That's pretty awesome, and a pretty grim testament to the lameness of Laptop makers if they aren't willing to put that panel in a laptop! (I mean seriously, you can't pull the display port connection out of an ivy bridge chipset?
Something I have always felt there would be a good market for would be a compact an inexpensive LVDS (TFT-LCD connector) to some sort of driver (small FPGA or maybe SOC). You could then salvage displays from "dead" laptops (where the display works) and build up some pretty awesome sorts of deca-panel UIs :-) I'm thinking 'home star trek bridge' to a whole new level.
Of course you can pull the DP connections out, that's not the problem. The problem is that when you put that display on a windows machine, you end up with a bunch of legacy windows software and web graphics content that looks unreadably small, followed unhappy users who return those fancy DP panels to your retailers.
Of course, the same thing is true of iOS and OS X software, which is precisely the reason "Retina" (as distinguished from arbitrary 200+ DPI hardware) displays are always shipped in resolutions that exactly double those of pre-existing devices. So even though the legacy software doesn't handle it Apple, by controlling the framework, is able to make it work more or less seamlessly.
Dell and HP don't have that freedom, they need help from Microsoft, and Microsoft, well... At least Windows 8 does high DPI natively, though the MS decision was basically to jettison legacy visuals and hope no one runs the old software anymore.
I have computers in datacenters that I can effortlessly SSH into when I am feeling the (frequent and chronic) urge to Dick About With Computers And/Or Software.
However, the one that connects to my display(s) is simply a tool for running Chrome, Skype, FaceTime, Mail.app, TweetDeck, Spotify, and SSH, and, as such, should not ever break or otherwise hinder the incredibly simple tasks which ARE ITS ONLY JOB.
This is why the only computers I own that I touch on a regular basis all have fruit on the back. It's a very simple calculus.
The same reasons [0] are why I have a Chromebook. CrOS is limited, sure, but the limits make the thing bloody near invincible (and amazingly productive).
[0] Hangouts are luckily the VoIP of choice for most I work with
Think about what your laptop is doing that we would have had no hope of doing with desktop Linux just a few years ago (if ever, depending on the hardware).
Sharing access to video capture and audio hardware between software written (sort of) by Apple, Microsoft (Skype, so -ish), Spotify, etc. would have been unthinkable. You would have been restarting your primary machine if you made the mistake of starting any two apps that tried to grab your QuickCam at the same time.
<-- Down votes go here and by all means, please don't respond.
Interesting. My personal workstation runs Windows and I too effortlessly SSH and RDP into many headless or virtual machines to dick around and (occasionally) to do actual work. Hey, I also run Chrome, Skype, TweetDeck, Spotify, SSH and the like and I have been for years...on the same install of Windows. It's like your calculus doesn't work at all in my universe.
Actually...I don't think you're doing calculus at all.
Also, have you ever seen the source of Chrome? It's not that simple. Lay-people think it's simple, but software developers know better.
"A browser" is simple from a "a process that talks to the network and draws things on screen" standpoint. I'm not suggesting that Chrome isn't a complex application - just that it's self-contained and very well tested and doesn't do anything special with the OS or hardware (drivers, USB, video modes, etc).
My workflow would work on Windows, too, aside from the fact that it'd suck. (Example: Unplug an external USB audio interface while a song is playing. OSX skips a beat and then plays via internal speakers, at the previous internal-speaker-volume-level.)
I could travel with luggage that isn't Pelican/Incase, too— but I don't.
I laughed at the "incredibly simple" bit myself. I'm sure it must seem that way to people who live at the top of the stack doing javascript hackery and MongoDB maintenance. It's amusing how few people, even in this world of open source everywhere, have ever bothered to do something as simple (in an only-slightly-ironic sense of "simple") as build an OS distribution. Android might be the "simplest" of this kind of thing, and it's still a 10G+ tree.
I had a Thinkpad with a 15.4-inch 1920x1200 screen, back when such wonders existed, and I had no big problems with its 150 DPI running Windows XP, even less with Windows 7. Yes, there were some really old applications that wouldn't react to system-wide DPI settings, but that was a small minority.
I think there is simply no excuse for the regression in laptop resolutions - I can't even find a laptop that will give me 1200 lines anymore, they give me "full HD" at best, which is less. And we can see proof in all the comments saying that they would buy this guy's hack to plug a hi-res screen into a laptop.
I had a Compaq in 2004 with the same display. And indeed, windows works pretty well up to about 150-170 DPI. But I have a 10" Acer tablet with a 1920x1080 screen sitting next to me, and the legacy windows desktop is basically unusable. Yes, it can be made to work if you are tolerant of VERY SMALL BUTTONS, largely unreadable text in graphics assets, and know where to find all the font settings. But it pretty much sucks, and no one sane would try to use it that way.
The problem are legacy apps. The Surface Pro is at 1.6X, so things begin to look small unless the app can be upscaled either by expanding the pixels (in which case, 1.6x is extremely inconvenient as not being a power of 2) or have the resolution natively baked into the app (not many yet).
2X (as in the Toshiba) is much easier to upscale transparently (2x being a power of 2), but its not really valuable until apps start supporting that resolution natively.
Edit: as pointed out below, N just has to be an integer, not a power of 2.
I've got a Surface Pro and I actually run the Win32/Desktop release of Chrome without scaling.
What legacy app do you think won't work well? I'll be glad to try it out. Everything I've thrown at it so far has worked great. Even the desktop apps work well with touch or the pen.
Visual Studio (2012) is one of the first things I installed and it works great.
Some other desktop apps I have installed: VLC Media Player, Notepad++ and Balsamiq Mockups. All work well but you are forced to use the pen (or type cover/touchpad) at times to grab little drag handles (if you're trying just to use touch).
Does Visual Studio upscale automatically according to your DPI setting or do you have to fiddle with the font settings? Also, is it possible to increase the font size for the menu fonts?
I didn't change any settings in Visual Studio after installing on my Surface Pro and everything looks fine to me. I'm embarrassed to admit that I really haven't peeked into what's going on with the DPI settings (because how it came out of the box has been perfectly fine for 99% of what I've been running on it), but I believe it's set at 150% DPI and VS respects that throughout.
Nice! I've been wanting to know if Visual studio was resolution independent or not. Being implemented in WPF, this was always a possibility, but I could find no evidence on the web.
Like people mention above, the hardware isn't the only impediment; Apple's control is what helps them a lot with this. Desktop Windows' support for DPI scaling is anemic at best, and this is only partially fixed in Windows 8 by having ad-hoc dimensions per-device, or on PCs just "big" vs. "really big" mode.
I put a macbook in a $39 supermicro 1u case and took it to a colo so I wouldn't have to buy an xserve to run one cheesy little osx app. It worked great.
> Something I have always felt there would be a good market for would be a compact an inexpensive LVDS (TFT-LCD connector) to some sort of driver (small FPGA or maybe SOC).
search ebay for "lvds dvi" (without quotes), you'll find inexpensive LVDS drivers. For about 30 bucks you get DVI in, LVDS out.
Does anyone know if there is some company I could send off a schematic of the adapter he uses and have one sent to me? (This would be UK, but I have no idea what term I'm searching for to get a single circuit board made)
PCB-Pool ( http://www.pcb-pool.com ) is a big player and should be shipping to the UK. But they are on the expensive side (production in germany).
Batch-PCB ( http://www.batchpcb.com/ ) takes longer (they have the boards made in china, at a reputable manufacurer), but cheaper.
Itead Studio ( http://imall.iteadstudio.com/ ) China, Very cheap (they buy empty spaces in fabs, you get 5 (or 10) PCBs buts its possible that one or two arent working.
I have no personal experience with any of those companies, though. Small batch PCB manufacturing is always expensive. Get in touch with a local hackerspace and etch it yourself ;-)
You'll have to buy the connectors and solder them on yourself.
It sounds like an adapter kit might have a little market when shipped with a displayport plug, a small power supply, and a fully assembled PCB. Small kickstarter?
You'll have to buy the connectors and solder them on yourself.
Just FYI, those SMT flexible-printed-circuit connectors are a bitch to get soldered onto a PCB correctly. You might survive with a lot of flux and a very thin tip (and a microscope since the plastic will melt at a pretty low temp. No mistakes allowed).
You might do a little better with a hot air gun. But this is not level-1 soldering.
OSH Park does an outstanding job. I've used them for work and hobby prototyping, and while the turnaround time isn't great (week or two) the quality is first rate.
From what I gather the boards are actually made by Amitron outside Chicago.
Unlike most quickturn PCB fabs they don't bury you with options. The standard options give you everything you need (ENIG, double mask+silk screen, either 2/4 layer, etc.). The OSH Park ordering wizard is a case study in how PCB orders should be done, and for the quality the price can't be beat. Best of all, it's just a "Fred in a shed" operation in Oregon that came out of a hackerspace there.
seeed studio (3 E's in seeed) makes PCBs, starting around $20 for 10. I've ordered a few from them, always good quality but shipping from China is a little slow.
I thought seriously about building and selling something similar to this about 4 years ago - really cool to see someone hack it together. Awesome work Andrzej!
Compact, portable screens seem like a no-brainer to me. My MBA display is less than half an inch thick. Why does my desktop screen need to be significantly thicker? Why not use the same panels from laptops to build a display that can be used both on the desktop and for portable use? I love to use one large/two smaller screens when coding - and I want a display that I can pack into my bag along with my main display and use at a coffee shop. The panels themselves are generally cheap (it's something of a commodity market, but depending on the display size/specs they run $35 to $90)
When I originally started researching this, there were two problems, both of which have been (pretty much) solved:
1 - Connectivity. Laptop panels generally use LVDS, which from my understanding are bit-reversed versions of DVI, possibly with some extra baggage. So you needed to have some intermediary processing, which was then (and is still) possible using a DisplayLink chip. The DisplayLink chip basically uses CPU power to create a virtual DVI/LVDS port over USB. The advantage of DisplayLink is that it negates the necessity for even using the Thunderbolt/DVI/Displayport output of your laptop, and allows you to simply plug in via USB and go. But the chips cost ~$15 each, require non-trivial integration, and use up CPU. As Andrzej mentions, displays are moving towards eDisplayPort, which means you can pretty much plug them in directly.
2 - Power draw. For the display to work well as a portable device, it should be possible to power it via USB alone. USB 2.0 standard is 5V and 0.5A, so 2.5W, which isn't sufficient. But USB 3.0 is 0.9A, bring the total up to 4.5W. As Andrzej mentions, the iPad display can be powered off the Displayport power (3.3W, 0.5A, ~1.9W), and with a step-up convertor you should just about be able to power the backlight up to max brightness from the USB port.
Both those problems are now fixed. You can now build an external display, either using an iPad panel, or that from a 13" MBA or (possibly that of a MBP Retina, not sure about the power requirements there). Then you have an extremely slim, awesome external display that's completely portable and powered by USB.
Some companies have been building crappy versions of this using DisplayLink chips, but I really want to see something with a clean, minimal design and way higher resolution. I'd try at it myself, but I'm busy trying to build software to enable better online political activism. If anyone out there is interested, I'd love to help, either with funding or advice.
A few examples of products that do this (not very well) using DisplayLink chips:
DisplayLink is a horror show these days. Their 1xxx chipset was embraced by Linux users and well supported, and it looked like the romance was going to continue- the website and announcement for the newer USB3.0 chipset declared Linux support.
However DisplayLink now uses an encrypted signal and has shown nothing but contempt for Linux, declaring that they're not satisfied with the state of affairs and unwilling to do anything to bring support to Linux. Which is a shame, because DMA-PRIME was more or less invented explicitly for DisplayLink's sake, to allow these external displays to be backed by the primary video card and have it's buffers shipped over to the DisplayLink, and DisplayLink packed it's toys, sealed the documentation, and left the party. The USB3.0 product which was launched advertising Linux support has never materialized any support.
http://displaylink.org/forum/showthread.php?t=1748&page=...
If one is interested in attaching displays these days, I recommend RTFA, and using eDP, embedded displayport. It's built in to monitors, and it'll plug into the badass display protocol of today that nearly all GPU's expose, and often run through PHY's to convert into lesser protocols.
Hopefully we see more use of the Aux channel in DisplayPort, for transporting USB or ethernet. And hopefully we see some display controllers for monitors coming out that support daisy chaining, so we can connect multiple displays to a single DisplayPort without a hub.
Not yet I'm afraid. My pet project is http://fixthedmca.org. I'm working on first implementing http://trydiscourse.org to help get the community engaged and communicating and building tools like click-to-call legislators to help drive action. I'll be abstracting out and offering the tools to other campaigns (anti-CISPA, patent reform, etc) as quickly as possible.
At the moment it's just me, so a little developer strapped as I try and both code and navigate getting an unlocking (+ DMCA hearings) bill through Congress. If anyone's interested in helping out, please please get in touch.
AMD's been shipping even their basic mobile chipsets with ~4 eDP for a while. Alas, laptop manufacturers are all cheap bastards, and only bother bringing one maybe two of these links to external ports.
Their discrete mobile GPUs and embedded GPUs still have all six eDP ports as found on their desktop kit. Would that someone step up with a laptop that brought 5 of them out the back! It's a niche feature, but a ragingly awesome niche feature.
It's not quite as easy, because if you're using, say, Intel graphics, you get maybe two clock signals, so you'd likely have to use your own. (Although you said AMD and discrete mobile GPUs :-))
In addition, the laptops that use eDP (Chromebook Pixel, which uses https://www.revo-sys.com/prodimages/LCD-Display-Panel/LG-12....) are using those lanes (and probably the aux lanes) for the primary touchscreen display. The external display interface is going to be pretty vanilla (mini-DP IIRC).
Next step: make four of them and get nice 20" 4096x3072 desktop. Of course you'd need to plan your DE to suit the tiled nature, but I imagine that it wouldn't be that bad.
Woah, $55 for that panel? Seems like it costs less than $100 including the PCB stuff, I'm really tempted to do this but I don't have the time :(. Heck, I'd buy a reasonably finished version of this for $150, maybe $200, without thinking too much. That's pretty awesome.
What's the cheapest most compact PC you could drive this level of display with? One could imagine packing them together to put together a open source, absurdly cheap (albeit bulky) iPad wannabe.
Minis can be easily disassembled so if I were you I'd try those ifixit reviews where they do it and compare the connectors. Chances are still unknown though.
I tried Xdmx a few months ago. It failed pretty badly for me, so I blogged about it ( http://idupree.dreamwidth.org/2151.html ; choice quote: "Xdmx is broken in various different ways for various different people on the Web. Xdmx isn't very well maintained. It doesn't support XInput2 [1], it relies on Xinerama which is somewhat maintained but whose architecture needs improvement [2], it may be incompatible with Composite [3], it just plain segfaults for a lot of people [4], and is generally unloved [5]." — citations in blog post.)
Or you could avoid all the trouble and bad a Surface RT or Pro, isn't it? Compatible out of the box with thousands of printers, scanner and every kind of peripheral you can possibly imagine.
Something I have always felt there would be a good market for would be a compact an inexpensive LVDS (TFT-LCD connector) to some sort of driver (small FPGA or maybe SOC). You could then salvage displays from "dead" laptops (where the display works) and build up some pretty awesome sorts of deca-panel UIs :-) I'm thinking 'home star trek bridge' to a whole new level.