DIY Oculus Rift - schematics, instructions and build pics!

Post by **android78** » Sun Sep 02, 2012 5:44 pm

Chriky wrote:Well I think I win the competition for worst looking Rift knockoff, but probably also the cheapest (screen+driver board was £60 delivered and lenses were £3 delivered).

I found out I have quite narrow IPD (around 60mm) which means I can't "join up" the stereoscopy if I render straight side-by-side. The screen is a bit larger than the Rift's (7in vs 5.6in) which also explains why Rift can have 100% overlap and I can't I guess.

The lens are 40mm and claim to be 10x magnification.
...

That is HOT!!! It's like porn for geeks; as naked as tech can get!

Post by **Dycus** » Sun Sep 02, 2012 6:15 pm

I was bored a couple days ago, so I decided to see how far I could trim one of these driver boards.

I could take off another 3/4" still if I wanted to rewire that little 8-pin IC and the regulators... but that's not my idea of a good time.

If you know what you're doing, though, this isn't too hard.

Post by **brantlew** » Sun Sep 02, 2012 11:39 pm

FYI on trackers. I've been playing with the Sparkfun Razor this weekend and I definitely think it's a better choice than the Hillcrest for these projects. The Hillcrest does work right out-of-the-box and the installed firmware does have a higher data rate at 120Hz. But the tracking quality of the default firmware is glitchy, noisy, and drifts a lot.

The Sparkfun requires a lot more work up front to assemble, load the firmware, and calibrate. But once that is done the tracking accuracy is very good, is glitch-free, and the drift is negligible. The AHRS firmware is locked at a 50Hz data rate (20ms latency) which sounds a little slow, but the tracking is smooth and accurate and no additional processing has to be done. The Hillcrest data on the other hand has to be smoothed. After which, the latency is equivalent between the trackers.

I have a very demanding VR setup and have tried 4 different head tracking solutions and the Sparkfun+AHRS firmware is easily the best experience that I have had of the four.

Post by **blazespinnaker** » Mon Sep 03, 2012 1:09 am

Woah, dycus, that's so cool.

I ordered two solidoodles yesterday, 8 loops, and 4 kg of filament from repraper Going to set up a mini factory and start selling these things ( just for parts and shipping, oculus all the way

online.

Will post on thingiverse so everyone can fork the case design.

Viva la revolution..

Post by **kunteper** » Mon Sep 03, 2012 7:45 am

hi, came by this thread through hackaday. great build and great info ...

I am curious about the efficiency of the single lens you use. usually they use triplets or 2+ lenses stacked to take care of chromatic aberration, do you see any noticeable aberration due to single lens ?

I have built a DIY projector with a 17" display and was thinking of going with a smaller display since any optical misalignment is very noticeable and the light source needs to be quite powerful to illuminate such a big display. so the 5.6" display you use is very suitable for projectors too. the spec sheet says 30ms response. is 30ms good for dynamic scenes ? I havent read anyone mention any negative comments about the response time but I thought I'd ask in case...

Post by **Marcel** » Mon Sep 03, 2012 8:08 am

kunteper wrote:I am curious about the efficiency of the single lens you use. usually they use triplets or 2+ lenses stacked to take care of chromatic aberration, do you see any noticeable aberration due to single lens ?

The Chromatic Abberation could be fixed by adding the 'opposite' abberation to the footage going to the Rift (just like the distortion is corrected by pre-distorting the image). With the low resolution of the dev kit it is probably not very effective, but with higher resolutions this should work very well.

I also think that the Chromatic abberation is probably not so much of an issue, it occurs mostly in the corners and your eyes will be looking mostly at the center. I hope

Post by **Jotschi** » Mon Sep 03, 2012 9:03 am

The Hackaday post can be found here:
http://hackaday.com/2012/09/01/diy-oculus-rift-vr/

Post by **rfurlan** » Mon Sep 03, 2012 11:08 pm

kunteper wrote:hi, came by this thread through hackaday. great build and great info. I am curious about the efficiency of the single lens you use. usually they use triplets or 2+ lenses stacked to take care of chromatic aberration, do you see any noticeable aberration due to single lens ?

Hi Kunteper, the very first prototype I built used two stacked lenses to correct the image (http://bit.ly/Uopm6Q), the image looked great but it was also bulkier and heavier. In the end the distortion isn't that bad because it mostly affects your peripheral vision and I don't think it is worth the trouble of correcting it - at least not at cost of added complexity and weight.

Post by **rfurlan** » Mon Sep 03, 2012 11:16 pm

Jotschi wrote:The Hackaday post can be found here:
http://hackaday.com/2012/09/01/diy-oculus-rift-vr/

Really cool, more the merrier

Post by **Dycus** » Mon Sep 03, 2012 11:58 pm

The distortion is actually a good thing; it puts more pixels in the center of our vision. This XKCD nicely illustrates our field of view: http://xkcd.com/1080/

Really, our peripheral vision is useless for reading or anything else that needs detail. I think I would actually want lenses with more distortion.

Post by **MaterialDefender** » Tue Sep 04, 2012 1:18 am

Marcel wrote:The Chromatic Abberation could be fixed by adding the 'opposite' abberation to the footage going to the Rift (just like the distortion is corrected by pre-distorting the image).

This is most likely possible, but it's more complicated than correcting the distortion. Chromatic aberrations spread the light like a prism does. With some simple rgb-channel pixel shifting (which with a few caveats is sufficient for distortion correction) it is only possible to counteract the color shift to some degree (not perfectly though), but the introduced blur still remains.

Post by **Marcel** » Tue Sep 04, 2012 1:38 am

MaterialDefender wrote:
Marcel wrote:The Chromatic Abberation could be fixed by adding the 'opposite' abberation to the footage going to the Rift (just like the distortion is corrected by pre-distorting the image).
This is most likely possible, but it's more complicated than correcting the distortion. Chromatic aberrations spread the light like a prism does. With some simple rgb-channel pixel shifting (which with a few caveats is sufficient for distortion correction) it is only possible to counteract the color shift to some degree (not perfectly though), but the introduced blur still remains.

In theory it might be like a prism, but in practice the chromatic abberation is almost always in pairs (red/cyan, green/magenta or blue/yellow). In one of his videos Carmack says it is pretty easy to correct (but it is not a big deal image quality wise). We'll see when we have access to the Rift

Edit: I have ordered a set of the lenses used in the Rift prototype (http://bit.ly/MALA4X). Once I have them I will see if I can accurately photograph the distortion and chromatic abberation. (International shipping so it will take a while before they arrive I'm afraid.)

Post by **MaterialDefender** » Tue Sep 04, 2012 2:06 am

It's still more complicated than the distortion correction, believe me. I've done a distortion correction shader with basic aberration correction, and simply shifting rgb channels independently is clearly not enough. It's not impossible and most likely it's not even difficult in the sense of 'next to impossible' to do this right, but it's definitely more complex - even for John Carmack - than the distortion correction, which in fact is extremely easy.

Post by **Jotschi** » Tue Sep 04, 2012 2:13 am

MaterialDefender wrote:It's still more complicated than the distortion correction, believe me. I've done a distortion correction shader with basic aberration correction, and simply shifting rgb channels independently is not enough. It's not impossible and it's not even difficult in the sense of 'next to impossible', but it's more complex - even for John Carmack - than the distortion correction, which in fact is extremely easy.

Any chances of sharing the sources for the shader? Is that a GLSL shader?

Post by **MaterialDefender** » Tue Sep 04, 2012 2:19 am

I got both a HLSL an GLSL version. The shader will be availabe when the source for my Rift ready version of Descent will be available, meaning after the Rift is released. Who wants to do Copy&Paste programming anyway? Where's the fun with that? You will have no problems finding material about the (not so complicated) math behind this.

Post by **druidsbane** » Tue Sep 04, 2012 11:23 am

MaterialDefender wrote:I got both a HLSL an GLSL version. The shader will be availabe when the source for my Rift ready version of Descent will be available, meaning after the Rift is released. Who wants to do Copy&Paste programming anyway? Where's the fun with that? You will have no problems finding material about the (not so complicated) math behind this.

I like the DIY attitude here. I think that there will be tons of these shaders out once the Rift is released so I wouldn't worry about others getting access Jotschi, but now is the fun time when each developer can roll their own and exercise their brain muscles for the next couple of months

It should be very straight forward to integrate these shaders with code even the day the Rift comes out! It can't come out soon enough but at least I have my hands full writing code for it

Post by **space123321** » Tue Sep 04, 2012 2:04 pm

I have now intrigrated my hacked wiimote to have the motion sensor on the the top of hmd with the extended cable to the wiimote at my side. Using this in GlovePie with a basic mouse emulator script and an xbox wireless remote is soooooooooooo sweet! For the time being, I am using this method in place of the hydra.

Post by **brantlew** » Tue Sep 04, 2012 2:25 pm

Sweet. Very jealous.

Post by **cybereality** » Tue Sep 04, 2012 8:00 pm

Good stuff!

Post by **TheLookingGlass** » Tue Sep 04, 2012 10:45 pm

Great work space123321! I love seeing everyone's progress!

Post by **rfurlan** » Tue Sep 04, 2012 10:48 pm

So I just ordered these (for head-mounting):

http://www.amazon.com/gp/product/B002GO05LM

http://www.amazon.com/gp/product/B0041PXQ38

I will try both and add the most suitable one to the parts list so everyone can skip some trial and error. Does anyone have any other suggestions?

Post by **Jotschi** » Thu Sep 06, 2012 2:52 pm

The new Kindle Fire HD was announced. 8.9" display with 1920x1200 resolution.

Maybe they are using this panel:
http://www.panelook.com/HV089WU1-1E0_HY ... 15453.html

Post by **gravity360** » Thu Sep 06, 2012 3:00 pm

That panel wouldn't do any good unless it has a controller that can handle HDMI/VGA I/Os. The Sharp IGZO is a good possibility as it has a high PPI which matches that of Toshiba's 6.1" display that has 498ppi. You are looking at a massive 2560X1600 resolution on a smaller 6.1" panel. With it being a HMD, you wouldn't want anything bigger than 7" tops and that's actually going a bit out there.

Post by **rfurlan** » Thu Sep 06, 2012 3:02 pm

Jotschi wrote:The new Kindle Fire HD was announced. 8.9" display with 1920x1200 resolution.
Maybe they are using this panel:
http://www.panelook.com/HV089WU1-1E0_HY ... 15453.html

An upgrade to 1920x1200 would be nice even despite the increase in size and weight, also the current Rift panel can only display 256k colors which is far from ideal.

Does anyone know what kind of controller board is needed to drive a MIPI panel?

Post by **rfurlan** » Thu Sep 06, 2012 3:04 pm

gravity360 wrote:That panel wouldn't do any good unless it has a controller that can handle HDMI/VGA I/Os. The Sharp IGZO is a good possibility as it has a high PPI which matches that of Toshiba's 6.1" display that has 498ppi. You are looking at a massive 2560X1600 resolution on a smaller 6.1" panel. With it being a HMD, you wouldn't want anything bigger than 7" tops and that's actually going a bit out there.

Hi Gravity360, is this Sharp panel already available for purchase?

Post by **Chriky** » Thu Sep 06, 2012 5:14 pm

Anything more than 6" isn't that useful. You have to render the centre of the images ~6.5cm apart, which means the edges of a screen >13cm aren't really used.

Post by **Fredz** » Thu Sep 06, 2012 5:43 pm

You're not forced to have the optical axis perpendicular to the screen, I think Palmer said that it wasn't the case for the Rift actually.

You can also use a larger display to cover more than the 90° FOV of the current Rift. Since you're limited to around 40 to 60° nasally, you need to use a larger screen to get more FOV outwards.

The problem is essentially a question of practicality I'd say, size and weight as well as ways of keeping the thing firmly attached to the head.

Post by **Dycus** » Thu Sep 06, 2012 6:51 pm

A wider screen would actually be better. With the prototypes we were showing at conventions and stuff, you could just barely see the left and right edges of the 6" screen if you looked. You couldn't see the top and bottom ones, though.

And, of course, if you want a larger, POV, you're gonna need a bigger screen too.

Post by **Krenzo** » Thu Sep 06, 2012 7:28 pm

rfurlan wrote:Does anyone know what kind of controller board is needed to drive a MIPI panel?

It says the specs are only available to members of the MIPI alliance, but it sounds almost exactly like HDMI (4 lanes = 1 clock lane + RGB lanes). Is anyone able to download the datasheet?

Post by **space123321** » Thu Sep 06, 2012 8:01 pm

I must say - I have been viewing various sbs videos on my DIY Rift prototype, and I am still in awe with the 3d effect that takes place - even with the warping present and the 'squished' images that are being used. Nothing that I have used in the past has come close to the 3d effect that I witness (older HMD's, 3DTV's, personal 3d theater, etc) - even compared to my LG passive TV's - which to my eyes have the best 3d avaliable (owned several active sets and eventually fell in love with the LG passives - however that is another story all together).

Now.... being a previous owner of the Sony HMZ, the 3d effect was there, however the picture was so natural that it almost wasn't there... confusing I know - however the 3d effect was very minimal compared to my LG TV's. Additional comments from individuals stated that due to our eyes not having any visual cues to reference - the 3d was more gentle on the eyes/brain... and I agreed as this was usually the case with HMD's that I had used in the past. Well I can proudly say that this is clearly not the case with the DIY rift... I guess with the picture being so large, immersive and in your face - the visual cues are no longer needed and in return - the 3d effect is outstanding! I just can't get enough lol!!!

Post by **rfurlan** » Thu Sep 06, 2012 8:05 pm

space123321 wrote:....the 3d effect is outstanding! I just can't get enough lol!!!

I am afraid that is called 3D madness and I am a sufferer too...

Post by **Fredz** » Thu Sep 06, 2012 8:08 pm

space123321 wrote:I have now intrigrated my hacked wiimote to have the motion sensor on the the top of hmd with the extended cable to the wiimote at my side.

It seems like a very good and cheap candidate for head tracking with the Rift, could you describe in more details what you've done ? Also, how was the latency ? I've read that it should be around 50ms with a Wiimote but I don't know what this could mean in terms of uncomfort.

Post by **brantlew** » Thu Sep 06, 2012 8:14 pm

@space123321: Same story. My HMD has a muted effect compared to my LG passive which I thought had a great 3D effect. But after looking through a Rift, all other 3D looks gimmicky in comparison. The effect is so profoundly natural you feel the urge to start grabbing things in front of you.

Post by **space123321** » Thu Sep 06, 2012 8:28 pm

Fredz wrote:
space123321 wrote:I have now intrigrated my hacked wiimote to have the motion sensor on the the top of hmd with the extended cable to the wiimote at my side.
It seems like a very good and cheap candidate for head tracking with the Rift, could you describe in more details what you've done ? Also, how was the latency ? I've read that it should be around 50ms with a Wiimote but I don't know what this could mean in terms of uncomfort.

I find the latency fine for testing and simply playing around with - not perfect but it works! When using the xbox controller, if any drifting occurs (or disorientation) - I simply use the right stick to gain back my required view.

I took the efforts from pierreye - seach wiimote cable and you should find the thread. Basically we had older wii motion plus adapters that connected to the first generation wiimotes. A cable was added in between (as well as a small soldering job) to allow for the wiimotion plus adapter to be mounted and the wiimote to be extended to the users side.

Post by **kunteper** » Sat Sep 08, 2012 5:17 am

here is a link for a lens manufacturer in china where I've bought DIY projector related stuf before. Infact I am about to order a new set of lenses for a projector from them that match the 5.6" display. Thought maybe some one might find it usefull. http://www.optolife.com

Post by **gravity360** » Sun Sep 09, 2012 1:17 am

rfurlan wrote:
gravity360 wrote:That panel wouldn't do any good unless it has a controller that can handle HDMI/VGA I/Os. The Sharp IGZO is a good possibility as it has a high PPI which matches that of Toshiba's 6.1" display that has 498ppi. You are looking at a massive 2560X1600 resolution on a smaller 6.1" panel. With it being a HMD, you wouldn't want anything bigger than 7" tops and that's actually going a bit out there.
Hi Gravity360, is this Sharp panel already available for purchase?

I am not sure on this. We would need to check with sharp. I think because of the type of display there may not be a proper controller for it to work like the current rift panels.

Post by **ak0** » Sun Sep 09, 2012 7:27 am

old auction with 5x loupes has ended, new one is
http://www.ebay.com/itm/5x-pocket-loupe ... 3a79987a19

Post by **rfurlan** » Sun Sep 09, 2012 10:35 am

ak0 wrote:old auction with 5x loupes has ended, new one is
http://www.ebay.com/itm/5x-pocket-loupe ... 3a79987a19

Thanks Ak0, I updated the parts list to point to the new auction

Post by **MaterialDefender** » Sun Sep 09, 2012 12:10 pm

space123321 wrote:...the 3d effect is outstanding! I just can't get enough lol!!!

Could someone please close this thread.

The more I read here the more I feel the urge to go shopping at ebay, despite having made a pledge for two dev kits...

Post by **space123321** » Sun Sep 09, 2012 12:16 pm

LOL - sorry! I am now onto version 3 of my HMD - will now be a more permanent solution getting away from the foam core sheets.