Shortchanging HDTV resolution

[edtyct]

Gary Merson, a long-time writer about conumer electronics and the pubisher of HDTV Insider on the web, recently completely a study revealing that many mfgrs have been bobbing 1080i signals--that is, upconverting each 540-line field to a display's native resolution (720p, 768p,1080i, or 1080p) rather than buffering the two 540-line fields to construct a proper 1080p frame before releasing them to the TV screen. The result can be a shortfall in the detail that consumers expected to get, even though they may not have noticed it (Merson claims that he can see the dfference). The situation has apparently improved with the 2005 sets, but 2004 was rife with this sort of bobbing. As you might expect, 1080p sets that go this route are particularly at fault, since consumers often bought them to convert 1080i signals, not 540 vertical lines, to 1080p (Sony's LCoS sets don't bob). The companies that failed the test consistently were Sharp and LG; the mixed bags were Sony, Panasonic,and Samsung (all of its DLPs). JVC, Hitachi, Pioneer, and Toshiba had no failures (except for one out-sourced Toshiba). Faroudja/Genesis and PixelWorks "admitted" that some of their chipsets worked in this way. A few companies have disputed Merson's findings for certain sets--Sony and LG, for example.

The seriousness of this problem is up for grabs. I certainly don't think that it falls squarely into the category of false advertising. After all, scaling and deinterlacing are almost as much art as exact science. But I don't think that it is just a semantic problem either, or a choice between equal methods. A 1080i signal is not simply two independent 540-line signals; it is one coherent unit. To divide it up into discrete pieces the easier to deal with it is to distort it. Merson's study does indicate that cost is a factor; the sets that don't bob are more expensive. At the very least, mfgs should reveal which kind of scaling/ deinterlacing option they use. After all, we know the difference between TI's HD2, HD3, and HD4 chips. We should be equally aware of the disadvantages, as well as any advantages (like cost), of the bobbing chips.

Ed

Anyway, more detail on the study is available at hdtvinsider.com, and in this installlment of The Perfect Vision. I'm not usre whether TPV's website, AVguide.com, contains the info.

[GMichael]

Can you tell me what the differences are between bobbing 1080i signals, rather than buffering them?

[edtyct]

You mean visually? Bobbed signals would be less sharp, and they might have more motion artifacts, like jagged lines, because of the degree of separation from the integral frame. If you mean technically, what happens in the buffer with true 1080i processing is that the TV processes both 540 fields together, combining them as if they were to be displayed as a single progressive 1080 frame, before the scaler takes that frame and turns it into the display's native resolution. Bobbing, however, scales each 540 field independently of its legitimate complement. In other words, bobbing treats each half of the 1080 frame as it should treat the whole 1080 frame, thereby reducing the effective resolution to 540 (which eventually gets scaled to 720p, 768p, 1080i, or 1080p). Each 540 field, not the entire 1080 lines of information, is processed to reach the native resolution of the display,.

[GMichael]

Thanks, that makes sense.

[cam]

Gary Merson, a long-time writer about conumer electronics and the pubisher of HDTV Insider on the web, recently completely a study revealing that many mfgrs have been bobbing 1080i signals--that is, upconverting each 540-line field to a display's native resolution (720p, 768p,1080i, or 1080p) rather than buffering the two 540-line fields to construct a proper 1080p frame before releasing them to the TV screen. The result can be a shortfall in the detail that consumers expected to get, even though they may not have noticed it (Merson claims that he can see the dfference). The situation has apparently improved with the 2005 sets, but 2004 was rife with this sort of bobbing. As you might expect, 1080p sets that go this route are particularly at fault, since consumers often bought them to convert 1080i signals, not 540 vertical lines, to 1080p (Sony's LCoS sets don't bob). The companies that failed the test consistently were Sharp and LG; the mixed bags were Sony, Panasonic,and Samsung (all of its DLPs). JVC, Hitachi, Pioneer, and Toshiba had no failures (except for one out-sourced Toshiba). Faroudja/Genesis and PixelWorks "admitted" that some of their chipsets worked in this way. A few companies have disputed Merson's findings for certain sets--Sony and LG, for example.

The seriousness of this problem is up for grabs. I certainly don't think that it falls squarely into the category of false advertising. After all, scaling and deinterlacing are almost as much art as exact science. But I don't think that it is just a semantic problem either, or a choice between equal methods. A 1080i signal is not simply two independent 540-line signals; it is one coherent unit. To divide it up into discrete pieces the easier to deal with it is to distort it. Merson's study does indicate that cost is a factor; the sets that don't bob are more expensive. At the very least, mfgs should reveal which kind of scaling/ deinterlacing option they use. After all, we know the difference between TI's HD2, HD3, and HD4 chips. We should be equally aware of the disadvantages, as well as any advantages (like cost), of the bobbing chips.

Ed

Anyway, more detail on the study is available at hdtvinsider.com, and in this installlment of The Perfect Vision. I'm not usre whether TPV's website, AVguide.com, contains the info.

Hey Ed, I don't know the differences between TI's chips. All I have heard is that the HD2+ is better then the HD3. And I haven't heard anything about the HD4 chips improvements over the HD2+. Could you take the time to explain to the less knowledgeable.

[evil__betty]

Hey Ed, I don't know the differences between TI's chips. All I have heard is that the HD2+ is better then the HD3. And I haven't heard anything about the HD4 chips improvements over the HD2+. Could you take the time to explain to the less knowledgeable.

The differences between the new TI chips are vitually none - at least no visibly. They have tweaked the chip slightly, and fixed anything that they found wrong with the previous year's chip. There is a big difference between the regualy '720p' chip and the new '1080p' chip. The DLP's that are 1080p use a method called "wobulation" (see S&V's explanation here) I wouldn't loose sleep over any of the minor differences between the HD2+, HD3 or HD4 chips that are being used. The TV's own processing engine will make more difference in the PQ than the chip will.

[edtyct]

The good things about the HD3 chip is that it reduces the price of DLPs and that it goes darker and is smaller than the HD2 chip. I beleieve that it also has an extra segment on the color wheel. The consensus, however, is that the HD2+ chip is the jewel in TI's crown at the moment, so far as contrast goes and distinct resolution. The HD3 chip wobulates to 1280x720 from 640x720, thus explaining the reduction in size and price. I'll leave it up to individuals about whether they can see the difference between the wobulators and the true 1280x720s. The HD4 chip wobulates 1920x1080p using half the horizontal resolution. I haven't seen enough of them yet, but I'm not sold that this type of chip will look as good as DLPs, LCDs, or LCoSs that have the full complement of pixels.

[cam]

Thanks ed. The HD4 chip you state wobulates 1920x1080p, but I check out Toshibas website and they have the HD4 chip on all their 720p dlp's also.

[edtyct]

Cam,

I'm not quite sure that I get what HD4 is all about. The xHD4 is a wobulating 1080p, but I can't tell whether the HD4 without the "x" wobulates at 720p or not. In other words, I don't know whether the distinguishing characteristic of any set with HD4 technology is wobulation at either resolution or a particular change to the mirrors and/or electronics. The contrast numbers go up with every generation; maybe that's the identifying characteristic for all "HD4" sets. As I said, I don't know. Maybe you do.

Ed