Just watched the TCU game (go Frogs) with great video. When the switch was made to the Oklahoma game on the same channel the video quality was only marginally better than regular TV? What gives?
DirecTV Video Quality
- Thread starter audioguy
- Start date
Steve Williams
Site Founder, Site Co-Owner, Administrator
Probably clouds or other interference between your antenna and the satellite which wasn't there before
Satellite companies deploy what is called "statistical multiplexing" or "stat mux'ing" for short. What this means is that the take they total bandwidth available and divide it based on:
1. What is precisely happening at that moment in video. If there is a peak of action requiring more bits, the data rate of another channel with easier to compress video is reduced and the extra bandwidth is given to this channel.
2. Priority of what makes money. A PPV sports channel will likely get far more bandwidth regardless of #1 as oppose to say, Turner Classics.
So at any one moment, the fidelity you get is unpredictable on the same channel and channel to channel. The system is constantly shuffling bits from one place to the other, attempting to optimize the above two factors. It is possible for example that some other important event came on a different channels and more bits where shifted to it, than the channel you were watching.
1. What is precisely happening at that moment in video. If there is a peak of action requiring more bits, the data rate of another channel with easier to compress video is reduced and the extra bandwidth is given to this channel.
2. Priority of what makes money. A PPV sports channel will likely get far more bandwidth regardless of #1 as oppose to say, Turner Classics.
So at any one moment, the fidelity you get is unpredictable on the same channel and channel to channel. The system is constantly shuffling bits from one place to the other, attempting to optimize the above two factors. It is possible for example that some other important event came on a different channels and more bits where shifted to it, than the channel you were watching.
Steve Williams
Site Founder, Site Co-Owner, Administrator
I've found DTV quality unpredictable even at different times on the same station. I knew there's a bandwidth issue, and some stations get more or less compression, but that doesn't fully explain the phenomenon. Great info. Amirm., makes sense of what I've experienced with DTV over the years.
Can there be a straightforward difference in the quality of cameras, ambient conditions for filming, and technical adjustments of the camera/broadcast equipment that would affect PQ as well?
Lee
Lee
It isn't that. You'll see the same show of a series look different. Particularly obvious sometimes is with the NBA stations -- it's always the same cameras in Chicago Stadium, and the local feed as seen on DTV can look very different from one game to the next. Sometimes barely SD, other times HD as you'd expect. You can see occasional compression artifacts as well on bad nights.
Oh. The original post mentioned switching from one channel to another (one game to another). As to the bandwidth issue, definitely. DirecTV even was shutting down some of its channels/stations when there was a high-demand event playing to preserve bandwidth.
EDITED TO ADD: NEVER POST LATE AT NIGHT WHEN TIRED.
Lee
EDITED TO ADD: NEVER POST LATE AT NIGHT WHEN TIRED.
Lee
Last edited:
Steve Williams
Site Founder, Site Co-Owner, Administrator
Steve-with digital tv, you either get a *perfect* signal or you get nothing. Unlike analog tv where you could receive weak signals with all sort of whacky artifacts, with digital it just isn't that way. Notice if you have direct tv and it starts raining hard or snowing hard-good-bye picture. It doesn't degrade, it just dsappears. So what you were seeing when your high-def picture degrade from one show to the next is just compression artifacts. The best high-def picture I think is what you can pull off from an outside high-def antenna from your local stations. They don't have the compression that you see from digtial that has been compressed to fit bandwidth concerns on a satellite.
I talked with a friend who is an engineer at Comcast. I can't speak to DirecTV or any other cable system but suspect it is similar. I'll paraphrase the discussion we had including a little of both our knowledge/opinions. This is all IMO, FYI, FWIWFM, blah blah blah - Don
1. Data dropouts do happen with digital signals and can cause artifacts, primarily pixelation and pixel dropouts when the EC (error correction) can't keep up. The degradation tends to be more noticeable than "minor" analog artifacts resulting in a loss of sharpening. In a digital feed, this may be a sharp-edged group of pixels that look out of place. Digital signals can generally operate with much lower SNR than analog before noticeable degradation occurs but, and partly because of that, it is true that digital tends to swiftly go from "good" to "gone" where analog goes through a host of artifacts before total signal loss.
2. Bandwidth allocation is dynamic, at least quasi-dynamic, and they can choose what compression level to use for any given show (and even within a time slot, e.g. the game and commercials may have different compression levels). So, they can choose to provide a highly-anticipated (read: well-funded by commercial sponsors) show at higher quality than another with lower audience (funding) to save (or, re-allocate) bandwidth.
3. There is some variation in quality from the content providers, by their choice and outside Comcast's control. Comcast can choose to further compress (or even upconvert) once they receive it, of course.
4. It used to be true in the analog days that quite often the best performance was from a local (big-city) newscast as the signal was local (less attenuated) and well-monitored by on-site engineers. The news was that local stations "flagship" program, after all. There's still some truth in that, though a lot of stations are still in transition with a mix of new and older equipment due to the extremely high cost of HD cameras, boards, etc.
5. Virtually all digital video feeds have some sort of compression; that's what the MPEG standard does for us. Neither he nor I are experts on codecs so beyond looking up a couple of quick articles I couldn't add anything to that statement. Compression artifacts tend to show up in motion blur/pixelation and a general "softening" of the picture as the highest-bandwidth components of the signal are reduced/lost. High-bandwidth elements are generated by motion and sharp edges. Sharp-edged motion thus receives the highest compression but edge-blurring on objects in motion is not as objectionable as a loss of sharpness on a static or slowly-moving picture element.
1. Data dropouts do happen with digital signals and can cause artifacts, primarily pixelation and pixel dropouts when the EC (error correction) can't keep up. The degradation tends to be more noticeable than "minor" analog artifacts resulting in a loss of sharpening. In a digital feed, this may be a sharp-edged group of pixels that look out of place. Digital signals can generally operate with much lower SNR than analog before noticeable degradation occurs but, and partly because of that, it is true that digital tends to swiftly go from "good" to "gone" where analog goes through a host of artifacts before total signal loss.
2. Bandwidth allocation is dynamic, at least quasi-dynamic, and they can choose what compression level to use for any given show (and even within a time slot, e.g. the game and commercials may have different compression levels). So, they can choose to provide a highly-anticipated (read: well-funded by commercial sponsors) show at higher quality than another with lower audience (funding) to save (or, re-allocate) bandwidth.
3. There is some variation in quality from the content providers, by their choice and outside Comcast's control. Comcast can choose to further compress (or even upconvert) once they receive it, of course.
4. It used to be true in the analog days that quite often the best performance was from a local (big-city) newscast as the signal was local (less attenuated) and well-monitored by on-site engineers. The news was that local stations "flagship" program, after all. There's still some truth in that, though a lot of stations are still in transition with a mix of new and older equipment due to the extremely high cost of HD cameras, boards, etc.
5. Virtually all digital video feeds have some sort of compression; that's what the MPEG standard does for us. Neither he nor I are experts on codecs so beyond looking up a couple of quick articles I couldn't add anything to that statement. Compression artifacts tend to show up in motion blur/pixelation and a general "softening" of the picture as the highest-bandwidth components of the signal are reduced/lost. High-bandwidth elements are generated by motion and sharp edges. Sharp-edged motion thus receives the highest compression but edge-blurring on objects in motion is not as objectionable as a loss of sharpness on a static or slowly-moving picture element.
Thank Don. That is a nicer explanation than the brief version I provided earlier 
. Cable companies did not used to use stat-muxing and variable coding but once they started to stuff a lot of channels into your home, the game was on.
On #5, I have been behind writing my primer on video compression. I wrote this intro article for Widescreen Review Magazine but never the second part that explained the basics of how video compression works: http://www.madronadigital.com/Library/Video Compression.html
For now, here is a very, very brief primer.
Video compression works by taking the initial frame of video and compressing it like you would with JPEG compression in your camera or computer. The image is divided into blocks and each compressed separately by reducing the amount of bits allocated to different "frequencies" (sharpness level) of the frame. Compress the block too much and it will then look like a square and hence the "blocking" artifacts you see when the camera pans in a sports game. MPEG-2 has fixed size blocks. VC-1 and MPEG-4 AVC have variable sizes.
Once you have the initial frame compressed, then the system attempts to compress the subsequent frames by computing the difference between the two. This done by tracking what happens to each of the above blocks. If I just move my head, the signal is sent to the receiver to simply shift those blocks. The frame is NOT transmitted and we save a ton of data. This is called "motion estimation" in that the system attempts to predict which way the blocks are moving.
Of course, real life is not that simple as the world is not made of blocks that nicely move in one direction or the other. If the new frame looks too different, then we switch to the initial mode above and send the entire frame (compressed). This takes a ton of data relative to just sending the movement of blocks. This will likely blow our bandwidth budget. The solution is to compress the full frame even more, causing its blocks to show up more. Hence the reason I used the above camera panning example.
MPEG compression as you read in my article, is the oldest scheme for compressing video. We are stuck with it as a legacy system which ironically costs 10X to license than the newer, better compression schemes! To give you an example, VC-1 (which my group at Microsoft developed ) has both fade to black and flash detection. If someone takes a picture of the crowd in a fashion show and the lighting becomes much brighter for a fraction of a second, MPEG-2 thinks the entire frame has changed and retransmits it all even though all that is changed is the brightness of each pixel. Result is that you see a ton of compression artifacts during that time whereas with VC-1, the image holds is quality. Ditto for when the video goes black. With MPEG-2 you see a lot of blocking artifacts whereas VC-1 is smooth and nice.
To catch up, MPEG-2 companies have put in all kinds of tricks such as softening the video when the codec gets in trouble. Less detail means the video takes less bits to compress for equiv. quality. To most people, a softer picture is better than a sharper one with artifacts. Next time you are watching sports, look at what happens when the camera chases a (American) football player or hockey. Often, you see the image get soft but as soon as the motion calms down, the picture becomes sharp again (note: this also occurs due to display response being too slow such as in older LCDs).
Going back to your point #5, the feeds that are sent to companies like comcast is lightly compressed. Different schemes are used depending on how much money one wants to pay for bandwidth. In general, it is true that some quality is lost in that "uplink" and we have a double-compression situation where the video is compressed yet again when it is delivered to consumers.
I better stop here and see if I have lost all of you or just some.
. Cable companies did not used to use stat-muxing and variable coding but once they started to stuff a lot of channels into your home, the game was on.On #5, I have been behind writing my primer on video compression. I wrote this intro article for Widescreen Review Magazine but never the second part that explained the basics of how video compression works: http://www.madronadigital.com/Library/Video Compression.html
For now, here is a very, very brief primer.
Video compression works by taking the initial frame of video and compressing it like you would with JPEG compression in your camera or computer. The image is divided into blocks and each compressed separately by reducing the amount of bits allocated to different "frequencies" (sharpness level) of the frame. Compress the block too much and it will then look like a square and hence the "blocking" artifacts you see when the camera pans in a sports game. MPEG-2 has fixed size blocks. VC-1 and MPEG-4 AVC have variable sizes.
Once you have the initial frame compressed, then the system attempts to compress the subsequent frames by computing the difference between the two. This done by tracking what happens to each of the above blocks. If I just move my head, the signal is sent to the receiver to simply shift those blocks. The frame is NOT transmitted and we save a ton of data. This is called "motion estimation" in that the system attempts to predict which way the blocks are moving.
Of course, real life is not that simple as the world is not made of blocks that nicely move in one direction or the other. If the new frame looks too different, then we switch to the initial mode above and send the entire frame (compressed). This takes a ton of data relative to just sending the movement of blocks. This will likely blow our bandwidth budget. The solution is to compress the full frame even more, causing its blocks to show up more. Hence the reason I used the above camera panning example.
MPEG compression as you read in my article, is the oldest scheme for compressing video. We are stuck with it as a legacy system which ironically costs 10X to license than the newer, better compression schemes! To give you an example, VC-1 (which my group at Microsoft developed
) has both fade to black and flash detection. If someone takes a picture of the crowd in a fashion show and the lighting becomes much brighter for a fraction of a second, MPEG-2 thinks the entire frame has changed and retransmits it all even though all that is changed is the brightness of each pixel. Result is that you see a ton of compression artifacts during that time whereas with VC-1, the image holds is quality. Ditto for when the video goes black. With MPEG-2 you see a lot of blocking artifacts whereas VC-1 is smooth and nice.To catch up, MPEG-2 companies have put in all kinds of tricks such as softening the video when the codec gets in trouble. Less detail means the video takes less bits to compress for equiv. quality. To most people, a softer picture is better than a sharper one with artifacts. Next time you are watching sports, look at what happens when the camera chases a (American) football player or hockey. Often, you see the image get soft but as soon as the motion calms down, the picture becomes sharp again (note: this also occurs due to display response being too slow such as in older LCDs).
Going back to your point #5, the feeds that are sent to companies like comcast is lightly compressed. Different schemes are used depending on how much money one wants to pay for bandwidth. In general, it is true that some quality is lost in that "uplink" and we have a double-compression situation where the video is compressed yet again when it is delivered to consumers.
I better stop here and see if I have lost all of you or just some
.
Had to check the map now and then but a nice read, Amir! Yeah, MPEG has a lock on things, and once standards are established it's hard to change even though with today's digital systems it would actually be a fairly easy thing to do (easy for us to say, I guess). Why OTA (over the air) BW restrictions are still with us I don't quite get, especially with the ability to tailor compression and thus BW as required...
Similar threads
- Locked
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |