RifTUP! FullHD Oculus DK1 upgrade kit

[rozsnyo]

Picture sizes for the impatient ones!

[Hornet]

Daniel congrats! You are the champion! Better than Cibulkova

[rozsnyo]

Reporting that the 5.5" ( LG LH550WF1 ) interface module is working too. So now you got a choice of 5.0" ( Sharp LS050T1SX1 ) or this 5.5" LG.
I had a terrible coding error when accessing the 16bit fifo in the MIPI bridge... the 5" sharp screen was okay to init with 1 byte commands - so those 2+ byte ones for the LG were corrupted.

[mr.uu]

So, the 5" sharp or the 5.5" LG.
Have to make up my mind about
- which one comes as a pair closer to the original rift DK screen dimension, especially width?
- which one is the "better" panel in response time, contrast, etc.?
- price difference and availability?
-400 ppi or 440 ppi

EDIT: OR the easiest, a single 5.9" swap, 370 ppi. Hmm. Does anybody KNOW, which one is the best panel (specs wise)?

EDIT: Daniel, do you have a RiftDK? Could you do pictures of the screens through a DK lens for comparison, how it would look like? Especially regarding screendoor...

[rozsnyo]

And the last today - the 5.9" big brother from Sharp - the mighty LS059S1TX1 for single screen DK1 upgrade

[3dvison]

And the last today - the 5.9" big brother from Sharp - the mighty LS059S1TX1 for single screen DK1 upgrade

I think that is the winner.
When can you send me one ???

[rozsnyo]

With Kernel32 user here on forum, we have tried the 5.9" with oculus DK1... the 5.9" would work - so I continue to make the HDMI-MIPI kit for that screen. It will feature a MINI HDMI plug on (screen bottom) = DK1 right side - you will have to cut a hole for it.

I am also going to test if the DK1 cable if it is suitable for HDMI transmission, because if yes, we can make a upgrade kit - replacing contents of both helmet and the breakout box with our boards. Anybody with a spare cable / broken screen or box?

Some pictures attached, but the cups / camera were not in proper distance, I have noticed it just now (the FOV should be much less and pixels larger). Also the persistence test is probably different phase on this 5.9" screen and the original screen - both show a cca 40ms persistence. (2 + partial frame)

Screen rotation is provided in Intel drivers (i7-3770s).

[ripcurl123]

this is great news finally hardware instead of vaporware ,you know who you are vaporware people say no more ,anyway on another note,
does anyone know where to get the sharp 5.9 screen from? as im having a lot trouble locating one. or at least what phone/note its in.
great pics Daniel by the way.

dave

[rozsnyo]

I will supply the screen with the interface board - make sense to buy them in bigger quantity and test every piece before shipping. Actually I've got six 5.9" screen in stock with me.

[mr.uu]

...I am also going to test if the DK1 cable if it is suitable for HDMI transmission, because if yes, we can make a upgrade kit - replacing contents of both helmet and the breakout box with our boards. Anybody with a spare cable / broken screen or box?

Daniel, i am ready to send one of my DKs to you for testing, which you could then convert for me. I pay for conversion, no problem. But you have one DK, at least you have access to one, so what do you need a spare cable for?

...Some pictures attached, but the cups / camera were not in proper distance, I have noticed it just now (the FOV should be much less and pixels larger). Also the persistence test is probably different phase on this 5.9" screen and the original screen - both show a cca 40ms persistence. (2 + partial frame)

Pictures through the lens at proper distance are important for me to judge the picture quality to expect. This is the most important factor for me when choosing the screen. I would be very happy if you could find time in the next week(s, don't hurry) to post them. Thank you.

[OzOnE2k10]

Hi, Daniel,

I completely forgot you had a DK1 already - is that a friend's Rift?

That 5.9" screen looks fantastic btw. Congrats!

I need to reply to your last e-mail too, but I've been really busy with the "other-other" HDMI project.

I think the original cable in the DK1 might work OK for HDMI.
LVDS is generally fussier, so will probably work through that fairly short length of cable.
Then again, I realize there are all sorts of impedance matching and other issues that could arise, so needs testing.

@mr.uu - thanks for the offer of sending your DK1 to Daniel, I think he will really appreciate it.
But, I don't think the final small version of the MIPI board is complete quite yet!?

It will take a little while for Daniel to finalize the design based on what he finds during testing with his MIPI dev platform.
Then he will know which features to keep, then finish the physical design to fit inside the HMD, and finish the breakout box.

We were discussing before about the possibility of powering the Rift directly from one or two USB ports (current consumption a tiny bit high for one port).
Then maybe leave a micro-HDMI or other socket in the HMD so people can choose their own HDMI cable?

Please keep in mind that Daniel is very busy with many other projects too. The progress looks great so far though.


@KBK - those new chips look very interesting, and will be great if they will support MIPI too.
May be a while before they hit the shelves and can be bought in smaller quantities though. Definitely worth keeping an eye on.
(and whether the datasheet will be available, as you say)


OzOnE.

[Kernel32]

With Kernel32 user here on forum, we have tried the 5.9" with oculus DK1... the 5.9" would work - so I continue to make the HDMI-MIPI kit for that screen. It will feature a MINI HDMI plug on (screen bottom) = DK1 right side - you will have to cut a hole for it.
...

At Daniel's it felt like visiting Willy Wonka Chocolate factory. Chocolate being some kick-ass screens with functional HDMI-MIPI boards.

We took the DK1 apart and replaced the original display with the FullHD 5.9" one. We wasn't able to head-mount the assembly, because the developer board just isn't meant to be moved around as much. Nonetheless we were able to see images and prerecorded video gameplay.

The increase of image sharpness is simply huge. Forgetting it's not real just got a lot easier. Especially text and objects at longer distances are much more defined and easy to recognize. Going back to 1280×800 feels like Stone Age.

Screen-door effect is much less pronounced. Although you can still recognize individual pixels and a tiny grid, it is more a microstructure now than the original in-your-face black grid. I think, if you still mind, a diffusive filter mod, as seen on the forum, will remove the screendoor effect altogether, while still leaving sharp image thanks to the high DPI and less diffusion needed.

The one inch smaller screen goes best with C cup lenses, as they magnify more than A cups and thus compensate for smaller screen size. FOV felt unchanged to regular C cups setup.

I was worried about increased persistence and judder, but it is not the case. Motion-blur is where it was and feels maybe even a little bit less present. Daniel made some persistent tests, as the moving car seen in his post. The screens were set to their native resolutions, making the 5,9" display car physically smaller, so it's really only to judge persistence, not sharpness/dpi.

All together, I'm very pleased by what I've seen and can't wait to get my hands on upgrade kit.

It should be easy to install the kit as Daniel most likely will fit the display and board into the Oculus rift front lid. There is quite a lot of space. He uses a little bit more expensive, but miniature, components and is able to produce a very low profile boards. The board will be most likely a simple swap solution including physical auto centering/leveling of the smaller display.

Unresolved question is the cabling for now. Daniel may use the original cable or put a mini HDMI port (not to be confused with micro HDMI) directly on the headpiece to carry all video/audio, power and USB trough it. HDMI port would require us to make a hole for the connector in casing on the right side, which I don't mind.

I'm glad we are finally moving somewhere thanks to Daniel. This is it

...I am also going to test if the DK1 cable if it is suitable for HDMI transmission, because if yes, we can make a upgrade kit - replacing contents of both helmet and the breakout box with our boards. Anybody with a spare cable / broken screen or box?

Daniel, i am ready to send one of my DKs to you for testing, which you could then convert for me. I pay for conversion, no problem. But you have one DK, at least you have access to one, so what do you need a spare cable for?

De-solder and use the LVDS connectors from both headpiece and box would be the fastest way to test parameters of present cable whether it is good enough.

[OzOnE2k10]

Oh, I missed the part about your testing, and about the MINI HDMI plug. lol

Big thanks for taking your Rift to Daniel and helping, Kernel32.

Good write-up too. The new panels sound every bit as good as we'd hoped.
Just to get rid of / reduce the "aliasing" along edges with the DK1 will be a massive improvement.

Did you get chance to try any 3D vids like the Rift Coaster yet?
I was wondering if panel makes the whole experience better, without so many distractions?

Maybe we can still implement backlight strobing on the final board?
Daniel - does the Dev platform already have control of the backlight driver?
If so, should be quite easy to add a simple variable delay to the backlight triggered by Vsync?
It could have a pre-delay as well, so it turns on the backlight before / after Vsync?

Backlight strobing could make a big improvement to the perceived persistence.
(I never got around to testing it on the DK1 / Hydis panel though, 'cos I thought it better to concentrate on the 1080p panels instead.)

EDIT: "I spy, with my big eye" - a PWM pin on the MIPI board.


Personally I don't mind at all about drilling a few holes in my DK1 - I don't think anyone will want to use the "stock" panel after seeing the 1080p.

@Kernel32 - from your initial testing, would you say that the 5" panel with C cups looked like it could give comparable FOV to the 5.9" panel with A cups?
Just to keep our options open, and it may be that the viewable PPI is slightly higher? (I'm not an expert on the optics stuff tbh).

btw, I don't suppose anyone could help hunt for the type of connectors use for the LVDS cable in the Rift?
I did have a long search on the Web before, but couldn't find the exact type.

IIRC, they are just 2mm pitch, but probably AMP / Molex, or something like that?
True though, Kernel32 - might as well just desolder the stock connectors for testing.

Regards,
OzOnE.

[rozsnyo]

Strobing could be done - but... when should the image be lit and when not? I see no point in that, as there is only one backlight and it either lits the whole surface or nothing. And the image is drawn from top to bottom, with almost none blanking (V: 1920 1928 1944 1954 = 1920:34 is active:blank ratio)... are you suggesting to drive the BL for that 1.77% of time at 56x of its designed intensity ? Because that solves only a sort of tearing, not persistence - as it is larger than 1 frame. (flashing the BL makes sense when the LCD response is shorten than one frame duretion - so lit it in the stable region, each frame - but we have no stable region)

[Kernel32]

Did you get chance to try any 3D vids like the Rift Coaster yet?
I was wondering if panel makes the whole experience better, without so many distractions?

We did try 3D video OzOnE. Image quality was great but IPD was a little off, because of another display width, resulting in sub optimal experience.
It got fixed in-game by increasing IPD, but I was not able to move the assembly to have full experience. It felt good, deep, but I would be glad to do some more testing.
I quess the right IPD to set is real IPD*(original display width/new display width).

@Kernel32 - from your initial testing, would you say that the 5" panel with C cups looked like it could give comparable FOV to the 5.9" panel with A cups?
Just to keep our options open, and it may be that the viewable PPI is slightly higher? (I'm not an expert on the optics stuff tbh).

We did test only the 5.9" panel trough Rift optics assembly as it is the widest. With A cups, you will also get higher viewable PPI on 5.9", at cost of some black space around the image. I would like to test the other screens if I have the opportunity, but 5.9" should be the best one given optics we have.
I would say that 5.9" with C cups is similar to 7" with A cups

Daniel can have my LVDS connectors for testing. There is no need for them after upgrade whatsoever.

Thank you Daniel for explaining the strobing effect. I guess it would make sense only for OLED display.

[3dvison]

I don't mind drilling holes, taping, gluing or whatever else it takes.
I think of this a DIY upgrade and having to do some of the work myself is fine.
No need to get too fancy with it and just focus on getting the 5.9 out as fast as you can.
I only say that, because I would like as much time as possible to use Daniel's upgrade, before the DK-2 comes out.
Can't believe one month is gone already and it February 2014.

EDIT: So were you able to disconnect the Rifts 7inch screen and only have the 5.9 running ? And if so, did the tracker still track and was the HMD still seen by software/games/demos as a Rift ? even with the Rift screen disconnected and only the 5.9 screen hooked up ?

[rozsnyo]

So were you able to disconnect the Rifts 7inch screen and only have the 5.9 running ? And if so, did the tracker still track and was the HMD still seen by software/games/demos as a Rift ? even with the Rift screen disconnected and only the 5.9 screen hooked up ?

The setup was: USB->RIFT, hdmi/dvi video on breakout box not connected + HDMI->MIPIdevkit ... with a totally nonrift EDID data. And yes - tracking worked with this kind of connection.

[OzOnE2k10]

Strobing could be done - but... when should the image be lit and when not? I see no point in that, as there is only one backlight and it either lits the whole surface or nothing. And the image is drawn from top to bottom, with almost none blanking (V: 1920 1928 1944 1954 = 1920:34 is active:blank ratio)... are you suggesting to drive the BL for that 1.77% of time at 56x of its designed intensity ? Because that solves only a sort of tearing, not persistence - as it is larger than 1 frame. (flashing the BL makes sense when the LCD response is shorten than one frame duretion - so lit it in the stable region, each frame - but we have no stable region)

Ahh, I see. So very little blanking for us to take advantage of. That's fine.

The motion blur / persistence on the DK1 screen isn't terrible anyway really.
I found that the low resolution was the main factor to a diminished experience.

Even better OLED screens will be just around the corner, and most of them will likely use MIPI, so we should be ready for them now.

Interesting that the tracker still worked with a non-Rift EDID on the HDMI?
I thought most of the Rift demos / games / SDK had to detect a valid ID string in the EDID before they will work?

(I haven't actually tried unplugging the HDMI cable on my DK1 to see if the apps still start up. I'll have to try that tomorrow.)

Of course, the EDID stuff is simple enough.
If it turns out that some games are fussy about it then I did see a 1080x1920 ID string in the SDK code many months ago that we could use.

@3dvision, yep I agree too - I don't mind at all to install into my Rift, and that was the original intention of the upgrade kit.
It's not difficult to get the Rift apart at all (apart from a few broken clips), and the stock panel can be easily prised out intact if you do it very slowly.

I'm still wondering exactly when the DK2 will be released.
I can't help thinking it could possibly just a be vapourware rumour to tide us over until the consumer version release? lol

@Kernel32 - thanks for the info. Can't wait to try it out now with the full head tracking etc.


OzOnE.

[3dvison]

So were you able to disconnect the Rifts 7inch screen and only have the 5.9 running ? And if so, did the tracker still track and was the HMD still seen by software/games/demos as a Rift ? even with the Rift screen disconnected and only the 5.9 screen hooked up ?

The setup was: USB->RIFT, hdmi/dvi video on breakout box not connected + HDMI->MIPIdevkit ... with a totally nonrift EDID data. And yes - tracking worked with this kind of connection.

Does that mean the EDID could be located in either the tracker or the rift control box and not the Rifts screen ?
I only say that because I thought I may have seen a post about that somewhere...but where I don't recall.

[OzOnE2k10]

EDID is almost certainly held in the Flash chip on the controller board (highlighted in orange)...

http://d3nevzfk7ii3be.cloudfront.net/ig ... f2A2A.huge

This is actually the SPI Flash chip for the Realtek scaler.
The Realtek firmware itself holds the EDID data, and simply acts like an I2C EEPROM interface to the HDMI port.

(some panels have an EDID EEPROM on them too, but the DK1 doesn't have the SDA / SCL control wires to the headset IIRC.)

But, the tracker's ID also gets picked up by the software, and many games / demos won't run without it.

EDID is very easy to spoof though. A few people have successfully used it already on their DIY Rifts.

I remember seeing a 1080x1920 "template" in the Oculus SDK shortly after their 1080p prototype was shown, so we likely just use that instead.
The SDK mainly just looks at the manufacturer ID string of the EDID to determine the template for the type of dev kit / screen size / warp.

Others will know a lot more on this though - I haven't looked at the source code for many months.

OzOnE.

[mr.uu]

...We did try 3D video OzOnE. Image quality was great but IPD was a little off, because of another display width, resulting in sub optimal experience.
It got fixed in-game by increasing IPD, but I was not able to move the assembly to have full experience. It felt good, deep, but I would be glad to do some more testing. ...

Well, did you notice wrong warping, too? Because if the screen has other dimensions, then warping should be adjusted. This could be reduced with the EDID of the first 1080p prototype, because it is using a smaler than 7" screen, too.

[OzOnE2k10]

@3dvision - here are the LVDS cable pinouts on the DK1 that I did a while ago...
(They are labelled on the screen anyway, but it's good to note them down.)

http://www.mtbs3d.com/phpbb/viewtopic.p ... 20#p139496

So, no possibility of EDID on the panel itself, it's only inside the Realtek firmware.

USB to the tracker is via four separate wires.

The posts below that include links to all the Realtek programming stuff I have.
Many of the cheap scaler boards you see can be reprogrammed via the VGA or DVI socket.

They just use a simple parallel port circuit and few pins in a certain way to reprogram the SPI Flash on the board.
I haven't tried this myself yet, but I know a few people on a PM who have apparently been successful with testing different code.

Spoofing the EDID is just a case of using a simple I2C EEPROM connected to the HDMI port though.
We don't really need to mess with the Realtek firmware, it's just for people who are interested atm.

The reprogramming stuff may come in handy for when new Realtek chips arrive, but they probably have a completely different SDK?

@Daniel - Not that it matters too much now with a nice 1080p panel, but does the 5.9" panel display lower resolutions at all, or does the gfx card need to output 1080x1920 @ 60Hz specifically atm? Of course, this doesn't affect the Desktop image, as almost all of the Rift demos work fine with a duplicated image / multi-monitor output.

@mr.uu - yes, it will be very interesting to see if the 1080p prototype EDID will suit the 5.9" panel.
The prototype used a very similar size I think? The warping should be very close indeed.

OzOnE.

[mr.uu]

Was looking up Panelook.com for panel specs:

Sharp 5": http://www.panelook.com/LS050T1SX01_SHA ... 17692.html
Active Area : 61.884×110.016 mm (W×H)
Outline Dimension : 64.3×118.2 mm (W×H)
Outline Depth : 1.4 (Typ.) mm

LG 5.5": http://www.panelook.com/LH550WF1-SD01_L ... 19410.html
Active Area : 68.04×120.96 mm (W×H)
Outline Dimension : 71.14×128.01 mm (W×H)
Outline Depth : 1.45 (Typ.) mm

Sharp 5.9": http://www.panelook.com/LS059T1SX01_SHA ... 18926.html
Active Area : 72.9×129.6 mm (W×H)
Outline Dimension : 76.2×139.1 mm (W×H)
Outline Depth : 1.65 mm

For comparison DK1-panel:
Active Area : 149.76×93.6 mm (W×H)
Outline Dimension : 161.2×107.2 mm (W×H)
Outline Depth : 2.5/4.5 (Typ./Max.) mm

So i am looking for a width of 149.8 and two 5.5" are 2x68 is 136 plus 2xoutline 1.45 (middle part) is 138.9 total width. Which is a difference of 10.9 (5.45 per side) mm.
Two 5.9" are 2x72.9 is 145.8 plus 2x1.65 is 149.1, which makes a perfect dual setup replacement.
Two 5" in portrait are to narrow. But in landscape its 220 plus 2x1.4 is 222.8mm. Height is one third less -> 5" doesn't fit.

Hmm... Brings me to either
- the 5.9" single swap option, 370 dpi, accepting a left and right 10.1mm reduced active screen width, or
- the dual 5.5" setup, 400 dpi, accepting a left and right 5.45mm reduced active screen width, or
- the dual 5.9" setup, almost same width (diff. <1mm)

...which brings me to the conclusion (so far ) to go with the dual 5.5" portrait setup. Okay, weight is probably an issue, but i hope that Daniel can shrink the HDMI-MIPI board considerably. The DK1 has 216 dpi, so 400 dpi would be almost four pixel where now one is, which i can imagine how it looks like .

EDIT: corrected summary

[mr.uu]

@mr.uu - yes, it will be very interesting to see if the 1080p prototype EDID will suit the 5.9" panel.
The prototype used a very similar size I think? The warping should be very close indeed.
OzOnE.

It is speculated to be the LG 5.5" panel. But maybe it was the 5.9" and daniel is doing what Oculus had ready many month ago...

The scaling has to be set to internally (gfx scaler) in the CCC (ati). Then you can drive it with almost any resolution you want.

[rozsnyo]

The scaling is done internally on GFX - when you present a fixed mode EDID, then the PC will scale anything.

It works even in boot stage - bios showing list of detected drives and memory, but even that I have a decent Z87 board, the bios is that shitty UEFI one and anything graphical is no go - no DEL to bios, no F8 to select boot media - thats graphical too (unfortunately I cant boot Win without explicitly specifying the drive... that bios is buggy as hell).

[Hadtstec]

This is fantastic news and I would love to put a new panel into my DK1!
Keep up the good work!

[fullban]

HELLO PEOPLE

i have just ordered a rift from there site im from uk so not sure when it will be hear but i would be really interested in this what would i need to be able to do
to get this incorporated into my rift when it comes and how much would it cost?

thanks chris

[albertopina1]

http://all-spares.com/es/LCD-for-China- ... 01-V2F.php

[OzOnE2k10]

@albertopina1 - Wow, that's cheap!

I hope it definitely is the FHD version, but it certainly looks like it.

There have been quite a few discussions about it as a display candidate before. It looks like a seriously nice panel.

http://www.mtbs3d.com/phpBB/viewtopic.p ... 80#p146680
https://developer.oculusvr.com/forums/v ... 7&start=40
https://github.com/gokulnatha/GT-I9505/ ... ll_hd_pt.c

The good thing is, there is plenty of info on the MIPI commands, and @cgp44 has done some great detective work in figuring out the pinouts / connectors.

I'm looking to buy one or two of these panels, and the Kindle HDX 7" one as well next week so Daniel can test them.

The only slight drawback of the S4 AMOLED is that it's only 5". That may be a tiny bit too small to work well in the Rift.
Well worth testing it though - I bet the latency and blurring is hugely reduced.

OzOnE.

[remosito]

proly angood time to repost this link back from page 13:

http://www.wayengineer.com/index.php?ma ... ts_id=5304


on another Note (hihi) : http://www.dhgate.com/product/100-new-o ... 2773580535 )

[mr.uu]

Wow, S4 AMOLED FHD lets me think about the 5" dual landscape setup (despite the 1/3 smaler vertical view), IF Daniel can drive it somehow "with" low persistence

I think we will see a lot of useable displays on the horizon in the next weeks. If we are lucky also >FHD panels...

EDIT: Kindle HDX 7" 1920x1200 is "only" 323ppi, S4 AMOLED 5" 1920x1080 has 441ppi...

[cgp44]

Ive gone off the Samsung S4 screen given panels at a third the price
and really the same true resolution.

Can we get the details off the unavailable datasheet for the LG 5.5"er LH550-WF1-SD01
Also the connector looks 'reasonable'. Do we have a connector part number?

ie connector pinouts, timing, DSC mode and any reset commands.

Could those who have access please report these. Thanks.

[Quppa]

Great to see progress being made.

Are there any candidate 7" panels with a landscape orientation, or are they all portrait? That KOE LVDS panel seems not to have materialised.

[cegli]

Really good work so far! I'm not sure I understand how a double screen would work, unless you window the game to keep it at the right aspect ratio. Wont it stretch everything terribly, because the games will not understand to crop the bottom 1/3rd of the screen off? Do games lock the aspect ratio correctly?

I think an interesting screen to use would be the Galaxy Note 3 screen (1080p Pentile OLED 5.5"), but I think it might be hard to find it for cheap. Probably a bit of a small FOV too. I guess it wouldn't help that much unless we can run it at a fast refresh rate. Anyone find somewhere to buy it cheap?

[rozsnyo]

Can we get the details off the unavailable datasheet for the LG 5.5"er LH550-WF1-SD01
Also the connector looks 'reasonable'. Do we have a connector part number?
ie connector pinouts, timing, DSC mode and any reset commands.
Could those who have access please report these. Thanks.

Datasheets are crap - having a typo in the connector number like it was written by a dyslectic could be forgiven, but stating a massive binary initialization sequence which does not work is an engineers nightmare. The datasheet also does not contain the timing. Anyway - the display can be operated as one can see from my photos

What I have now noticed amongst 5.0 / 5.5 / 5.9 is the driving method - the exact type of "column inversion". Some method seems somehow flickering - like it is refreshing only the even/odd lines. Not sure what to think of this.

[mr.uu]

Really good work so far! I'm not sure I understand how a double screen would work, unless you window the game to keep it at the right aspect ratio. Wont it stretch everything terribly, because the games will not understand to crop the bottom 1/3rd of the screen off? Do games lock the aspect ratio correctly?

Sounds complicated, is in reality very easy. I can only tell you the way it works with AMD cards:
First you setup via CCC a eyefinity dual screen setup. Then you let the gfx card do the scaling (also CCC). And at this tab you can select the scaling behavior when driving the screen with different than native resolutions. You can either select native - picture will be centered, pixel by pixel, leaving a black border around it, or aspect ratio - picture will be streched but aspect ratio is beeing kept, or streched - picture will be streched to fill all pixels of the dual setup.
So you have full control what and how it is displayed...

[cgp44]

Rozsnyo,

what is the sharp 5.99er datasheet like? I've grabbed it from the link previous.

I hope you can shrink your board to no more than two pieces of pcb.

I know your commerical imperative, but could you share any secrets
you find regarding getting displays to work where the datasheet is
misleading. I understand if you don't.

[cegli]

Really good work so far! I'm not sure I understand how a double screen would work, unless you window the game to keep it at the right aspect ratio. Wont it stretch everything terribly, because the games will not understand to crop the bottom 1/3rd of the screen off? Do games lock the aspect ratio correctly?

Sounds complicated, is in reality very easy. I can only tell you the way it works with AMD cards:
First you setup via CCC a eyefinity dual screen setup. Then you let the gfx card do the scaling (also CCC). And at this tab you can select the scaling behavior when driving the screen with different than native resolutions. You can either select native - picture will be centered, pixel by pixel, leaving a black border around it, or aspect ratio - picture will be streched but aspect ratio is beeing kept, or streched - picture will be streched to fill all pixels of the dual setup.
So you have full control what and how it is displayed...

Wow, that's pretty slick. I've never messed around with the eyefinity settings on my AMD card, but maybe I'll have to check it out!

[ExZero16]

I've been following this post for a really long time now and it looks like it has finally gotten figured out.
Who do I contact to acquire one of the HDMI to MIPI boards? I can do all the modding myself I would just like to get one of the boards to connect a screen to.

Thank you!

[cegli]

I've been following this post for a really long time now and it looks like it has finally gotten figured out.
Who do I contact to acquire one of the HDMI to MIPI boards? I can do all the modding myself I would just like to get one of the boards to connect a screen to.

Thank you!

They're not acquirable yet. Rozsnyo/Daniel is showing off his completed board, but it sounds like he hasn't finished miniaturizing it and hasn't started mass producing it yet. We've also seen examples of the HTBM5 HDMI to MIPI board working, but it seems to cost almost $200.00!

It's taking so long, but getting closer. I was convinced I'd have the KOE panel working by Christmas, then it turned out to be vaporware... It seems hopeful Rozsnyos will follow through on this!