RifTUP! FullHD Oculus DK1 upgrade kit

[zacherynuk]

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.

I'm afraid that once the damage was done I used snips and metal screwdrivers on the rest of it. I was hoping to take some nice after pictures but with the frame now in half I couldn't see the point. Sorry.

[Inscothen]

Sorry to hear about the trouble zacherynuk.

Will these 7" panels be available without the touchscreen? The Galaxy s4 replacement screen is only available with the touchscreen, I hope these will be able to be purchased separately for anyone doing a 7" mod.

About the tracker's USB being just passthrough, my devkit's screen won't work without the USB being plugged in. Does that mean the display is powered through USB?

[zacherynuk]

Sorry to hear about the trouble zacherynuk.

Will these 7" panels be available without the touchscreen? The Galaxy s4 replacement screen is only available with the touchscreen, I hope these will be able to be purchased separately for anyone doing a 7" mod.

About the tracker's USB being just passthrough, my devkit's screen won't work without the USB being plugged in. Does that mean the display is powered through USB?

Unsure about panel availability, seeing how flimsy it is without any external structure it may well not be (even the batter is squishy!). Asus are not helpful and the parts and 3rd party repair people have no stock of any internals yet.

The display is powered from USB - but you could plug into a phone charger, rather than PC if required.

[MSat]

Sucks that you broke your panel, zacherynuk

But now that it's trashed, you think you could take a peek at the other side of the PCB on the panel now and see how many traces go from the ribbon connector towards the glass itself (which I assume has the LCD driver in die form bonded to the glass)?

[zacherynuk]

Sucks that you broke your panel, zacherynuk

But now that it's trashed, you think you could take a peek at the other side of the PCB on the panel now and see how many traces go from the ribbon connector towards the glass itself (which I assume has the LCD driver in die form bonded to the glass)?

I have never seen anything like this on a panel: this, literally, is it: http://www.wastedspace.co.uk/cms/wp-con ... 092811.jpg

I'll get some macro shots tomorrow, but there is nothing to it; not even anything I would recognise as backlights - it's incredible.

[KBK]

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.

I'm afraid that once the damage was done I used snips and metal screwdrivers on the rest of it. I was hoping to take some nice after pictures but with the frame now in half I couldn't see the point. Sorry.

I had no idea it was yours!

I would have been a bit more politic, as they say. (I didn't quite fart in the elevator. Just near the door)

[MSat]

Sucks that you broke your panel, zacherynuk

But now that it's trashed, you think you could take a peek at the other side of the PCB on the panel now and see how many traces go from the ribbon connector towards the glass itself (which I assume has the LCD driver in die form bonded to the glass)?

I have never seen anything like this on a panel: this, literally, is it: http://www.wastedspace.co.uk/cms/wp-con ... 092811.jpg

I'll get some macro shots tomorrow, but there is nothing to it; not even anything I would recognise as backlights - it's incredible.

Is that flex board glued on? Can you peel it and flip it over?

[Peva3]

http://www.engadget.com/2013/08/16/benc ... h-2-560-x/

Looks like amazon is working on a kindle with 2560x1600 screen. Assuming they will be upgrading one (or both) of their current product lines (7" & 8.9") and not introduce a larger kindle, the resulting PPI would be 431.27 for a 7" and 339.2 for a 8.9".

Also the current 8.9" kindle has a 1920x1200 screen. Does anyone know the specs of it? (MIPI?)

Edit: Here's a picture of its current connector along with a fixit teardown.

http://ecx.images-amazon.com/images/I/5 ... L1000_.jpg

http://www.ifixit.com/Guide/Installing+ ... en/12617/1

[LaserEdge]

I got my HDMI to MIPI board up and running with the LS050T1SX01. I had only one bug with the PCB itself which I easily worked around. I didn't think the panel would want the power up of AVDD- and AVDD+ 1ms after IOVDD. The enables for the switching regulators had to be separated. I was going to flesh out the power management features on the next board rev anyway so I can add a stand-by mode for portable applications.

Still need to experiment with a few things in firmware like CABC. Its a on panel brightness control feature. Also want to experiment with settings in firmware to see if can lower active power consumption.

Will post pictures later. It's 3:40 am !

[Peva3]

I got my HDMI to MIPI board up and running with the LS050T1SX01. I had only one bug with the PCB itself which I easily worked around. I didn't think the panel would want the power up of AVDD- and AVDD+ 1ms after IOVDD. The enables for the switching regulators had to be separated. I was going to flesh out the power management features on the next board rev anyway so I can add a stand-by mode for portable applications.

Still need to experiment with a few things in firmware like CABC. Its a on panel brightness control feature. Also want to experiment with settings in firmware to see if can lower active power consumption.

Will post pictures later. It's 3:40 am !

Of course someone would get the sharp screen working the day after I buy my two 5.6" 720p screens haha.

But really congrats! Can't wait to get my hands on the board when the bugs are worked out

[Endemo]

Awesome news laser!
Cant wait to get my hands on a working solution. Also as a non tech geek i cannot thank you enough for all the invested time.

[Flim]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

[Peva3]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

I actually would love to use this with the Sharp screen. Hannibalj2 said he would release the files once a few of the bugs were worked out. It's designed for two 7" screens, so two 5.6" screens should be able to fit with a ton of room to spare.

http://www.mtbs3d.com/phpBB/viewtopic.php?f=26&t=18335

[jquantum]

I'd be ok with using the sharp for now just to get my hands on a working display this is awesome news as is. Was the wiring straightforward?

[LaserEdge]

Here are some screenshots of the board connected to a LS050T1SX01. I choose to display this thread as evidence of it up and running. It works just like a mini-monitor so I can display anything I want on it. Its running at 1080x1920x60Hz 24bpp. The second screenshot is the panel magnified.

[Peva3]

Here are some screenshots of the board connected to a LS050T1SX01. I choose to display this thread as evidence of it up and running. It works just like a mini-monitor so I can display anything I want on it. Its running at 1080x1920x60Hz 24bpp. The second screenshot is the panel magnified.

two please.

[remosito]

That is wonderful work! I hope you realize the boost you are giving DIY HMD enthusiasts! HD HMDs! Incredible!

It always blows my mind the things people know and master. You did that PCB by yourself! To someone who can tell a resistor from a capacitor, but not much more, that's crazy stuff!

Can the PCB be reprogrammed for other panels? Like the LS059 or nexus 7-2?

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

[LaserEdge]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

For the 5.5 LG I am in the process of trying to obtaining the PCB connector for it. The part number is GB042-30s-H10. My domestic partner is Chinese which helps a ton with making connections in China. She is helping me get the part. Will work with her today on it. She helped me get connectors for the LS050T1SX01. I will go ahead and get PCBs fabbed for the 5.5LG. My strategy is to have PCBs designed for each board to keep everything as small as possible.

Of course if you or anyone else can help obtain the GB042-30s-H10 that would help.

[remosito]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

For the 5.5 LG I am in the process of trying to obtaining the PCB connector for it. The part number is GB042-30s-H10. My domestic partner is Chinese which helps a ton with making connections in China. She is helping me get the part. Will work with her today on it. She helped me get connectors for the LS050T1SX01. I will go ahead and get PCBs fabbed for the 5.5LG. My strategy is to have PCBs designed for each board to keep everything as small as possible.

Of course if you or anyone else can help obtain the GB042-30s-H10 that would help.

Cant help you with the connector. Though with the LS059 now available ( see link I edited into previous post above), maybe LG no longer really needed? Should be same connector as your LS050 and 85$ seems very good price!

I'll gladly sponsor you one if you'd like!

[LaserEdge]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

I actually would love to use this with the Sharp screen. Hannibalj2 said he would release the files once a few of the bugs were worked out. It's designed for two 7" screens, so two 5.6" screens should be able to fit with a ton of room to spare.

http://www.mtbs3d.com/phpBB/viewtopic.php?f=26&t=18335

Hi Peva3,

Took a quick look at the thread. Looks like work based on Foisi's InfinitEye design. I am still a believer that dual screens can make high quality HMDs. The weight of two 5" Sharps with this board will be around 60g. Already choose to have the MCU off board so dual screens can share one MCU. I will read that thread in detail later.

[LaserEdge]

That is wonderful work! I hope you realize the boost you are giving DIY HMD enthusiasts! HD HMDs! Incredible!

It always blows my mind the things people know and master. You did that PCB by yourself! To someone who can tell a resistor from a capacitor, but not much more, that's crazy stuff!

Can the PCB be reprogrammed for other panels? Like the LS059 or nexus 7-2?

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

Hi, remosito.

Thanks for the compliment. It was fun working on this board. I think it turned out pretty well for a first revision.

Yes. I am using an Arduino mini to program the board. It is highly configurable. The issue for any board is going to be the connector and the pinout. I am guessing the EDID and the Arduino firmware is going to be mostly the same for the majority of these 1080P panels. There are all using the same driver IC in the LCD panel.

[Peva3]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

I actually would love to use this with the Sharp screen. Hannibalj2 said he would release the files once a few of the bugs were worked out. It's designed for two 7" screens, so two 5.6" screens should be able to fit with a ton of room to spare.

http://www.mtbs3d.com/phpBB/viewtopic.php?f=26&t=18335

Hi Peva3,

Took a quick look at the thread. Looks like work based on Foisi's InfinitEye design. I am still a believer that dual screens can make high quality HMDs. The weight of two 5" Sharps with this board will be around 60g. Already choose to have the MCU off board so dual screens can share one MCU. I will read that thread in detail later.

Thanks for the reply

I'm currently building a Foisi HMD, but adding two 1080p screens would really be an upgrade for it. Is there any timeframe for these boards?

[LaserEdge]

Laser, you will be a very popular person here soon! Just give me the ability to fire up the 5.5 LG and I can CAD up the new housing for everyone...

For the 5.5 LG I am in the process of trying to obtaining the PCB connector for it. The part number is GB042-30s-H10. My domestic partner is Chinese which helps a ton with making connections in China. She is helping me get the part. Will work with her today on it. She helped me get connectors for the LS050T1SX01. I will go ahead and get PCBs fabbed for the 5.5LG. My strategy is to have PCBs designed for each board to keep everything as small as possible.

Of course if you or anyone else can help obtain the GB042-30s-H10 that would help.

Cant help you with the connector. Though with the LS059 now available ( see link I edited into previous post above), maybe LG no longer really needed? Should be same connector as your LS050 and 85$ seems very good price!

I'll gladly sponsor you one if you'd like!

For the LS059T1SX01 only changing the connector and updating the EDID will be required. The firmware will be exactly the same as the LS050T1SX01.

I have to think more about if this board is something I want to sell. My goal is to get reality recording and reality playback off the ground. We need affordable very hi-res HMDs to do that. You can think of it as instantaneous traveling without traveling. I believe this is the application that will make VR mainstream and as common as the TV is today. Having the right application to meet something that everyone wants is the key!

[remosito]

Cant help you with the connector. Though with the LS059 now available ( see link I edited into previous post above), maybe LG no longer really needed? Should be same connector as your LS050 and 85$ seems very good price!

I'll gladly sponsor you one if you'd like!

For the LS059T1SX01 only changing the connector and updating the EDID will be required. The firmware will be exactly the same as the LS050T1SX01.

I have to think more about if this board is something I want to sell. My goal is to get reality recording and reality playback off the ground. We need affordable very hi-res HMDs to do that. You can think of it as instantaneous traveling without traveling. I believe this is the application that will make VR mainstream and as common as the TV is today. Having the right application to meet something that everyone wants is the key!

this page has connector info I think. It's chinese or korean or sth. so not sure.

http://www.kupanel.com/P-LS059T1SX01


接口类型:Connector
接口品牌:Panasonic
接口型号:AXT630124
接口脚距:0.4 mm
接口数量:1 pcs
接口脚数:30 pins


page for the LS050 has different info there:

信号大类:MIPI
信号小类:MIPI (4 data lanes)
面板电压:5.0/-5.0V (Typ.)(AVDD(+)/AVDD(-))
面板电流:11.9/5.6mA (Typ.)(AIDD(+)/AIDD(-))
接口类型:FPC
接口数量:1 pcs
接口脚数:31 pins


same connector would have been nice At least it's an easy available one. Assuming that info is correct.

[studdart]

I have to think more about if this board is something I want to sell.

If you do, I will happily take one! If not, perhaps we could convince Adafruit or Sparkfun to carry it for you? It's a very useful little board even outside of the HMD realm.

I'm somewhat late to this party, but I would be able to help on the software side (both Arduino or SDK tweaking) if needed. If the EDID idea doesn't work out I'm fairly sure we could modify the Oculus SDK (as long as this isn't a commercial HMD, it seems it's okay by Oculus' licensing terms). I'd be a bit happier if Palmer would give it his blessing though. I don't really like to step on anyone's toes.

[LaserEdge]

this page has connector info I think. It's chinese or korean or sth. so not sure.

http://www.kupanel.com/P-LS059T1SX01


接口类型:Connector
接口品牌:Panasonic
接口型号:AXT630124
接口脚距:0.4 mm
接口数量:1 pcs
接口脚数:30 pins


page for the LS050 has different info there:

信号大类:MIPI
信号小类:MIPI (4 data lanes)
面板电压:5.0/-5.0V (Typ.)(AVDD(+)/AVDD(-))
面板电流:11.9/5.6mA (Typ.)(AIDD(+)/AIDD(-))
接口类型:FPC
接口数量:1 pcs
接口脚数:31 pins

Based on the LS059T1SX01 datasheet the board to board connector on the LCD panel itself is the Panasonic AXT630124. We need to get the corresponding connector to that which is the Panasonic AXT530124. These are socket/plug type board to board connectors. The LS050T1SX01 requires a different B2B connector... the Panasonic AYF333135.

[LaserEdge]

If you do, I will happily take one! If not, perhaps we could convince Adafruit or Sparkfun to carry it for you? It's a very useful little board even outside of the HMD realm.

I'm somewhat late to this party, but I would be able to help on the software side (both Arduino or SDK tweaking) if needed. If the EDID idea doesn't work out I'm fairly sure we could modify the Oculus SDK (as long as this isn't a commercial HMD, it seems it's okay by Oculus' licensing terms). I'd be a bit happier if Palmer would give it his blessing though. I don't really like to step on anyone's toes.

Yea. I agree this board has many uses even outside of HMD applications.

I want to see Oculus succeed just like everyone else. What I am doing is working on a project that Palmer was helping with before he started Oculus and the Rift kickstarter. Palmer knows about it. What we are faced with is even a 1080P Rift would not be enough resolution for the projects needs. Now a dual screen 1080P starts to get us in the ballpark for our res needs. I assure you whatever comes out of this project wouldn't be stepping on any toes. In fact if our project succeeds Oculus would gain as well.

Will get back to everyone later today about the possibility of selling this board, so they can purchase for their own projects.

[studdart]

Now a dual screen 1080P starts to get us in the ballpark for our res needs.

On that tangent, I've been wondering how to make sure two screens are v-sync'ed. I don't believe (but very happy to be proven wrong) that it's not possible from the computer/hdmi end of the chain. So that would mean some kind of timing lock/signal between two of your control boards synchronizing the two displays.

I read through (very briefly) the MIPI DCS spec you posted and it looks like it may be possible to v-sync, but I'm not sure based on the different display architectures they discuss. I assume this is the 3rd type, no framebuffer on board, just blast video at the screen?

If you needed numbers to make selling the PCB work out, perhaps you can sell it in packs of two!

[goodl]

Id have at least four customers for you

[LaserEdge]

On that tangent, I've been wondering how to make sure two screens are v-sync'ed. I don't believe (but very happy to be proven wrong) that it's not possible from the computer/hdmi end of the chain. So that would mean some kind of timing lock/signal between two of your control boards synchronizing the two displays.

I read through (very briefly) the MIPI DCS spec you posted and it looks like it may be possible to v-sync, but I'm not sure based on the different display architectures they discuss. I assume this is the 3rd type, no framebuffer on board, just blast video at the screen?

If you needed numbers to make selling the PCB work out, perhaps you can sell it in packs of two!

MIPI-DSI supports both sync pulses and and sync events. The difference is events are done with short packets and sync pulses are done directly to the wires. All the 1080P panels that I have seen only support sync packets.

Since HDMI continuously streams the video data and there is no way to pause it a FIFO is needed for the video data stream that is ahead. It is pretty easy to do with parallel RGB. No data will be taken out of the FIFO until both VSyncs have been received.

[MSat]

First of all, congrats on the board, LaserEdge! Looks great! Using an external MCU seems like a good idea not just for the ability to have it control two separate interface boards, but it could also be used for interfacing with the motion sensors to create an IMU.

As far as sync goes, it seems HDMI 1.4a supports 48-bit "deep color" mode over two cables. That has me thinking - those two HDMI channels must be in sync in the first place, so a dual HDMI -> Parallel 48-bit interface IC (https://www.google.com/url?sa=t&rct=j&q ... 8961,d.aWM) can be used in place of the current HDMI -> 24-bit parallel device to drive two RGB -> MIPI devices. And instead of displaying the image in SBS mode, the images would be merged with one taking the upper 24-bits and the other would take the lower.

[Hannibalj2]

This thread keep getting more interesting. I am staying in China for work related reasons. I have looked for the Mipi converted, and nothing. So the option LaserEdge could be potentially offering could help greatly to DYI HMD's!

Definitely will be nice to see the Portal Dual HMD with dual Hd's display

[LuckyDog]

Here are some screenshots of the board connected to a LS050T1SX01. I choose to display this thread as evidence of it up and running. It works just like a mini-monitor so I can display anything I want on it. Its running at 1080x1920x60Hz 24bpp. The second screenshot is the panel magnified.

[LaserEdge]

First of all, congrats on the board, LaserEdge! Looks great! Using an external MCU seems like a good idea not just for the ability to have it control two separate interface boards, but it could also be used for interfacing with the motion sensors to create an IMU.

As far as sync goes, it seems HDMI 1.4a supports 48-bit "deep color" mode over two cables. That has me thinking - those two HDMI channels must be in sync in the first place, so a dual HDMI -> Parallel 48-bit interface IC (https://www.google.com/url?sa=t&rct=j&q ... 8961,d.aWM) can be used in place of the current HDMI -> 24-bit parallel device to drive two RGB -> MIPI devices. And instead of displaying the image in SBS mode, the images would be merged with one taking the upper 24-bits and the other would take the lower.

Hi MSat,
Thanks for the compliments. Yea. Having MCUs on their own board and communicate over I2C or SPI is generally a good idea.

The solution you proposed will not work. The output format is not compatible for two screens. More processing would be required to make it work.

The issue always goes back to the GPU. Is the GPU treating the output as a single monitor or two? I think it is a bad choice to go down the route having the GPU treat the two panels as one large monitor over HDMI. Long term we are aiming to move to OLED displays for their high refresh rate capabilities. Dual screen OLED at 1080P @ 120Hz will be very nice for HMD application. The solution I have in mind will take advantage 1080P @ 120Hz without adding much cost and have no latency increase.

[MSat]

The solution you proposed will not work. The output format is not compatible for two screens. More processing would be required to make it work.

The issue always goes back to the GPU. Is the GPU treating the output as a single monitor or two? I think it is a bad choice to go down the route having the GPU treat the two panels as one large monitor over HDMI. Long term we are aiming to move to OLED displays for their high refresh rate capabilities. Dual screen OLED at 1080P @ 120Hz will be very nice for HMD application. The solution I have in mind will take advantage 1080P @ 120Hz without adding much cost and have no latency increase.

The way I see it, the only processing necessary would be to combine the 24-bit 1080 stereo frames into one 48-bit 1080 frame which would be simple enough - just need to make sure the GPU supports HDMI 1.4a. Splitting the "48-bit" frame back into two separate 24-bit frames would be even easier - just route the 8-LSBs from each subpixel to one RGB->MIPI chip driving one panel, and the 8-MSBs to another RGB->MIPI for the second panel. The panels would always be in sync because they're working from one frame.

At any rate, is it known that multi-monitor outputs are not synchronized? After all, they're somewhat common for gaming.

I couldn't say whether it's better to treat the output as one display or two. If you're looking at long term, could you say HDMI will be a better option than DisplayPort or something like ThunderBolt?

[LaserEdge]

The way I see it, the only processing necessary would be to combine the 24-bit 1080 stereo frames into one 48-bit 1080 frame which would be simple enough - just need to make sure the GPU supports HDMI 1.4a. Splitting the "48-bit" frame back into two separate 24-bit frames would be even easier - just route the 8-LSBs from each subpixel to one RGB->MIPI chip driving one panel, and the 8-MSBs to another RGB->MIPI for the second panel. The panels would always be in sync because they're working from one frame.

At any rate, is it known that multi-monitor outputs are not synchronized? After all, they're somewhat common for gaming.

I couldn't say whether it's better to treat the output as one display or two. If you're looking at long term, could you say HDMI will be a better option than DisplayPort or something like ThunderBolt?

So you want to have the card output 24-bit 1080p60 stereo frame packing and then combine into one 48-bit 1080p60 non-stereo frame? If not what is the HDMI 1.4a output format the GPU will be using? Just want to be clear on what you have in mind.

Not sure about how wide spread knowledge of the problem exists. I first learned about it when looking at ways to measure the processing latency of monitors. The old way was to dual screen output a microsecond clock on a CRT monitor next to a LCD monitor and then take a short shutter shot of them together. It was noticed there was a large variability with this method and other less error prone solutions were desired.

I personally like DisplayPort from an engineering standpoint. But, the market has requirements that will make DisplayPort harder to adopt. DisplayPort will like make in roads mostly in the computer monitor market. HDMI will remain strong in the TV market and MIPI-DSI will become the main standard for mobile markets.

[LaserEdge]

Decided that the HDMI->MIPI board will be sold as a product to support the DIY community. Not ready to launch it yet. I want to do one more rev of the board to make it smaller, consume less power and fix the problems I have found with the first revision. Will start another thread in the next day or two so the community can discuss any desired features before I finalize the next rev.

[jquantum]

Decided that the HDMI->MIPI board will be sold as a product to support the DIY community. Not ready to launch it yet. I want to do one more rev of the board to make it smaller, consume less power and fix the problems I have found with the first revision. Will start another thread in the next day or two so the community can discuss any desired features before I finalize the next rev.

You sir, are beyond awesome This just solved my dilemma big time (iPad screen vs 5.6" 1280x800, resolution vs size)

TBH this actually just made my day (well finished the day off) now I can hardly sleep lol

I'll want a couple for myself to play with


Oh and did I say Thank you for the great and awesome work you've done

[studdart]

The way I see it, the only processing necessary would be to combine the 24-bit 1080 stereo frames into one 48-bit 1080 frame which would be simple enough - just need to make sure the GPU supports HDMI 1.4a. Splitting the "48-bit" frame back into two separate 24-bit frames would be even easier.

Though it's an intriguing idea, I'm not sure this really solves a problem that needs to be solved. We have an easy way to output to two screens right now, why make it more complex? Even if it's a simple idea to merge and split the image, it's going to add some complexity. Complexity adds to the size, weight, power and cost. Digi-key lists that chip at around $10, that doesn't include any other components that are needed to support it. Speaking of power, I'm having a little trouble adding the datasheet together, but it looks as if it'd take nearly all the 500ma of available USB power.

In my ideal world (that I'm too ignorant to actually create) two of LaserEdge's shiny PCBs could be connected or perhaps plugged into to same MCU and would v-sync. You could mount them where and how you like and use one or two depending on one's budget.

Decided that the HDMI->MIPI board will be sold as a product to support the DIY community. Not ready to launch it yet.

That's fantastic, I'm down for one, probably two. Any thoughts on my "ideal world" above? It would be quite fantastic to be able to chain the HDMI->MIPI boards together for genlock/vsync. Anyway, I await the thread and launch with excitement!

[MSat]

So you want to have the card output 24-bit 1080p60 stereo frame packing and then combine into one 48-bit 1080p60 non-stereo frame? If not what is the HDMI 1.4a output format the GPU will be using? Just want to be clear on what you have in mind.

I'm making assumptions and could be wrong about some of it as I never heard about Deep Color HDMI until earlier today. To correct a mistake I made earlier, Deep Color was introduced in HDMI spec 1.3 and not 1.4a. Also, you don't need two HDMI cables for it. Apparently, the PS3 supports it. Pulled from the web:

"HDMI 1.3 was released June 22, 2006 and increased the single-link bandwidth to 340 MHz (10.2 Gbit/s). It optionally supports Deep Color, with 30-bit, 36-bit, and 48-bit xvYCC, sRGB, or YCbCr, compared to 24-bit sRGB or YCbCr in previous HDMI versions. "


The way I see it, you have a left frame and a right frame rendered at 24-bit into their own respective frame buffers. Then, each is copied to a 3rd frame buffer at the same resolution but 48-bit color depth (assuming the GPU supports this - need to look into this). The way you would do that is to prefix all the subpixel (RGB) data bits from one frame with 8 '0s' and postfix subpixel data bits from the other frame with 8 '0s' and then OR them together.

Now two 24-bit frames are packed as one 48-bit frame at the proper display resolution. If you use an HDMI receiver that can output 48-bit RGB (such as the ADV7619 I posted earlier) "unpacking" them should be as simple as routing the upper and lower 8 bits per subpixel to the appropriate RGB->LVDS/MIPI chip and driving them both with the same clock and sync lines from the HDMI->RGB chip. This works because the received frame is at the native resolution of the LCD panels. No data buffering necessary, and no additional latency.