velcro cable ties
    1/4 bolts, washers and wing nuts
    3mm screws and lock nuts
    Scotch Thermal Laminator Pouches – to help with screen door affect
    STM32F3DISCOVERY board for head tracking, programed with instructions from http://yetifrisstlama.blogspot.com/2014/03/the-foculus-rift-part-2-reverse.html
    A EDID spoofer, using a 24c32 chip, and two 47k resistors, look here for the circuit https://lh3.googleusercontent.com/-oTLJAVwHzQc/U5KVnup2Q0I/AAAAAAAADoA/nPF_RJyoxMs/s800/DIY-HMD_028.png and here for the programmer http://www.riccibitti.com/quickdesigns.htm

    What you get is a cheap DIY Rift that looks to the software and demos like the real thing, and works well, but the resolution is way to low, I am planning to upgrade to a 7″ 1280×800 LCD, just wish there were more 3D printable designs for the 7″ LCDs, the only 3D printable design that I have seen for 7″ LCDs is from http://www.mtbs3d.com/phpbb/viewtopic.php?f=26&t=18035

  14. Fredz says:

    It’s not an Oculus Rift clone. The development kit uses a 7″ panel and 7X aspheric lenses, not a 5.6″ panel (was used only in the prototype) and 5X aspheric lenses (not enough magnification). It won’t be compatible with games and demos developed for the Oculus Rift, without content it’s pretty much worthless even if it’s cheaper.

    • Whatnot says:

      The drivers you use for oculus are general 3D thingies with various options to tweak depending what you use as display. The perception driver allows you to make game profiles with settings for FoV and yaw/roll/pitch multipliers and all such things for instance.

  15. Barnaby says:

    Another option is to use a android cell phone. They have extreemly high-resolution screens, just the right size, and good quality IMUs and enough processing power to use them with low lag. http://projects.ict.usc.edu/mxr/diy/fov2go/

    • Teddy says:

      I have the Virtual io original. the solution for me for fast head tracking was to use a gyro mouse, only have two axis but works in realtime with many older games and was/is a blast to play.

  16. David says:

    Would a 7″ 800×480 lcd display be good enough to do this? Cause i have one lying around not being used and if i could use it to make this i would go for it.

    • Fredz says:

      No 800×480 is far too low. I and others have tested with 1024×600 panels and it’s still too low, 1280×800 is really the minimum to get any sense of immersion.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 94,655 other followers