RifTUP! FullHD Oculus DK1 upgrade kit

Post by **zacherynuk** » Mon Aug 12, 2013 1:52 pm

You cannot run Oculus software on a 5.5 screen without modifying the software. This is why I have been pushing for a 7" replacement. HOWEVER, I do feel Oculus are overdue a technological sitrep.

A quick an easy way of upgrading the resolution, which I have already done (admittedly though, not with Oculus tracker) is two of the original vitro light displays in portrait spanned mode. But this only gives 33% higher horizontal resolution on what is arguably a worse panel anyway. (3 in portrait around you with a sheet of flexi fresnel makes a great 2D experience though!)

Frankly, if tweaking software is what you want to do, then DIY InfinitEye is the way to go, rather than 5.5 Rift. (http://www.youtube.com/watch?v=nH86II2CwdE - http://www.mtbs3d.com/phpbb/viewtopic.php?f=138&t=16434 ) This thing looks the NADS.

Indeed, using an iPAD screen and the mipi interface metioned in the DIY thread and some software tweaking could yeild you amazing things. I had a great experiance with an asus eee pc screen again in 2D.

In the last few years of being a member of this site I have had a fair few HMD's built and played with, scrapped, rebuilt, tweaked and discarded. The one thing that we were always missing was conformity to a spec, an open spec, which allowed you to actually use the damned things for more than a single proof of concept. Hence at the moment, any upgrades or rebuilds or tweaks will be built to work with the developers amazing creations for the rift.

ANYWAY

Today I talked to Asus, who wouldn't give up any info either.

""Thank you for contacting the Asus Support team. We are sorry to hear you are having difficulty with your Nexus 7 , Unforunetly we are not able to supply any information for repair and deconstruction of your tablet. This information is confidential to Asus. If you tablet requires to be reparied then you can apply for a repair with ourselves. Please contact back with further information on the fault itself."" (sic)

I've been here before, so I know that patience is the best way of keeping the wallet alive and sanity in check.

I wouldn't even know where to begin with LVDS or MIPI signals. I could just about manage to verify where the sync is on a component video feed, on a good day!

Anybody in the UK up to the task ? I have a shiny new Nexus 7.2 screen here.....

Post by **MSat** » Mon Aug 12, 2013 2:01 pm

LaserEdge wrote: Assuming the master to slave communication stays in low power mode you could probably get by with just probing the positive data lane. The voltage for a high bit on the positive lane will reach around 1 volt. You will of course be seeing only half the bits in low power mode as the negative pair carries the other bits. For high speed mode transmission the voltage swing is only 0.2 volts so it might not register at all with the logic analyzer. In short it is gamble if this short cut will work.

So in low-power mode the data0 pair operates as a 2-bit parallel bus? At .2v, I can see how working with high-speed mode might be tricky. I was hoping a scope (though I guess it would still have to be really fast) rather than a logic analyser would be sufficient to find telltale signs of which is clk, data0 and the remaining data pairs. Sure, being able to hook it up to a logic analyser would be great, but it won't do us any good if no one here has access to one.

Thanks for the MIPI spec sheet. I'm going to take a loot at it later.

As far as whether it's better to use the Nexus7 panel or this 5.5" LG unit is a matter of balance in terms of cost, and ease of replacement IMO. Yes, the Nexus7 offers the chance of a direct physical panel replacement without modification to the optics and housing, but is the most expensive by far (you have to buy a Nexus7 as of right now), you have to carefully remove it from the gorilla glass it's glued to, and you still need a custom controller even if it is LVDS (it looks like the Rift's controller board only has a single LVDS link). The 5.5" LG on the other hand is significantly cheaper, and you don't have to worry about breaking it while prying it from a glass face because there isn't one. However it does require different optics and at least some way of mounting it in the current Rift's enclosure (not sure if it would need to move closer to the lenses which would require further modification as there's an optical divider).

Or we can say "F it" and wait for the Rift-HD dev kit to be released soon as probably some of us are expecting.

Post by **LaserEdge** » Mon Aug 12, 2013 3:59 pm

MSat wrote: So in low-power mode the data0 pair operates as a 2-bit parallel bus? At .2v, I can see how working with high-speed mode might be tricky. I was hoping a scope (though I guess it would still have to be really fast) rather than a logic analyser would be sufficient to find telltale signs of which is clk, data0 and the remaining data pairs. Sure, being able to hook it up to a logic analyser would be great, but it won't do us any good if no one here has access to one.

Thanks for the MIPI spec sheet. I'm going to take a loot at it later.

As far as whether it's better to use the Nexus7 panel or this 5.5" LG unit is a matter of balance in terms of cost, and ease of replacement IMO. Yes, the Nexus7 offers the chance of a direct physical panel replacement without modification to the optics and housing, but is the most expensive by far (you have to buy a Nexus7 as of right now), you have to carefully remove it from the gorilla glass it's glued to, and you still need a custom controller even if it is LVDS (it looks like the Rift's controller board only has a single LVDS link). The 5.5" LG on the other hand is significantly cheaper, and you don't have to worry about breaking it while prying it from a glass face because there isn't one. However it does require different optics and at least some way of mounting it in the current Rift's enclosure (not sure if it would need to move closer to the lenses which would require further modification as there's an optical divider).

Or we can say "F it" and wait for the Rift-HD dev kit to be released soon as probably some of us are expecting.

The master still has to initiate a BTA ( bus turn-around) for the slave to transmit. It does so by issuing the sequence LP-11 / LP-10 / LP-00 / LP-10 / LP-00 on the Data 0 lane in low powered mode. After this sequence the master stops driving on Data 0 and the slave begins to drive it. It is usually used to receive a response after write. The slave has to use the same turn-around procedure to give Data 0 control back to the master.

If you have a logging DSO you could probably use it to capture the transmission and analyze it. An educated guess is a 500 MHz DSO should be sufficient. You will need at least 2 channels, so you can figure out which channel is which. Data 0 will always have to be logged so you can capture the packet header.

The issue with using the 5.5" panel in the dev kit housing and optics is that your going to get some cropping on the far left and far right edges. To solve that you need shorter focal length lens (higher magnification). Shorter focal length means you have to move the panel closer.

The best option might be the 5.9" panel coming out next month. I already have the data sheet for the LS059T1SX01. Based on the dimensions I believe you will be able to used the dev kit optics with it. This panel is identical to the LS050T1SX01 except for the required FPC connector and pinout is different. They both use the Renesas R63311 driver IC. If someone has adobe illustrator or similar program and dev kit you can create a quick test pattern with max rectangular dimensions of 72.9 mm X 129.6 mm to confirm if the dev kit optics will mate well with the LS059T1SX01. If the test pattern is in focus and you don't see the bounding edges of the test pattern the dev kit lenses will work.

Post by **3dvison** » Mon Aug 12, 2013 4:42 pm

So why is there no problem with my Diy Rift with a 5.6" screen and its also much better in that it shows less screen door than my 7 inch Dev.Kit.Rift.
Why would a 5.5inch 1080p screen using the real Rift tracker not be great and whats up with lenses being a big problem now also ?
I think there is a reason the HD Rift uses the 5.5 LG...

YES LaserEdge, that 5.9 would be the best I think..but I like now also..LOL...I forgot this was the 7inch screen thread...I need to stop posting 5.5, 5.9 stuff here.

Post by **Flim** » Mon Aug 12, 2013 5:47 pm

Ya, I don't get it... the lenses are wrong now for the 7 inch. That is why the original had a 5.6 inch lcd with the 1.5 inch lenses. Even if you get the 7 inch in there, you will not use half of the 1200 res.

Since the 7 inch is not a fast replacement, then in my opinion we should go about it as Oculus did.

Post by **LaserEdge** » Mon Aug 12, 2013 6:14 pm

The 5.5" might work. Only way to be sure is to create a test image and print it and out and test! You can use a program like Adobe Illustrator to create extremely accurate 1:1 printouts that match the dimensions of the panel in question. The active area of the 5.5" panel is 68.04mm X 120.96mm. Take the printout, open up the dev kit and place the printout where the LCD screen should be. You can add anything to the printout you think will be helpful. For example I like to create a cross pattern so I can test for geometric distortion as well as chromatic aberration. You will probably need a small LED light to illuminate the printout while you are doing visual inspection.

Post by **MSat** » Mon Aug 12, 2013 6:43 pm

Thanks for your excellent assessments, LaserEdge. The 5.9" sounds like it would be great, but what's the chance of being able to buy it in small quantities when it's released?

If the timing controller is the same between the 5.5, 5.9, and possibly the panel in the Nexus7, would it be feasible to trace out the path of the signal lines between the ribbon cable connector and the TCon IC so that we can use what we know from one of the panels based on an available datasheet and apply it to another?

Post by **LaserEdge** » Mon Aug 12, 2013 8:14 pm

MSat wrote:Thanks for your excellent assessments, LaserEdge. The 5.9" sounds like it would be great, but what's the chance of being able to buy it in small quantities when it's released?

I imagine the 5.9" will have a much larger production run than the 5.5". The HTC One Max will be a smartphone application where as the Nexus 7" 2nd gen is a tablet application. In general smartphone volume is much higher than tablets. I would not be surprised to see JDI follow up with a 5.9" panel as well given that the HTC One Max is a smartphone. Hard to say when OEM panels will be available for either. It is based on when supply exceeds the demand of their tier 1 customers.

MSat wrote:If the timing controller is the same between the 5.5, 5.9, and possibly the panel in the Nexus 7, would it be feasible to trace out the path of the signal lines between the ribbon cable connector and the TCon IC so that we can use what we know from one of the panels based on an available datasheet and apply it to another?

In short no. Most likely JDI followed whatever was requested of them in terms of pinout for the 7" panel. Here is a brief summary of the data lane ordering of all the 1080P panels using the Renesas R63311.
Sharp 5" LS050T1SX01 D2P-N / D1P-N / CLKP-N / D0P-N / D3P-N
JDI 5" ACX450AKN-7 D2P-N / D1P-N / CLKP-N / D0P-N / D3P-N
LG 5" LH500WF1-SD01 D1P-N / D0P-N / D3N-P / CLKN-P / D2N-P
LG 5.5" LH550WF1-SD01 D2P-N / CLKP-N / D3P-N / D1P-N /D0P-N
Sharp 5.9" LS059T1SX01 D3_N-P / D0N-P / CLKN-P / D1N-P / D2N-P

Post by **Inscothen** » Tue Aug 13, 2013 2:09 am

7to55.png

I made this for simulating the devkit modded with 5.5" screen. 64mm IPD. View it on the Rift fullscreened. It's probably not exactly 1:1 dimensions but it should be real close. I redid the stereo/warp to match the smaller display. White is obviously no screen. Not as many pixels as the 5.5" screen though

If you don't like the reduced FOV you'll want a larger screen. I use my Rift with the housing "out" and the straps tight enough to hold the Rift on my face but not too tight. I find it acceptable for me. Anyone that likes to get close to the lenses won't like it I think.

Post by **em4mars** » Tue Aug 13, 2013 10:33 am

1.jpg

This is the only image I can find of the inside of the HD prototype - It looks to me like they have changed the location of the lenses when compared to the DK1 - they look much closer together. I am assuming that, if we have to move our lenses, we will have to account for this in the SDK. I also assume that with the lenses closer together oculus can keep a very similar FOV with the smaller screen.

I know nothing about the ability to change the SDK - is this even possible right now or would we need help from Oculus? I think our assumption right now should be that Oculus is busy with their own development and do not want to spend time on software modifications for DIY'ers.

So, if we are going to drop a new HD screen into our dev kits and make it work with the current optics, do we need a 7" screen? If so thenI think we should gear the mod to the first available 7" screen that has a high enough res to make it worthwhile to change.

Just my not-to-technical thoughts,

Marshall

Post by **Flim** » Tue Aug 13, 2013 10:51 am

The goggles are the same, they only changed the attachment that holds the lcd. If they did move the lenses, then probably it was custom 3d printed eyecups, because the mounting is the same. I assume the difference is you see more screen compared to the 7 incher.

Post by **em4mars** » Tue Aug 13, 2013 12:23 pm

@Flim - In my dev kit the eye cups are individual, in the photo above the lenses are both attached to one holder that has less space between them. I do not claim to know what this means for the possibility of upgrading our dev kits but just wanted to point it out to those of you deciding whether the upgrade should be to a 5.2" screen or a 7" screen.

I am eagerly awaiting any solution that will allow me to use the Rift SDK with my dev kit in HD! I am willing to cut into any part and make a mod so long as it is proven to work - any happy guinea pigs out there?

Marshall

Post by **Inscothen** » Tue Aug 13, 2013 12:56 pm

em4mars, I don't see any lenses in that pic. Looks just like my devkit with the eyecups taken out.

Post by **Endemo** » Tue Aug 13, 2013 1:14 pm

Inscothen wrote:em4mars, I don't see any lenses in that pic. Looks just like my devkit with the eyecups taken out.

There are lenses. Just checked. Looks as if they just glued the new, bigger in diameter lenses just into the holes and left the eyecups out.

Post by **Inscothen** » Tue Aug 13, 2013 1:27 pm

Doesn't look like it to me

http://www.dvice.com/2013-6-13/oculus-r ... queasyness

http://www.oculusvr.com/blog/oculus-at-e3-2013/

Post by **geekmaster** » Tue Aug 13, 2013 1:39 pm

em4mars wrote:

This is the only image I can find of the inside of the HD prototype - It looks to me like they have changed the location of the lenses when compared to the DK1 - they look much closer together. I am assuming that, if we have to move our lenses, we will have to account for this in the SDK. I also assume that with the lenses closer together oculus can keep a very similar FOV with the smaller screen.

I know nothing about the ability to change the SDK - is this even possible right now or would we need help from Oculus? I think our assumption right now should be that Oculus is busy with their own development and do not want to spend time on software modifications for DIY'ers.

So, if we are going to drop a new HD screen into our dev kits and make it work with the current optics, do we need a 7" screen? If so thenI think we should gear the mod to the first available 7" screen that has a high enough res to make it worthwhile to change.

Just my not-to-technical thoughts,

Marshall

Those images are obviously with eyecups removed. That is a photo of the actual LCD panel visible through the eyecup holes. You can see the pre-warp, not "undistorted" by lens pincushion distortion. No glare or lack of focus from missing the lens "sweet spots" either. Just a direct view of the LCD panel through the holes. Nothing exotic here...

The "location of the lenses" would be in the eyecups that may perhaps be stored in the Rift carrying case?

Post by **Flim** » Tue Aug 13, 2013 1:59 pm

All of that looks the same to me... just different 3d printed lcd housing. Easy enough to replicate.

Post by **zacherynuk** » Tue Aug 13, 2013 2:24 pm

Yes all very simple of course.

If you tweak the software to remove the partial overlap - like the original rift and all the 5.x DIY ones.... If you tweak the software you can have stereo display on any freakin screen you want.

Not having partial overlap is the primary reason the 5.x screens have a lower FOV

Post by **em4mars** » Tue Aug 13, 2013 3:20 pm

Sorry guys, I thought the image I found had the lenses inserted into a single housing - my bad.

So, I have a dev kit as someone who wants to make the thing work outside of gaming for architectural visualization but have very little experience with the technicalities of the SDK. As such I have what may be a silly question but here it goes:

Is the SDK code easily modifiable? I ask because I think the benefit of a DIY HD Rift will be much more attainable if we can use it with games/software that have native rift implementation.

Sorry if I am covering ground that has already been tread. I am trying to understand the potential upgrade path. If someone such as myself will be able to download a revised version of the SDK for use with the HD DIY Rift then I think any size screen in HD is preferable to the current dev kit even if some FOV is compromised. If the SDK can be tweaked I agree with others who have voted for the use of a screen that is available now. Maybe we can have working HD Rifts by year's end!! THAT would be fantastic, even at a reduced FOV.

Thanks,

Marshall

Post by **KBK** » Tue Aug 13, 2013 8:22 pm

maxOh wrote:I did get a feedback about the HDMI/DVI->MIPI board. Sounds quite promising. Here the facts:
HBTM5(MIPI Bridge)_1.pdf
- the board will cost 150 USD
- Minimum quantity seams to be 100(http://www.alibaba.com/product-free/144 ... HBTM5.html), on buykorea it was 10, so let's see how to get a board

Here, Hold my beer. I'll do it.

(Sent them an email, directly off the hybus website, with 'official' company(s) connections explicitly laid out)

Post by **LaserEdge** » Tue Aug 13, 2013 11:47 pm

Suppose to receive delivery of the HDMI->MIPI PCB Wednesday. I will work on assembly, programming and testing as I have time over the next few days. If all goes well will have it working this weekend. I am starting with the LS050T1SX01. Will work on LH550WF1-SD01 after I get that working.

Post by **silentbrain** » Wed Aug 14, 2013 2:02 am

Hi, I make some diagram for panel size and MIPI lcd.

And I called HBTM5 Bridge board engineer and received positive answer for general customer. As you know, HBTM5 is work well for 2 panel model(LH500WF1-SD01, LS059T1SX01). And can support another lcd panel by S/W modyfing in the future.

Now, They need more time about a month for customizing and mass-production.

Post by **zacherynuk** » Wed Aug 14, 2013 6:39 am

silentbrain wrote:Hi, I make some diagram for panel size and MIPI lcd.

Great diagrams! They really show how a narrower (16:9) aspect ration makes better use of the screen.

Post by **jquantum** » Wed Aug 14, 2013 10:35 am

Hey, been lurking around (found the forum recently) and just had to register to give some words of encouragement. I'm excited to see this get some traction

Post by **MSat** » Wed Aug 14, 2013 1:15 pm

Great first post, and excellent diagrams, silentbrain! I know others were mentioning modifications on the software-side but I couldn't see it, but you post cleared that up. Thanks!

Needless to say, this makes compatibility a big issue. There's little sense in modifying a Rift this way if we can't take advantage of any content made for it.

Post by **Inscothen** » Wed Aug 14, 2013 2:12 pm

If we can change the stored parameters(resolution, display dimensions, lens distortion, lens separation, eye to screen distance) in the controlbox(unless it's in the tracker) we can mod our devkits with new screens and/or lenses and maybe not have to worry about compatibility anymore. If the info is stored in the controlbox we would need to still have that plugged in. If it's on the tracker, that will be plugged in anyways.

Post by **zacherynuk** » Wed Aug 14, 2013 2:33 pm

Inscothen wrote:If we can change the stored parameters(resolution, display dimensions, lens distortion, lens separation, eye to screen distance) in the controlbox(unless it's in the tracker) we can mod our devkits with new screens and/or lenses and maybe not have to worry about compatibility anymore. If the info is stored in the controlbox we would need to still have that plugged in. If it's on the tracker, that will be plugged in anyways.

It's all in the SDK

Post by **Inscothen** » Wed Aug 14, 2013 2:50 pm

cybereality said(https://developer.oculusvr.com/forums/v ... f=44&t=253) the hardware stores info and the SDK reads that info to automatically adjust stereo separation, and warp correction. Doing it that way means the SDK can be used for any version of the Rift. A game made for the devkit will be usable for the consumer version and vice versa. If you go into the SDK you can see how if identifies the screen dimensions, resolution, lens separation, and eye to screen distance. IPD is user configurable but all the other info is automatic.

The reason for this is because the stereo/warping algorithms pull parameters from the headset in order to work

Well the distortion constants are stored in the headset, which as used to feed into the warp shader. There are also things like IPD, and other eye related parameters that are used for the stereo.

The idea was that storing parameters on the Rift would help with compatibility, since we would like games created with the developer kits to work on the consumer version.

Post by **MSat** » Wed Aug 14, 2013 4:41 pm

Inscothen wrote:cybereality said(https://developer.oculusvr.com/forums/v ... f=44&t=253) the hardware stores info and the SDK reads that info to automatically adjust stereo separation, and warp correction. Doing it that way means the SDK can be used for any version of the Rift. A game made for the devkit will be usable for the consumer version and vice versa. If you go into the SDK you can see how if identifies the screen dimensions, resolution, lens separation, and eye to screen distance. IPD is user configurable but all the other info is automatic.

The reason for this is because the stereo/warping algorithms pull parameters from the headset in order to work

Well the distortion constants are stored in the headset, which as used to feed into the warp shader. There are also things like IPD, and other eye related parameters that are used for the stereo.

The idea was that storing parameters on the Rift would help with compatibility, since we would like games created with the developer kits to work on the consumer version.

If that's the case, then there might be hope. So as you point out, we need to find out if that's part of the EDID data, or something stored in the tracker. Should be able to find out via the SDK.

Post by **LuckyDog** » Wed Aug 14, 2013 4:53 pm

MSat wrote:
Inscothen wrote:cybereality said(https://developer.oculusvr.com/forums/v ... f=44&t=253) the hardware stores info and the SDK reads that info to automatically adjust stereo separation, and warp correction. Doing it that way means the SDK can be used for any version of the Rift. A game made for the devkit will be usable for the consumer version and vice versa. If you go into the SDK you can see how if identifies the screen dimensions, resolution, lens separation, and eye to screen distance. IPD is user configurable but all the other info is automatic.

The reason for this is because the stereo/warping algorithms pull parameters from the headset in order to work

Well the distortion constants are stored in the headset, which as used to feed into the warp shader. There are also things like IPD, and other eye related parameters that are used for the stereo.

The idea was that storing parameters on the Rift would help with compatibility, since we would like games created with the developer kits to work on the consumer version.
If that's the case, then there might be hope. So as you point out, we need to find out if that's part of the EDID data, or something stored in the tracker. Should be able to find out via the SDK.

The tracker is merely an accelerometer. It doesn't store data. It's most probable the info is part of the EDID.

Post by **MSat** » Wed Aug 14, 2013 6:05 pm

LuckyDog wrote: The tracker is merely an accelerometer. It doesn't store data. It's most probable the info is part of the EDID.

I wouldn't say it's merely an accelerometer. There is a microcontroller on board which might store that information.

Post by **geekmaster** » Wed Aug 14, 2013 8:57 pm

A limited amount of information can be stored in the Tracker using a writable USB HID report. It was reported that this may possibly used to store some limited amount of configuration data in the Rift controller at some point.

Post by **KBK** » Wed Aug 14, 2013 9:36 pm

More than likely it is holding some very specific custom filter parameters. (That's what I'd do, or propose, in my feeble state of knowledge, at this juncture)

Post by **Inscothen** » Thu Aug 15, 2013 12:09 am

I don't know where the info is stored, but OzOnE2k said it could be stored on the tracker's MCU. He said the controlbox's PCB appeared(from a glance at the teardown pics I think he said) to just pass the tracker's USB through.

Post by **zacherynuk** » Thu Aug 15, 2013 1:22 am

I stand corrected then - I was under the impression that the constants required were coded into the SDK configuration and the SDK only checked to see if the edid existed - I'll have to have another look. But yeah, if warp constants (if that's possible) and more importantly IPD is stored in the headset then... well just need to ensure the software uses that data!

Post by **silentbrain** » Thu Aug 15, 2013 2:02 am

I think We need 2 things in SDK or Tridef 3D, Real-IPD & Zoom control.
If 2 options available, Oculus can support various LCD size & ratio not only 7"

Post by **Inscothen** » Thu Aug 15, 2013 2:41 am

I'm not quite following. Oculus already supports different size LCD's and lens separation. Do you mean modded devkits? That's just a matter of using custom SDK implementations or changing the stored parameters on the Rift itself.

Post by **zacherynuk** » Thu Aug 15, 2013 3:01 am

Removing the panel from the Nexus 7.2 ...

http://www.wastedspace.co.uk/cms/2013/0 ... ut-weight/

Never mind.

Post by **Flim** » Thu Aug 15, 2013 6:49 am

That sucks.... but good to know it is tough! I still have mine in one piece...

So, how about that 5.5 LG?

Post by **KBK** » Thu Aug 15, 2013 8:10 am

I don't think I'd have a problem getting one out. Just be careful. What would be necessary, is a few more photos of the operation. Knowing exactly where the glue is on the faceplate.

For the amount of sacrifice done by this person, they could have taken more images. An image of the faceplate and the glue areas, would have been good. With a dead and disassembled unit at hand.. this should be no problem.