DBSTalk Forum banner
Status
Not open for further replies.
1 - 3 of 3 Posts

·
DirecTV 10yr+ Customer
Joined
·
1,011 Posts
Discussion Starter · #1 ·
Well,

I'm here is Atlanta and call on Scientific Atlanta, Motorola BCS and Echostar settop development group engineering. I've been doing this for 20 years and lately have seen alot of lab work on MPEG4 streams. What has been interesting I've seen alot of lab testing where the cable guys take their fixed Coax bandwidth and chop it up/degrade (called transcoding/transrating) in order to get more bandwidth (ie. channels) out of the coax. So, basically it is a "how low of a bit rate stream and quality can we drop the HD images that is still acceptable to the consumer". Like most, I was under the impression MPEG-4 main profile (H.264) was inherantly worse than MPEG-2. Yesterday I was educated technically by a Encoder engineer from this the company EGT that I was in fact wrong (http://www.egtinc.com/). They make MPEG encoders that bolt to the transmission headends. EGT in general does only MPEG-4. However, what I thought was interesting is that while MPEG-4 runs at a lower bit rate than MPEG-2 it actually is a higher quality stream ! Yup, higher quality. The key to quality as explained to me is in the encoding of the baseline native broadcast video into MPEG and that processes effectiveness/efficiency in doing motion compensation (basically what has changed/moved in a frame). To bottom line it....MPEG-2 has a 1/2 pixel resolution and MPEG-4 a 1/4 pixel resolution. So, they showed me in their lab a MPEG-2 stream and MPEG-4 stream running the identical encoder and the MPEG-4 was substantially noticably better ! I couldn't believe it !!!! They agreed that most people/consumers have the same viewpoint that MPEG-4 is worse because the cable folks use its lower bit rate to even play more games on increasing channel bandwidth. So, the long-n-short is that the MPEG-4 we see is not the MPEG-4 that is possible. It comes down to the battle of quality vs. quantity of channels.....no surprise.

Now for what it is worth....I have no idea whose encoders D* uses nor their transcoding/transrating efforts. Clearly, D* markets their quality over the cable folks and I hope that continues. The reason I post this is there is alot of threads on the quality of the local HD MPEG-4 streams. I guess the word is to be patient as they continue to tweek the firmware and upgrade the equipment. I also recall a post that D* and the affiliates are going thru and extra conversion step (forget why) and this would be eliminated in the future.

Anyway, I found this tutorial interesting...thought others might.

Take a look side-side: http://www.balooga.com/display.php3?dir=mpeg4&fileName=mpeg4-2_mpeg4-10_1200.jpg
I know.....folks will say.....well the MPEG-2 is at a low 1.2Mbps bit rate.....that's not apples-apples (fair) since most MPEG-2 is at 4Mbps+. My point is...oh yes it is...because as stated above the folks above are transrating their streams so it is apples-apples in real-world consumer environment.
 

·
Hall Of Fame
Joined
·
1,222 Posts
That's all good and well, but if you take something that was already encoded in MPEG2, decode it, and re-encode it in MPEG4, it will never be better and will almost certainly be worse than the MPEG2.

What you say would only be applicable if DirecTV could get the uncompressed feed from the local station before they encode it to MPEG2 for ATSC broadcast and encode it themselves to MPEG4 for LiL broadcast. I'm not even sure such a thing is possible (i.e. wouldn't the things they broadcast have already been compressed by the network before being sent to affiliates?)

Put another way: I don't think anyone was suggesting that MPEG4 was inherently lower quality than MPEG2 (that would be a meaningless statement). They were saying it was the additional compression step that was the cause. Try editing a JPEG several times to see the problem: the file may actually get larger each time, but quality will get worse.
 

·
DirecTV 10yr+ Customer
Joined
·
1,011 Posts
Discussion Starter · #3 ·
walters said:
That's all good and well, but if you take something that was already encoded in MPEG2, decode it, and re-encode it in MPEG4, it will never be better and will almost certainly be worse than the MPEG2.

What you say would only be applicable if DirecTV could get the uncompressed feed from the local station before they encode it to MPEG2 for ATSC broadcast and encode it themselves to MPEG4 for LiL broadcast. I'm not even sure such a thing is possible (i.e. wouldn't the things they broadcast have already been compressed by the network before being sent to affiliates?)

Put another way: I don't think anyone was suggesting that MPEG4 was inherently lower quality than MPEG2 (that would be a meaningless statement). They were saying it was the additional compression step that was the cause. Try editing a JPEG several times to see the problem: the file may actually get larger each time, but quality will get worse.
Actually, no....the broadcasts are analog..........composite (called CVBS) out of the headend....they then go into encoders....then downstreamed. However, and again I forget where I saw this, D* may be making that M2 to M4 conversion now as an inrterim approach to get the locals MPEG-4 stream out quickly. Again, I'm not sure of their transmission architecture at all so this is just speculation ! As an example Rogers Cable headends use this product....composite (CVBS) in.....MPEG-4 or IP video out: http://www.egtinc.com/Page3346.aspx

I posted this thread because I have seen alot of folks that stated MPEG-4 was natively worse quality. To be honest, I thought this myself because of the low native bit rate required. Anyway, obviously alot of variables on what "could" happen to the image but hopefully in the long run if the headends get the latest MPEG-4 encoders (Gen 2) with the increased DSP processing power we may someday get those 150+ HD channels all broadcast in 1080i....all CVBS to MPEG-4 directly encoded. Here's to the statement...."I have a Dream". :joy:
 
1 - 3 of 3 Posts
Status
Not open for further replies.
Top