[opendtv] Re: Popular screen aspect ratios

  • From: Craig Birkmaier <craig@xxxxxxxxx>
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Sat, 15 Jan 2011 11:23:50 -0500

At 4:58 PM -0600 1/14/11, Manfredi, Albert E wrote:
When you set the STB to 4:3 or to 16:9, you are giving it either of two very specific display attributes. 

Correct!
If you set the STB to 4:3 it assumes it is delivering the decoded source to a legacy 4:3 display and that any aspect ratio accommodation will be performed by the STB.
If you set the STB to 16:9 it assumes that you are connecting the box to a DTV display. It does not know or care what the aspect ratio of the display is - that is the responsibility of the display processor in the TV. hence the same output worked just fin on my first 4:3 HDTV display AND on my present 16:9 display.
Like the ATSC, the MVPDs have assumed that widescreen means they are putting the source into a 16:9 container. The source "may" be 4:3 in pillarbox or something wider than 16:9 in letterbox. The TV then is set to display this content using a customer selected accommodation mode.
This was a practical solution, as intelligent interfaces that can deliver the resolution and aspect ratio metadata did not exist when the DTV transition began.
The computer industry, which has a far greater variety of displays to serve, chose to develop a solution based on the metadata that accompanies ANY source. Since all of this is taking place inside an intelligent device running software (and most likely hardware acceleration) for video streams, the metadata is used to automate the accommodation of the source to the display. 

THAT IS:
1. It knows the resolution and aspect ratio of the display;
2. It knows that the display uses square pixels;
3. It knows the resolution of the source samples and whether these samples are square pixels or some other defined sample aspect ratio. 4. And it knows how to scale the source samples to fill at least one axis of the display, with throw away areas for the rest.
IF the viewer wants to distort the image to fill the screen they can do this manually to override the automatic display accommodation. 



Again, PCs can ONLY do this because their monitors come with a CD or DVD.


Acnient history. Get a Mac and throw those discs away.
That disc enters the required settings into the video card. It's not like I haven't dealt with this time and time again. You get a new display, it might be some odd shape (like 1.25:1 or 16:10 or other oddity), the existing video card entries **don't** support that display shape, and you see **distortion**.
Using modern interconnect technologies like DVI or display port, all of this is negotiated between the display and the graphics card. I'm not sure why, but for some reason on Wintel PCs you may need to load a specific driver.
So the solution is, play the doggoned disc, then go back to the display settings, and like magic, there is/are the correct setting(s) suddenly in front of you. Now, you can either continue to see distorted images, or you can use one of the new settings available to you. That's how PCs do it, Craig. But TV STBs do NOT typically come with an optical disc reader or other such device.
Nor do they need such an archaic solution. Worst case, they could download a driver via their network settings. Of course this assumes some intelligence in the TV and a netwrok connection, both of which were announced by virtually every TV manufacturer at CES. 


 > My cable STB outputs a variety of formats via the analog component

 connectors; SD in both 4:3 and 16:9, 1280 x 720, and 1920 x 1080. The
 display processor in the TV deals with all of these, scaling SD up to
 1280 x 720 and 1920 x 1080 HD down to 1280 x 720. Since the STB assumes
 that the analog component connections are going to a DTV, it does
 expand 16:9 SD to 854 x 480 ands present 4:3 (640 x 480) in pillarbox,
 in a 854 x 480 raster.
Craig, try connecting the cable box via component, composite, or Y/C, and set the display aspect ratio in the cable box setup to 4:3. Now tell me whether you can undistort the images. Also, this limited set of options is what your "we don't need standards anymore" ignored. 

Why would I do this?
The 4:3 output is for legacy TVs. These devices will never have the ability to deal with multiple aspect ratios or to accommodate them.
Having said that, it IS possible to build external boxes that can handle any format and accommodate it to a display of any resolution and aspect ratio. The simple fact is that the CE and MVPD industries did not do this; they chose to support a limited number of formats.
The fact that the computer industry has solved this problem and can support any format should be sufficient evidence that this is the case.
It is obvious that different displays will paint their screens using whatever technique they need. A DLP screen will have to do different things internally than a CRT. Ditto for a non-square pixel LCD or plasma, compared with square pixel displays. Ditto for my 1366 X 768 display. That's obvious, but the fact is, the STB does not need to support all of the native screen pixel counts or painting techniques out there. That's the WHOLE POINT.
Yes, as I said above, the CE and TV industries chose the path of limited options. But it would add very little cost to do the job properly. Most STBs and DTVs are using similar, or the same GPUs as computers, and the computer market is now larger, and helping to drive down the price of these chips. If a computer can automate the process of accommodating any source to any display, then a STB or "Internet ready" TV can do the same.
We are arguing technology semantics and industry politics here. The underlying technology is converging.
Just one example. What if I wanted to create an image with 2.7:1 aspect ratio? Can I do that without standards? Heck no. Can I just blindly transmit that image, say over an analog interface, *even if* I assumed one of the standard options for number of horizontal lines? No. In order to make it work, I have to send that image either in a 4:3 or in a 16:9 container. And at best, I can squeeze it down horizontally by a factor of 1.77/1.33. That's it. I can't vaguely mention "MPEG-2 metadata" to the unwashed masses, and expect such a no-standards system to work.
A very poorly constructed example. The fact is that is is trivially easy to create 2.7:1 source at ANY resolution, encode it using h.264, and expect a computer to display it properly, filling the width of ANY display attached to that computer.
You may even be able to do this on your computer that is connected to your new LG display, at least for a single frame (otherwise known as a still image or photo).
It involves little more than starting with a high resolution image (i.e. oversampling), cropping to the desired aspect ratio, and scaling to the desired output resolution, which you can then save as a JPEG image. Any program that can display a JPEG can then fill the horizontal width of your display with that image. 

What has this got to do with video?
Video is simply a sequence of individual frames. The first non-linear editors that I helped to market for multiple companies used Motion JPEG for video compression. The symmetry of the JPEG algorithm, and the cheap cost of JPEG compression hardware made this a far more practical solution than using MPEG-1 or MPEG-2; video editing requires quick retrieval of any frame, which is more complicated when you must decode a GOP and recreate individual frames.
And MANY digital video imaging cameras today handle the oversampling - in fact this is no different than the good old days in the film industry, when if you wanted to create a new aspect ratio you put a new piece of glass in the viewfinder with graticules for the desired protected image area for that aspect ratio.
Just for the record, I used a non-linear editing system at the 1995 SMPTE Conference in San Francisco to demonstrate a technique for extracting a 16:9 image from 4:3 source. I called it tilt and scan; essentially the same as pan and scan except I moved a 16:9 window up and down within the 4:3 frame. Paramount like the idea so much that they continued to shoot TV episodic shows in high resolution 4:3 using tilt and scan to extract the 16:9 HD format, then scaling the 4:3 to SD resolution for the rest of the world.
Bottom line, the TV industry enjoyed a Golden Age, built atop a technology solution that was revolutionary (and appropriate) in the middle of the 20th Century; then they attempted to extend that business model by trying to control the evolution of digital technology to protect their legacy business model. In the short term this strategy has been successful, at least in terms of flattening out the product life cycle curve for broadcasting. In the long term it will be viewed for what it was: 

A side street leading to a cliff...

Regards
Craig


----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org 
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts:

  1. Home
  2. opendtv
  3. Archive
  4. 01-2011