RifTUP! FullHD Oculus DK1 upgrade kit

[OzOnE2k10]

Excellent!

Always good to see a batch of shiny new PCBs.

I've been reading about the Gobe thing on Indiegogo. The comments are quite worrying.
I hope for everyone's sake that Indiegogo sticks around for the long-term.

@cgp44 - was that you stirring it up on the Oc forum? hehe
Do you get your 50% discount now? What on Earth is a "Claire FullHD"? :p

https://developer.oculusvr.com/forums/v ... UP#p102334

btw, does the DK2 panel really have a diamond pixel arrangement? Can somebody clarify the DK2 pixel type?
Not that it matters too much for me personally, but how would this look compared to the RGBG arrangement shown on the RiftUP! campaign?


Anyway, great update about the RiftUP! kits. Very happy to see the progress so far.
It will certainly be a lot quicker than the desperate wait for my DK1 to arrive.

OzOnE.

[rozsnyo]

btw, does the DK2 panel really have a diamond pixel arrangement? Can somebody clarify the DK2 pixel type?
Not that it matters too much for me personally, but how would this look compared to the RGBG arrangement shown on the RiftUP! campaign?

It will look same, except being rotated in 45 degrees. Given that there is the same number of pixels in both "arrangements", such a rotation makes no real improvement. The difference will be in perception of such an array - while a standard low-res screen would be more like a specific mosaic... the rotated screen would look like a blurred, detail-less image.

If you dig out a really old digital still camera, where the viewfinder was like 100k pixels or less, they used this rotation to put the pixels in a "triangle" form, to reduce the effect of having square pixels which would show the preview as a mosaic... and instead of that, you get some more natural image. While this applies very well on photographic use (natural pictures), I doubt it can have better performance than a full RGB array. The image details wont be there, especially the color ones.

Look at something like this http://www.dansdata.com/images/c2100uz/ ... der320.jpg (cant find a true macro shot of such a screen, they are quite old.... and now they are returning!)

[OzOnE2k10]

It will look same, except being rotated in 45 degrees.

oic, thanks.
That's what I was thinking, 'cos it's pretty much a "diamond" shape already.

I remember the effect on older cameras and viewfinders.
It made sense on those, but more important now is the total resolution as you say.

Some people are definitely over-reacting to the RiftUP! on the Oc forum tbh.

Yes, the DK2 will be nice as well, but I've personally been waiting for a 1080p DK1 upgrade for many months.
So I'm more than happy to buy both.

I think people are forgetting that the DK1 will still be a great device once the RiftUP! is fitted, and should have good resale value if they so wished?

Just look at how much this DK1 sold for recently, and that was after the Facebook announcement...

http://www.ebay.co.uk/itm/Oculus-Rift-d ... 1295172032

With a 1080p panel, it would surely be worth a fair bit if people don't want their DK1 any more after their DK2 arrives?

OzOnE.

[cgp44]

What on Earth is a "Claire FullHD"?
OzOnE.

I was thinking it was a blowup doll!

[OzOnE2k10]

What on Earth is a "Claire FullHD"?
OzOnE.

I was thinking it was a blowup doll!

LOL

I hope so. I'd pre-order that.

OzOnE.

[OzOnE2k10]

I REALLY want to know what new stuff Rozsnyo is working on atm.

Can we have a hint at all?

Higher resolution? Low-persistance? Blow-up doll?

Also, what is the "closest" pixel structure to the DK2 display, does anyone know?
https://www.google.co.uk/search?q=penti ... =isch&sa=X

OzOnE.

[woodsmoke]

Just look at how much this DK1 sold for recently, and that was after the Facebook announcement...

http://www.ebay.co.uk/itm/Oculus-Rift-d ... 1295172032

Hmm, with new plastics and cable the RiftUP! is pretty much a new unit, besides the lenses of course. Almost tempted to sell the original DK1 with it's current markup as I make a new housing for the RiftUP!

..Bah, if only it wasn't so historically significant, I just might have. Perhaps it's only destined to go up in value as a collectors item anyway.

Can we have a hint at all?

rozsyno recently admitted he was a resolution junkie, perhaps that's a clue?

[OzOnE2k10]

I did contemplate that, woodsmoke.

Especially seeing that the Rift tracker has been pretty much cloned as well now.
It would just need a 3D-printed housing and the strap / goggle / lens parts.

I do want to keep my DK1 though, as it's now sure to be a pretty decent collectors' item since the FB thing.

If Roz is looking to use even higher res panels, I'm slightly worried that it will get to a point of diminishing returns due to the DK1's lack of positional tracking etc.

Of course, as @cgp44 said on the Oc thread, us VR guys will be happy to buy the new stuff regardless.

It's just that if a better panel will be much more expensive than the current RiftUP! kits, then maybe it would be more worthwhile Roz producing panel upgrades for the DK2 instead?

OzOnE.

[Kernel32]

Excellent!

Always good to see a batch of shiny new PCBs.

I've been reading about the Gobe thing on Indiegogo. The comments are quite worrying.
I hope for everyone's sake that Indiegogo sticks around for the long-term.

@cgp44 - was that you stirring it up on the Oc forum? hehe
Do you get your 50% discount now? What on Earth is a "Claire FullHD"? :p

https://developer.oculusvr.com/forums/v ... UP#p102334

...
OzOnE.

Indiegogo is a much opener platform than for example Kickstarter. However they should really act in case of such an obvious hocus.

I meant more about our hack Opener utility coming to open the SDK to any hardware news Still counts. You will get a hell of a deal.

[OzOnE2k10]

I meant more about our hack Opener utility coming to open the SDK to any hardware news Still counts. You will get a hell of a deal.

Cool.

It will be interesting to see how tweakable the different features are.
If you can tweak the FOV and warp settings on-the-fly, then that's even better.

Compatibility is the main thing though.
Sounds like it's working with many SDKs / demos / games?

OzOnE.

[Kernel32]

I meant more about our hack Opener utility coming to open the SDK to any hardware news Still counts. You will get a hell of a deal.

Cool.

It will be interesting to see how tweakable the different features are.
If you can tweak the FOV and warp settings on-the-fly, then that's even better.

Compatibility is the main thing though.
Sounds like it's working with many SDKs / demos / games?

OzOnE.

Yes you can.

We tested it with Oculus SDK Unity/Unreal games/demos so far. It will contact our servers in case there is some incompatible software and we will resolve and add profile to your hack libraries.

[ripcurl123]

One thing has occurred to me about the riftup is the 1080p screen gonna need more computing power compared to dk1 720p screen does anyone know?
or is it minimal

Ripcurl.

[mr.uu]

One thing has occurred to me about the riftup is the 1080p screen gonna need more computing power compared to dk1 720p screen does anyone know?
or is it minimal

Ripcurl.

Just run it with 720p (16:9 like 1920x1080 RiftUP!) to avoid more needed computing power. RiftDK had 800p (16:10).
But of course less AA or details settings ingame and FHD is a better choice...

[3dvison]

So if you don't have a Rift tracker will a hillcrest or YEI 3 space work ? Or is there a clone of the Rift Tracker you can buy now ?

[eidahl]

Sold my DK1 last week... so tempted to pick up one of the RiftUp Nows. :X

( Were screen latency and contrast levels discussed earlier in the thread? Checking.. )

[cgp44]

You only get a display, board to feed the display, and a board that goes into the DK1 controller box.
So no head mask, light box, lenses, straps.

[rozsnyo]

I would love to see a box of 120 sharp 5.9ers

They have arrived today!

If you place the 120 screens in a matrix of 12x10, you get a total resolution of 23040 x 10800 ... which is almost 250 MPixels. We could show those fancy "gigapixel" class photos in native resolution And this matrix system would be only 76 inch diagonal. Alternatively you can make a 8K screen of 24" size from 16 displays.

[Hornet]

Really nice view - pěkné

[OzOnE2k10]

I thought it was a given that a higher resolution would require a bit more GPU power tbh?

Modern graphics cards do a very good job of scaling to higher resolutions without the massive impact to frame rate that older cards did, but it will certainly need a fair bit of power to do true 3D stereo, low-latency rendering, HLSL warping / aberration correction.

I bought a GTX 670 for my "Rift" PC, and it seems pretty good so far. It's not the newest card of course, but should be plenty for now.

btw, when you say "run it at 720p", I'm fairly sure the RiftUP! kits won't scale other resolutions to the 1080p panel?
So, you'll have to enable GPU scaling in your graphics card options if you want to run in a res other than 1080p.

@rozsnyo - could you possibly confirm that GPU Scaling needs to be enabled for displaying lower / higher resolutions on the RiftUp! kit?

(great to see the new panels arrive btw. )


@3dvision - I'm not sure about the Hillcrest and other 3rd-party trackers, as I've never played around with them nor the software.
But, there was some fantastic work done by the Yeti dude for "cloning" the Oculus Tracker and protocol...

http://yetifrisstlama.blogspot.co.uk/20 ... verse.html

If this could be refined somewhat by using the exact same sensors as the Oc tracker, and a new custom PCB, then it should in theory perfrom just as well as the DK1.
The Oc tracker doesn't appear to do any major calcs on the data itself, it seems to send the raw data to the PC for the sensor fusion to take place.

Making our own tracker would also make it easier to modify the panel and warp coef data for different panels.


OzOnE.

[cgp44]

beautiful. That and the blowup doll. Things are happening in Prague.

[OzOnE2k10]

beautiful. That and the blowup doll. Things are happening in Prague.

Seriously, I really want to know what a "Claire HD" is now. lol

Is it a blow-up doll with "High Definition" features?

OzOnE.

[cgp44]

If four were placed vertically side by side, and two horizontally on the bottom
you would get a 4K x 3K screen. Now this arrangement would be on a curved
line in front of your face WITHOUT lenses. The screens would be at a distance
where you focus at infinity, so you have no eye strain.

The seams between screens would be obvious, so the effect would be as if you were
in a caged cockpit.

What in the hdmi outputting modes could handle this. Maybe we would need
to have six separate cables or can we go with 2x 2kx2k and 2x fHD i.e., four
cables. I see the latest hdmi standard has 4K at 60fps, but that is for future GPU cards.

[KBK]

Since this is mipi, I'm not all that adverse to taking apart a nexus 7 gen II and pulling the 16:10 7" 1920x1200 panel out of it,and dropping that into my rift.

Only problem is the software match is not there, it's done quite logically, for what appears to be a 5.9"..Sony (JDI) panel?

Sony, Toshiba, and Hitachi. And Hitachi is no slouch.

[cgp44]

@KBK,

you will need to wait for rozsnyo to offer a small pcb addon that offers various connectors.
Otherwise you have to get the connector and do a smd on board large pinnouted thingy
replacement. You would need fancy heat guns for smd rework.

Same thing for putting hdmi connectors on the riftUP main board. But I like the idea of the
secondary board that goes in the Rift controller box for its own sake.

As well we cannot reprogram the fpga/rom on the RiftUP!, get the other board.
EDIT: Opps , wrong here. You can reprogram the board. No fpga, but an MCU (microcontroller unit)

[cgp44]

One of the important questions I have regarding hdmi and EDID (DDC) is:

can we just put detailed timing figures way off standard video timing in EDID
(ie all those multi-paneled setups) and you should expect the GPU to do it.

The new hdmi 4k standard has 4K @60fps on the same high speed cables today.
There is no tmds signalling changes (low level physical signalling).

[cgp44]

Alternative to a fast signal on a few hdmi cables is standard signalling on many cables.
Now these cables would be special custom cables where the four tmds differential signal
wire pairs are on the thinnest wires possible. I cut up hdmi and vga cables and you could
use say three wires rather than the seven in each wire. Wrap them with foil somehow every
tightly and get a super-thin hdmi cable. Cluster say six of these together.

[OzOnE2k10]

@cgp44 - The EDID thing will depend on a few things...

Will your GPU support the high resolutions?
Will the HDMI / DVI port(s) handle it?
Will the current drivers allow your custom mode, or does it have certain limits?

Assuming all of that, you'd have to design the board around the 4K (or higher) panels you'd want to use.

I don't think the RiftUP! boards have an FPGA on them as such.
AFAIK, they are based around the Solomon SSD2828 chip that we were looking at in the Nexus thread.

The new controller board has an HDMI / DVI switcher / cable driver chip in the controller + a small MCU to configure / control that.
At the headset end, I believe it has the HDMI receiver chip, then the parallel RGB stream goes directly to the SSD chip.

(@rozsnyo / Kernel32, please correct me if I'm wrong on any of this.)

So, to use multiple DVI / HDMI channels and / or multiple panels would need a complete custom board with and FPGA or similar on it.

Looks like the max res for a single SSD2828 chip is 1080p @ 60Hz, or "2K" @ 30Hz...

http://www.solomon-systech.com/en/produ ... p/ssd2828/

I'm not sure how you would go about using multiple SSD chips for multiple channels tbh, but I'm guessing the 4K panels will be using 8 MIPI lanes or something like that?

Getting a GPU to output more than dual-channel TMDS using multiple outputs is not something I've looked into for many years either.
I'm sure it's possible to do all this stuff, but it could be a hell of a challenge to get multiple 4K panels working on a Franken-Rift.

OzOnE.

[rozsnyo]

I don't think the RiftUP! boards have an FPGA on them as such.
AFAIK, they are based around the Solomon SSD2828 chip that we were looking at in the Nexus thread.

The new controller board has an HDMI / DVI switcher / cable driver chip in the controller + a small MCU to configure / control that.
At the headset end, I believe it has the HDMI receiver chip, then the parallel RGB stream goes directly to the SSD chip.

There is no FPGA. The breakout board has a switcher and a I2C expander chip which handles the LED and buttons, there is no MCU.

The headset board has another switcher (not populated on DK1 upgrade kits) and then a HDMI receiver, SSD converter and a processor which configures and operates the whole system. Extra to that, there is a USB hub with one onboard device (FTDI based MCU programmer) and 3 open ports. These are also not populated on upgrade boards.

Regarding >FHD displays, they are "2 channel" MIPI types, that means 4+4 lanes. Such displays exist: 6" AUO from VIVO Xplay3S (2560x1440), then 8.9" Kindle HDX (2560x1600) and a 10" sharp screen (2560x1600). The first I do not have, but it is possible to get from china, but no docs. The second I have physically with me, but no docs. The third has documentation, but no interest in making driver for it (too big for VR).. maybe later for the 4K cam project we make something with it.

[OzOnE2k10]

Excellent, thanks for the reply.

So, the MCU is on the HMD board, that makes sense now.

I'm assuming the software / firmware for the standalone version will only be made available to people who purchased that version of the kit?
ie. if people were to solder the extra components for the programmer, would you be releasing the code as open-source?

(Obviously, it's understandable if this support will only be for the standalone kits.)

Also, would it still be possible to update the firmware on the upgrade kits for testing the backlight strobing, or are those parts not populated?

Sounds like the boards are very well designed though, with lots of possibilities for future features.


Yep, that 10" display sounds huge. Even the 8.9" display would be pushing things. lol

The 6" AUO sounds very good though.
I'll have a hunt around for any info I can get hold of on it.

OzOnE.

[rozsnyo]

Also, would it still be possible to update the firmware on the upgrade kits for testing the backlight strobing, or are those parts not populated?

Will be populated. And there are enough buttons to control the effect I am thinking to make the final firmware this way:

POWER = input switch
CONTRAST = strobing ratio
BRIGHTNESS = backlight intensity

[OzOnE2k10]

Oic, I forgot you can just map the buttons for different functions.

Good to know the duty cycle stuff will be included.
That button mapping makes good sense.

Does this mean that you're intending to add the basic strobing function to the upgrade kits as well before shipping them?
Can the firmware on the upgrade boards be easily updated at a later time if needed? (TTL serial pins etc?)

OzOnE.

[rozsnyo]

Does this mean that you're intending to add the basic strobing function to the upgrade kits as well before shipping them?
Can the firmware on the upgrade boards be easily updated at a later time if needed? (TTL serial pins etc?)

Yes, the functionality will be there, the upgrade is not easy - you would need an AVR ISP programmer (on software side: avrdude)

[OzOnE2k10]

OK, no problem.

I've got AVR programmers, so I may be able to help out with testing new firmware features etc.

Thanks,
OzOnE.

[geekmaster]

But, there was some fantastic work done by the Yeti dude for "cloning" the Oculus Tracker and protocol...

http://yetifrisstlama.blogspot.co.uk/20 ... verse.html

Why did he decide the "only way" to get rid of the anomalous magnetometer readings was to run it at 75Hz instead of 220Hz, when he could just discard statistical outliers like I do in my Rift+Hydra USB/HID drivers for my RasPi?

Likewise, discarding statistical outliers allows the Hydra to run faster without heavy filtering, with much more precision and much lower latency. Instead of heavy filtering of bad data points, or running it so slow there aren't any bad data points.

I MUCH prefer detecting and discarding the bad data to maximize accuracy and minimize speed. Strangely, I do not see others doing this in the various code I have read on the net.

That "foculus" code could be improved by implementing my suggestions. I have a lot of experience in the motion control industry with bypassing the speed v. accuracy speed ramp profile nonsense. It took many years before others began implementing my "magical" ideas (as others called this "nonstandard but superior" motion control method). I just applied a different (but similar outcome) idea to sensor data, and it works wonders, as I mentioned above. No more valid claims of the Hydra have accuracy and latency problems. That was just symptoms of a firmware bug and crappy software workaround. I tunneled through all that, just like in motion control decades ago...

I am not bad-mouthing anyone here. I truly appreciate bringing such products to market, and especially open-source sharing of code and ideas. Perhaps I am hard to impress when I look at the details within, wondering why suboptimal workarounds were used to "solve" a problem instead of recognizing the obvious solution sitting right there (or perhaps only obvious to an aspie). If I knew long ago that I was an aspie, I would have been less disappointed in code that used the "wrong" solution. http://www.youtube.com/watch?v=At4Vmo13vJE

I really like where this project is headed, and this post was just meant as a little tip to perhaps make it a bit better. I only shared my own personal background here to support my "unusual" claims a bit...

[OzOnE2k10]

Hi, geekmaster,

I can't admit to understanding much about what you're saying, as I don't know a great deal about coding these complex algorithms etc.

It does sound interesting though, and if you could suggest ways to improve the Foculus code I'm sure it would help DIY Rift builders.
I imagine the solution Yeti went with for the Foculus was the quickest solution at the time, not necessarily the best.

It sounds like it works pretty well though.
I'd like to see some custom "Rift API compatible" tracker boards being made at some point, to see if any other improvements can be made over the DK1 tracker.
(as good as it is, components and software is improving all the time. Some very clever stuff being done by modders / hackers these days.)

When you're talking about the Hydra, do you mean the accelerometer board or the Razer Hydra?
I've been meaning to build a small board which emulates the joystick data for the Razer Hydra, so I can experiment with different inductors etc. for the tracking part.

Fellow Aspie here as well. You would be amazed how common it is these days...

I've done quite a few Hangout sessions over the past year with people on the retro computer groups, and we suddenly realized that around a third of us were on the AS spectrum.

I was only diagnosed about three years ago. It was a bit of a surprise at first, but then not really, 'cos it hasn't really changed anything as such.
But, it means that I never really leave the house without somebody else, and don't really have my independence.

There are benefits though as you say - I don't think I'd be so heavily into electronics or computers if it wasn't for the Aspie thing.
It also means that I can take a fresh look at certain problems where most "normal" people wouldn't.

If only I could use that knowledge to actually make a living out of it. lol

Anyway, don't want to get this thread too far off-topic.

The Foculus thing was worth mentioning as it's a good option for DIY modders as a Rift-compatible tracker.
As an obvious tie-in, the Foculus tracker could be used alongside a RiftUP! kit to build a very nice DIY Rift.


OzOnE.

[geekmaster]

Fellow Aspie here as well. You would be amazed how common it is these days...

I've done quite a few Hangout sessions over the past year with people on the retro computer groups, and we suddenly realized that around a third of us were on the AS spectrum.

I was only diagnosed about three years ago. It was a bit of a surprise at first, but then not really, 'cos it hasn't really changed anything as such.
But, it means that I never really leave the house without somebody else, and don't really have my independence.

There are benefits though as you say - I don't think I'd be so heavily into electronics or computers if it wasn't for the Aspie thing.
It also means that I can take a fresh look at certain problems where most "normal" people wouldn't.

If only I could use that knowledge to actually make a living out of it. lol

One quick thought -- I hear rumors that only a "machine like" aspie is neurocompatible to interface with our AI overlords after the Technological Singularity. Perhaps we can walk among them instead of being kept on a reservation. Any truth to those rumors? Perhaps that is why "they" are evolving us this way...

[geekmaster]

Fellow Aspie here as well. You would be amazed how common it is these days...

I've done quite a few Hangout sessions over the past year with people on the retro computer groups, and we suddenly realized that around a third of us were on the AS spectrum.

I was only diagnosed about three years ago. It was a bit of a surprise at first, but then not really, 'cos it hasn't really changed anything as such.
But, it means that I never really leave the house without somebody else, and don't really have my independence.

There are benefits though as you say - I don't think I'd be so heavily into electronics or computers if it wasn't for the Aspie thing.
It also means that I can take a fresh look at certain problems where most "normal" people wouldn't.

If only I could use that knowledge to actually make a living out of it. lol

One quick thought -- I hear rumors that only a "machine like" aspie is neurocompatible to interface with our AI overlords after the Technological Singularity. Perhaps we can walk among them instead of being kept on a reservation. Any truth to those rumors? Perhaps that is why "they" are evolving us this way...

The recent WSJ article that claims companies must be one-percent aspie to remain competetive, may help more of us find gainful employment inside the "beast", instead of staying mostly "self-employed" (1099 in the USA)...

Re: the Hydra, not IMU or accelerometer. Magnetic absolute position and orientation sensing (relative to the base). It has lots of little firmware glitches (privately acknowledged by sixense) but I can work around all of them I have found. The hydra has a cool buffer overflow that lets me read 4K of RAM where the variables are stored, so perhaps I can use "insider" data to do things the firmware normally does not allow.

I am way overloaded, schedulewise. Too many personal hybrid VR/robotics and related projects (with sponsors for parts and trips). If I get time, I would like to try putting some snippets of my "outlier exclusion" code into that STM32F3DISCOVERY foculus code (I have 3 of those boards, and 4 hydras, and countless other dev kits and VR peripherals, but not enough time despite only sleeping 2 or 3 hours per day most of my life -- an Aspie thing). Outlier exclusion and thresholding can replace high-latency sensor filters in many cases (isn't it "obvious", or is that an aspie thing?), just like superior "bang-bang optimal control" can replace "speed profile ramping" in motion control. Having more productive waking hours in the day than NT "normals" is cool too, eh? AI-compatibility, man! Now where is my Direct Neural Interface to the AI matrix of cyberspace? But we must remain compassionate and lead those who cannot keep up with the Technological Singularity (next 45 years?) into a comfortable and productive life too (perhaps a life of leisure, but with a purpose). Party on lords of geekdom!

I want my DNI/BMI, but until then, a FullHD FrankenRift will have to suffice...

Okay, so it seems I coined a new term here, so it is now time for an official proclamation:

A DIY HD-upgraded Rift shall henceforth be known as a "FrankenRift".

[OzOnE2k10]

I definitely don't have the low sleeping hours thing, quite the opposite. :p
I do have problems with reversed sleeping patterns however.

I'd like to hear more about the Hydra stuff if you want to PM me?
It has been a while since I even powered up my Rift PC though (been busy repairing arcade PCBs and retro consoles for people).

I was tracing a fair bit of the PCB layout a few months ago after I saw the teardowns on another thread (I bought my Hydra after the price drop).
I noticed it uses a fixed oscillator, then has a mux for switching between the three coils.
It would be cool to tweak the osc freq slightly on-the-fly to see what the effect would be, as well as changing the inductors to see if their quality makes a difference.

I know what it's like with these projects though - still not enough hours in the day, especially if you're working.
(I'm kinda "between jobs" right now. lol)

If you do get chance to look into the Foculus code, it would be good to maybe see a new thread on it showing your findings etc.

OzOnE.

[rozsnyo]

See the future's back! Little sister is a 5" Sharp and the bigger brother is the 5.5" LG. If you want to ask if they work... yes they do!
And now I need to seriously clean-up my firmware, as it has more #ifdefs than effective code - after all those hardware versions, screen and feature combinations

[OzOnE2k10]

#ifdef ROZSNYO_IS_COOL!

RiftUP!_Enabled();

#endif