The State of Video Connections

mikemuch writes "Joel Durham provides a nice rundown on what's happening in video interfaces as we leave VGA behind and move through the DVI flavors, visit HDMI along the way, and look forward to UDI and DisplayPort."

Comments

HDCP: it still sucks

[schwaang]

This article pimps UDI, which uses an HDMI-backwards compatible plug and can do higher bandwidth (10.8Gbps) and HDCP (copy protection enforcement).

Unfortunately, HDCP implementation sucks. Standard procedure for the problems almost everyone has with HDCP-enabled cable boxes is to *reboot the box*. Apparently, in the exchange of encryption keys a handshake sometimes gets dropped, and nobody has a firmware solution.

Of course, even it worked right, HDCP would still suck.

Re:HDCP: it still sucks

[Matthew Bafford]

You guys all must be using that Westinghouse example as an industry wide example now of failure.
They talked about how they would come to your house and fix it for you because they had the wrong timing for the handshake on there.

It's still a very real issue. I returned a Philips DVD player because it would consistently fail to renegotiate HDCP when I switched the TV's inputs off of the DVD player. Now, maybe this was just a timing problem and Philips failed to implement the specs correctly. Of course that's not HDCP's fault. It's entirely Philips's fault.

Still, no matter where the blame lies, the fact is it wouldn't be an issue AT ALL if HDCP didn't exist. It's because of HDCP that there are specs to implement incorrectly. I still don't buy the idea that HDCP is going to have any meaningful impact on piracy. It's just yet another inconvenience because the consortiums assume everyone is planning to do the wrong thing. It's insulting.

Important: Intel Opinion Center

[Anonymous Coward]

As I'm sure many of you have noticed, Intel and OSTG went into some kind of marketing deal with the Intel "Opinion Center" on Slashdot. There is nothing inherently wrong with that as all of the "stories" (rehashed press releases) were posted in Intel's own section; none of them were on the front page or in any of the other sections. AMD had a similar deal a while ago, but that appears to have been over for a while now. The strange thing about Intel's deal is that the link on the front page is in a somewhat prominent position and has a different color scheme in order to make it stand out. But what is more interesting is that the link IS NOT A DIRECT LINK. Instead it redirects through DoubleClick for some reason. I am not trying to make this sound sinister, but I found that a little odd.

Anyway, Intel posted a number of press releases and got a few comments here and there. But sometime last week they decided to get out of the deal. There is nothing wrong with that, but they DELETED all the previous stories and posted some lame excuse. Not that this means anything, but the comments on Intel's previous stories could still be viewed if you knew the exact url. In other words only the stories were deleted; the comments were not. This action generated a number of negative comments on the whole Intel "Opinion Center" idea. Today I went back to check on it and lo and behold they have DELETED ALL THE COMMENTS and marked the story as READ ONLY. While Slashdot claims that they can't or won't delete comments, I think it is pretty clear that things can be done if the price is right. Although I suppose we all already knew this from previous incidents, this time in particular it caught me by surprise. While a few of the comments were trolls, most of them voiced honest but negative opinions of the "Opinion Center". If you want to call it an "Opinion Center", then you should be ready to hear opinions. Otherwise just call a spade a spade: Intel "Marketing Center".

I never liked the idea in the first place, but deleting all the previous stories AND comments is really weak and speaks a lot about the integrity of both Intel and Slashdot. If you think Intel and Slashdot did the wrong thing here, please mod this post up.

Huh?

[JMZero]

Such cards were typically capable of addressing only 256K of local memory and displaying 256 colors at 640x480

My VGA card had 256k of RAM, and it did 640x480 at 16 colors. I wonder why...

640*480=307200
256k=262144 bytes

That's also why most early "VGA" games ran at 320x200x256. I understand that 640x480 is sometimes referred to as VGA regardless of color depth, but that doesn't seem to be what he's doing here.

One cable to rule them all

[VoltageX]

I wish they'd hurry up and standardise the damn things. I just bought a Chimei LCD and the cable supplied is a DVI-I to DVI-I but my video card (Xpertvision Geforce 6600GT) has a DVI-D port, and for the life of me I can't find a shop here in Australia that sells a DVI-I to DVI-D cable! I can see why so many people don't like computers. Standards like SATA (small cables!) and USB (plug just about anything in) are going the right way. Hey, why couldn't we use USB2, wouldn't ~400mbits be enough?

Re:Print Version

[Second_Derivative]

Oh but come on, they couldn't let you print without leaving that oh-so-lucrative product placement in there

Jesus christ. Why does absolute sewage like this get posted?? It's about five paragraphs per page and three or so screens' worth of useless garbage.

Re:article text to avoid annoying 6 pages

[drewzhrodague]

Yup. Analog here. Many tubes, and an analog NEC projector. HDMI can go screw itself. I'll be analog with coax until anything else is far far far cheaper (or better).

What's the straight dope on HDMI?

[Matey-O]

I've heard from the home theater folks that HDMI was a seriously broken implementation. v1.1 wasn't necessarily compatible from device to device, v1.2 only carried stereo, and at the time I was in the ,market, only the PS3 used v 1.3....and they weren't necessarily backwards compatible.

They ended up with the comment that the video quality wasn't up there with component.

So, were they blowing sunshine up my skirt, or is HDMI really the tarpit they describe?

Plus....

[shmlco]

"Why do we need another display connector?"

If you move into TV-land you also have coaxial, composite, s-video, component, and HDMI, as well as 1/8 and 1/4" phone jacks, RCA, digital-coax, and digital-optical for audio.

My personal theory [putting on tinfoil hat] is that's it's all a vast conspiracy by the cable and connector manuafactuers. Every new connector requires new cables, adaptors, and, in the end, replacing "obsolete" equipment that can no longer talk to one other.

And why does an optical or HDMI cable of sufficient length end up costing more than most DVD players? It's a CABLE for Pete's sake.

Re:Plus....

[shmlco]

S-video carries the video data as two separate signals (brightness and color), unlike composite video which carries the entire set of signals in one signal line. S-video is better. Component typically breaks them out to three different RGB or Y'PbPr signals. Component beats s-video hands down. All are analog.

HDMI is a all-digital Transition Minimized Differential Signaling (TMDS) method that also includes audio data. HDMI beats component on todays "digital" LCD and plasma screens when used with digital sources like cable or DVDs, as you're not converting from digital to analog back to a digital matrix.

I try to make sure everything I buy today has HDMI inputs/outputs, as the single cable is "cleaner" and easier to route.

Re:Piss off!

[RzUpAnmsCwrds]

How about just analog RGB and quit pretending we need digital connections at all?

How about we stop pretending that analog RGB looks good? Ever try screwing with the contrast setting on an LCD? That's analog technology at work.

DVI lets me see the image outputted by my graphics card - pixel and value precise. Neither my monitor nor my graphics card supports HDCP, so DRM isn't a problem.

As a public service, let me remind you that high-bandwidth analog signals are problematic. It doesn't take much for noise, crosstalk, or other issues to show up on an analog monitor at high resolutions.

Try connecting your monitor to your desktop with a 20 foot DVI cable - then try doing the same thing with an analog RGB cable.

Try using a crappy KVM. Most screw up resolutions greater than 1600x1200.

Analog is the reason my cable signal looks like shit. It's the reason why broadcast TV looks crappy. It's the reason why AMPS cellphones have static.

So, hell, why shouldn't we take a nice clean digital signal, run it throguh a DAC, throw it through a cable, and try to reconstruct it into a digital signal with an ADC at the other end. Extra components, extra complexity, and more chances for interference. What a great idea.

DVI and KVM switches..

[antdude]

What bugs me about DVI is that KVM with them are still expensive. I am still using VGA with my old Belkin OmniCube KVM switches that I bought back in 2001.

Re:Piss off!

[fyngyrz]

Really? You mean pixels on an LCD aren't signaled in a (more or less) similar way as cells in a RAM chip? I'm no electrical engineer, but I would have assumed...

There are two issues that relate directly to your concern. One is addressing; in order to pick a ram location or a screen pixel, a signal needs to be sent to the particular location that says "hey you, and not any other." Displays and RAM can be similar in this regard, though it is more likely that a simpler scheme of sequential counters is used to drive the selections, as the signal is dependably progressive. This is always done digitally, as far as I know. This step, however, does not process the actual video levels, it routes them to a particular display location.

The other issue is the brightness of the cell triplets, and that is, in the end, an analog issue within the LCD cells. Several approaches are possible; none of the really easy ones would have a DA converter at every cell, though, which would make the access arrangement essentially similar to a multi-bit ram chip (though still analog internally, there's no way out of that.) So typically three analog signals (RGB) are routed via analog gates to neighboring LCD cells and these are either selected via a simple sequential X:Y matrix along rows and columns, or via a more complex full addressing scheme. The three signals are used to tell the LCD cell triplet how opaque (or transparent) the crystals should be until they receive their next regular update. The light behind the cells of a relatively constant brightness passes through the three cells in a ratio determined by the opacity/transparency of those cells, and the resulting color is determined in additive color space: red plus green = yellow, and so forth as it reaches your eyes, which are very good general purpose color adders. If you look very closely, you can see the triplets. You may need a loupe; on modern LCDs, they're amazingly small.

Here's a model of one way to connect all this stuff together. Run RGB analog and power signals to all the triplets on the screen, just one after another. Run a single X-select line for each column of the display; run a single Y-select line for each row of the display. Now, if you think about this, you'll realize that if one X line is turned on, and one Y line is turned on, there will be only one triplet that has BOTH the X and Y lines turned on. That triplet has just enough circuitry to open a "gate" and take in the RGB information, storing it temporarily and using it to drive the liquid crystals appropriately. Now all it takes to make that work as a display is to sequence the X and Y lines across the display in an orderly fashion. You can even rotate the display by simply switching the X with the Y lines, or switching the drives and "next row/column" outputs of the counters that sequence them, more likely. The first requirement for this general approach is that the LCD triplet store the RGB value, and the related LCD opacity/transparency, until it is next accessed - about a 60th of a second would not be unusual. The second is that the triplet needs to be able to change fairly quickly when it is told to, or the display will appear to "blur" as the picture sends information that conveys motion, but the triplets hold on to previous brightness settings for too long, resulting in motion blur. The third is the more accurately the levels convert to opacity, the more color gradations can be achieved. Finally, and this has always been the achilles heel of LCDs, how much light can be blocked in the most opaque state and how much light gets through in the most transparent state determines the ultimate contrast ratio. There are plenty of other issues - operating temperature range and its effect on response time and depth, cell density, alternate color addition schemes, response to physical pressure on the LCD cells, power consumption - but they're not critical to understanding how the general approach works.

I keep saying "opaque/transparent"

we don't, we need a *device* connector

[RMH101]

...what I want is for my satellite box, dvd, amplifier, projector etc to all hook up intelligently to each other: no multiple remotes, no daisy-chaining scart cables, no having to switch the TV to AV1 and the amp to DVD each time - I want it to just work. I want an interconnect that can do video, audio AND control devices. I want my DVR to change my satellite box to the right channel at the right time without messing around with IR blasters and the like. No-one wants complexity, yet look at the average geeky AV setup.
I also don't want to have to buy all my AV gear from the same manufacturer to get any of this: there should be an open standard for this.