or Connect
AVS › AVS Forum › HDTV › HDTV Programming › Comcast HD Quality Reduction: Details, Screenshots
New Posts  All Forums:Forum Nav:

Comcast HD Quality Reduction: Details, Screenshots - Page 3  

post #61 of 2079
Quote:
Originally Posted by mrvideo View Post

I totally disagree. I get excellent image grabs.

You do NOT want to deinterlace 1080i. Deinterlacing can cause other problems. I can understand wanting to de-interlace when the source is video, since each field will be different and it is more evident with motion.

The images that I capture are from film sources or 23.976p video. For example:

http://vidiot.com/images/CBS-Sprite-080317-03.png
http://vidiot.com/OneTreeHill/images...ill-080226.png
http://vidiot.com/LifeIsWild/images/...ild-071202.png

With regard to the examples given, even if VRD's deinterlacing isn't as good, the point is that the images from the two source are obviously different. As with any experiment, you have to have a control. In this case, the control is the capture software. It is VRD, which will apply the exact same deinterlace function to both video sources. So, whatever it "adds" to the image will be the same for both.

Then how do you explain the frame grabs he posted?

The point is even if the difference between the two captures are obvious I want the images to look as close as possible from the original source. For accuaracy's sake.
post #62 of 2079
Quote:
Originally Posted by markofmayhem View Post

From what I've seen with Quantum HITS, the "receiving head-end" only has to modulate to 256QAM with no encoding/decoding needed, just modulation (from Q to 256). [see above posting for the rest of the posting]

The modulation schemes that I know of for satellites are QPSK, 8PSK and 16QAM. But the 16QAM is not the same as the 256QAM used on cable. In the case of these HITS feeds, I believe the only modulation available for DCII is QPSK. I've not seen 8PSK DCII satellite feeds.

That said, by sending the MPEG-2 stream at the bitrate needed for squeezing 3 HD channels into a single QAM channel, no recoding is required of the MPEG-2 video. Just demod it from the receiver and feed the MPEG-2 directly into the QAM modulator.

It is the same kind of scheme that FOX does with their splicer system at their affiliates. They encode the MPEG-2 in LA, send it out over a bird, when then goes through the splicer box, to get joined up with the audio and all of the other ATSC PSIP data bits and sent to the transmitter. No re-encoding by the local affiliate.

There is FEC with the sat feed. It is a must. In this case it is 3/4. The FEC that is in the DCII sat feed is not the FEC used for the QAM channel

The reason that ABC, CBS, NBC and The CW have bitrates up into the 35 Mbps+ range is to satisfy a kind of rule regarding reencoding of MPEG-2. As a general rule, the incoming bitrate should be at least twice as high as the final encoded rate. So, for ATSC, the final rate maxes out at about 18 Mbps. So the sat feeds are 36 Mbps, or better. It seems that there can be all kinds of funny MPEG-2 encoding artifacts when the source is MPEG-2. The closer the source MPEG-2 bitrate is to the final bitrate will result in the errors being worse.

In the case of these HITS sat feeds, no reencoding of the MPEG-2 is taking place, so those encoding problems don't appear.

With the total payload of the QAM channel being limited to 38.8 Mbps, so no way in Hell should they be shoving three 1080i HD video streams into that limited space. As was seen, the video can look like crap.
post #63 of 2079
Quote:
Originally Posted by mrvideo View Post

The mux is QPSK, so the most you can get out of it is about 45 Mbps. The CBS mux is 32762 SR, also at 3/4 FEC (IIRC). Since this mux is at 29270, I suspect the total payload is about 42 Mbps (I don't have the formula handy for figuring out the exact payload datarate).

29270000 (SR) * 2 (QPSK) * 3/4 (FEC) * 188/204 (RS) = 40.46 Mbps

Ron
post #64 of 2079
Wow, good to see that this issue is finally seeing the light of day. I 1st noticed this issue back in January, had 3 techs come out, one of which wanted to gut the plugs in my house. I knew all along this was a Comcast issue and not some wiring or node load issue.

http://www.dslreports.com/forum/r197...and-Pixelation

http://www.youtube.com/watch?v=nT30UW80N0I
post #65 of 2079
just special-

I'm on a small system with great HD and we're being assimilated by the borg. Comcast takes over at the end of the month.

I hope they dont fiddle with anything until they dump analog.
post #66 of 2079
Quote:
Originally Posted by michaelk View Post

just special-

I'm on a small system with great HD and we're being assimilated by the borg. Comcast takes over at the end of the month.

I hope they dont fiddle with anything until they dump analog.

If you have HDNet, say bye-bye to it.
post #67 of 2079
Quote:
Originally Posted by sansri88 View Post

If you have HDNet, say bye-bye to it.

Comcast left HDNet on when they took over here in Houston
post #68 of 2079
Quote:
Originally Posted by dr1394 View Post

29270000 (SR) * 2 (QPSK) * 3/4 (FEC) * 188/204 (RS) = 40.46 Mbps

Thanks. I'm think for 8PSK it is 4, not 2 (for some reason 3 sticks in my mind). Instead of 188/204 for the Reed/Solomon, what is it for Trellis?
post #69 of 2079
Quote:
Originally Posted by mrvideo View Post

Thanks. I'm think for 8PSK it is 4, not 2 (for some reason 3 sticks in my mind). Instead of 188/204 for the Reed/Solomon, what is it for Trellis?

QPSK = 2, 8PSK = 3 and 16QAM = 4 for the bits/symbol. The trellis or convolutional code is the 3/4 number. The convolutional code and the RS code are always used together.

If the transponder is really 8PSK, then the bitrate is much higher.

29270000 (SR) * 3 (8PSK) * 3/4 (FEC) * 188/204 (RS) = 60.69 Mbps

40.46 Mbps seems to match the TSReader data from vj9999 much better than 60.69 Mbps.

Ron
post #70 of 2079
Thread Starter 
By request, I've replaced all the VideoRedo frame captures with those taken using Media Player Classic with Dscaler5-IVTC. I captured the full resolution frames and then resized them to half resolution with XnView (Lanczos).

The new half-resolution captures on the first page are clearly less detailed than the previous VideoRedo captures, but that may have something to do with differences between the algorithms VideoRedo and XnView use to resize.

I used ZIP instead of RAR for the 1920x1080 captures download.
post #71 of 2079
Quote:
Originally Posted by dr1394 View Post

QPSK = 2, 8PSK = 3 and 16QAM = 4 for the bits/symbol. The trellis or convolutional code is the 3/4 number. The convolutional code and the RS code are always used together.

Man, I gotta go back and re-read this stuff. The expensive Trellis is in the FEC (Forward Error Correction) value? Too bad Trellis wants both arms and both legs to use their scheme

Quote:


If the transponder is really 8PSK, then the bitrate is much higher.

No, it isn't 8PSK. Just asking for my notes. I'm keeping the formula handy.
post #72 of 2079
Quote:
Originally Posted by Xylon View Post

Then how do you explain the frame grabs he posted?

I looked at the same samples of both deinterlaces that you did and the VRD version is definately worse. The descaler5 also has its shortcomings, but it isn't as bad as the VRD version.

Quote:


The point is even if the difference between the two captures are obvious I want the images to look as close as possible from the original source. For accuaracy's sake.

Then they shouldn't be deinterlaced at all. That would make it tougher to look at, if there was rapid motion, as the two fields would be vastly different.

That said, for me, the VRD comparisons would be fine, since the point was to show the vast difference between to two systems. Comcast just plain sucked.
post #73 of 2079
Quote:
Originally Posted by mrvideo View Post

You chopped off the bottom of the TSReader output that gives the total mux datarate.

But, it is interesting to note that my guess regarding the bitrates of the three video streams was higher than the actual bitrates.

Here you go. This was captured with Fusion HDTV and TSReader source I modified to be able to connect to Fusion's stream (I use an external app to change the channel and TSReader hooks on to the stream).
LL
post #74 of 2079
Quote:
Originally Posted by risk235 View Post

Comcast left HDNet on when they took over here in Houston

Others weren't so lucky.
post #75 of 2079
Quote:
Originally Posted by mrvideo View Post

In the case of these HITS sat feeds, no reencoding of the MPEG-2 is taking place, so those encoding problems don't appear.

With the total payload of the QAM channel being limited to 38.8 Mbps, so no way in Hell should they be shoving three 1080i HD video streams into that limited space. As was seen, the video can look like crap.

So it is your opinion that the poor quality is solely based on Imagine's ICE Broadcast solution; the actual squeezing of 3 HD feeds into 1 QAM? I concur, I was being optimistic in thinking it could have been the second sat bounce.

The only HITS modulation I've seen is QPSK. It sounds as if there is plenty of headroom on the QPSK transponder to introduce enough FEC to ensure the sat transmission is not causing interference, so it must be the "package". That is not good, considering Comcast has sold themselves on the idea of "national" releases of national channels via HITS where Imagine's ICE Broadcast solution will be squeezing 3HD and 15SD into 38.8 pipes (actually, more like 36, leaving headroom)... yuck. I had high hopes for this solution, but it is quickly becoming apparent that MPEG4 is the only quality-viable compression option out there today. "Bit Rate Shaping" and other gimmicks make logical sense at lower scales, but Comcast is exceeding Imagine Communication's capabilities. Perhaps 2 HD along with 2 SD per QAM would have been a better starting point than 3 HD.

Didn't Cable Labs announce that 15 Mbps was the "safe harbor" for HD content?
post #76 of 2079
You keep saying vyyo, vyyo, vyyo... I thought it was Imagine communications.

I've just checked out vyyo's website and they are heavily pushing their 3 GHz spectrum overlay. If Comcast invested in vyyo, wouldn't they be doing 3 GHz spectrum overlay instead?

The Vyyo solution adds an 1300-2750 MHz "UltraBand".

Here's Imagine's website explaining the product:
http://www.imaginecommunications.com...ions/video.php
post #77 of 2079
I am SO sorry. It IS Imagine... lack of sleep is catching up! 15-18 hour work days are taking their toll.
post #78 of 2079
Quote:
Originally Posted by Foojay View Post

Wow, good to see that this issue is finally seeing the light of day. I 1st noticed this issue back in January, had 3 techs come out, one of which wanted to gut the plugs in my house. I knew all along this was a Comcast issue and not some wiring or node load issue.

http://www.dslreports.com/forum/r197...and-Pixelation

http://www.youtube.com/watch?v=nT30UW80N0I

I think we need to be careful here... The discussion in this thread is about macroblocking, or artifacts in the picture, caused by insufficient bit rates. The "pixelation" that is demonstrated in the youtube video is a completely different issue. That should instead probably be referred to as "break ups" as that's what's happening. The picture is breaking up. Break ups are caused by too little signal reaching the receiver for it to decode the whole stream properly. There is a threshold for digital reception. If you are above it, you get a clean picture, if below it, you get no picture. If your signal level is right at or slightly above the threshold, small variances in the signal strength would cause the break ups that you are seeing.

In your case it's certainly likely that when Comcast added or re-arranged channels, channels that used to work fine for you no longer work as they used to (now have break ups). If Comcast moved them to a different frequency (could still be on the same "virtual channel" that you actually see), then it's possible that the signal strength on those new frequencies is worse. Different frequencies, higher or lower could be more susceptible to interference, or existing wiring may not be sufficient for them.

If what you are experiencing is "break ups" then it most certainly could be a problem with your wiring and the Comcast tech was correct in attempting to address the issue in that way.
post #79 of 2079
Quote:
Originally Posted by markofmayhem View Post

I am SO sorry. It IS Imagine... lack of sleep is catching up! 15-18 hour work days are taking their toll.

Heh thats understandable. For awhile there I thought maybe Vyyo bought out Imagine Communications so quickly! I did some research and couldn't find anything on it. I get CED news in my e-mail every day and never saw it there as well. Anyway get some rest this weekend!
post #80 of 2079
I can tell you here in Knoxville, TN, Comcast PQ on some channels has deteriorated. I looked and there are several groups of channels where 3 channels are crammed into one QAM. All these channels are the ones with worse PQ and occasional breakup in the signal.
post #81 of 2079
Not gonna happen this weekend. I got Vyyo on the brain, what a screw-up!

The "break-up" could be a signal issue that suddenly appeared. I'm more on the theoretical and business case side of things, so those with engineering experience can tear me apart. Most HD feeds hover around 17-18.5 Mbps, with the remaining "19.9" pipe (2HD per 256QAM) left to headroom mostly filled with FEC, correct? So, on average, 1.5-3.0 Mbps of FEC exists per "channel". On the tripple-loaded QAMS, each "channel" is reduced to sub-12 Mbps with less than 1 Mbps of head room available for FEC. (3-12 Mbps per channel leaves 2.8 in headroom divided by 3). This leaves VERY little room for a signal to survive environmental molestation. Where signals on certain frequencies were receiving less than picutre altering errors, the lowering of correction data would then exploit these frequencies causing break-ups and drop-outs. I can see that happening in rare situations, but the more I read and talk to Comcast customers, the more "wide-spread" this appears! Is there a single Comcast customer out there who has watched an action-packed movie on Sci-Fi HD or USA HD that hasn't experienced a full-screen pixelated mess during fast-full screen changes?

The other explanation could be with the ICE server. It is only able to receive compressed MPEG2 feeds, unlike NBC/CBS affiliates that decompress their feeds and then recompress them into superb final products in the sub-16 Mbps bitrates; the Imagine solution FURTHER compresses without decompression into a sub-12 Mbps bitrate. Pixelation and drop-outs could easily be the result of buffer overflow.
post #82 of 2079
they really should just give everyone a box for free and switch to full digital. there's really no reason not to, that i can think of.
post #83 of 2079
Quote:
Originally Posted by briansemerick View Post

they really should just give everyone a box for free and switch to full digital. there's really no reason not to, that i can think of.

I can think of two.
1) Boxes are expensive.
2) People hate them.
post #84 of 2079
Quote:
Originally Posted by aindik View Post

I can think of two.
1) Boxes are expensive.
2) People hate them.

And once you've breached that barrier, sat becomes an option for customers as they all use boxes as well.
post #85 of 2079
Quote:
Originally Posted by markofmayhem View Post

Not gonna happen this weekend. I got Vyyo on the brain, what a screw-up!

The "break-up" could be a signal issue that suddenly appeared. I'm more on the theoretical and business case side of things, so those with engineering experience can tear me apart. Most HD feeds hover around 17-18.5 Mbps, with the remaining "19.9" pipe (2HD per 256QAM) left to headroom mostly filled with FEC, correct? So, on average, 1.5-3.0 Mbps of FEC exists per "channel". On the tripple-loaded QAMS, each "channel" is reduced to sub-12 Mbps with less than 1 Mbps of head room available for FEC. (3-12 Mbps per channel leaves 2.8 in headroom divided by 3). This leaves VERY little room for a signal to survive environmental molestation. Where signals on certain frequencies were receiving less than picutre altering errors, the lowering of correction data would then exploit these frequencies causing break-ups and drop-outs. I can see that happening in rare situations, but the more I read and talk to Comcast customers, the more "wide-spread" this appears! Is there a single Comcast customer out there who has watched an action-packed movie on Sci-Fi HD or USA HD that hasn't experienced a full-screen pixelated mess during fast-full screen changes?

The other explanation could be with the ICE server. It is only able to receive compressed MPEG2 feeds, unlike NBC/CBS affiliates that decompress their feeds and then recompress them into superb final products in the sub-16 Mbps bitrates; the Imagine solution FURTHER compresses without decompression into a sub-12 Mbps bitrate. Pixelation and drop-outs could easily be the result of buffer overflow.

19.28 Mbit/s is the payload, not the actual broadcast rate including FEC, R-S, and whatnot. The difference between 19.28 and the video bitrate is taken by audio, nullpackets, PAT, PMT, EIT, ETT, and various other ancillary streams. In addition, the FEC is not broken down by the contents of the stream; it operates at a totally different layer and would have the same effect recovering a stream full of nullpackets as it would recovering 3 overcompressed HD channels or 15 overcompressed SD.
post #86 of 2079
Quote:
Originally Posted by aindik View Post

I can think of two.
1) Boxes are expensive.
2) People hate them.

if those boxes are expensive then they need to find another place to buy from because they're sh*t. the 2" cablecard i just got has better picture on regular channels than my 2' X 1' box ever did. plus they don't seem to give a damn about the boxes, most of the time they just tell people to keep them because they're so old and outdated. i had one of the black ones and it sucked.

and people may hate them (as do i) but it's a necessary evil right now to ensure higher PQ. that or buy a new TV w/cablecard access. don't degrade my signal because people don't want a box.
post #87 of 2079
Quote:
Originally Posted by keenan View Post

And once you've breached that barrier, sat becomes an option for customers as they all use boxes as well.

no satellite in an apartment, plus can you have more than one TV hooked up to it? i know you used to not be able to, but that's probably changed now.
post #88 of 2079
Quote:
Originally Posted by briansemerick View Post

if those boxes are expensive then they need to find another place to buy from because they're sh*t. the 2" cablecard i just got has better picture on regular channels than my 2' X 1' box ever did. plus they don't seem to give a damn about the boxes, most of the time they just tell people to keep them because they're so old and outdated. i had one of the black ones and it sucked.

and people may hate them (as do i) but it's a necessary evil right now to ensure higher PQ. that or buy a new TV w/cablecard access. don't degrade my signal because people don't want a box.

Comcast currently charges $8.90 a month for every digital box. (The first $8.90 is included in the price of a digital package if you have one). Unless they drastically reduce the price, customers with five TVs and 1 box (these people, I think, are pretty common) will see their cable bills go up by $35.60 if Comcast switches to all digital and requires boxes in order to watch those other four TVs. A not-insiginificant number of them will switch to Dish, DirecTV or FIOS if that happens.

Not needing a box for every single TV is the only thing that keeps cable anywhere near price-competitive with satellite.
post #89 of 2079
Quote:
Originally Posted by briansemerick View Post

the 2" cablecard i just got has better picture on regular channels than my 2' X 1' box ever did.

CableCards don't tune channels, they are removable security cards used to authorize channels, to be tuned by the TV's internal tuner.

CableCards have also been used in third-party DVRs, and now in all cableco. tuners.
post #90 of 2079
But with more HD channels being added to the digital lineup and more people going that direction, having no box will no longer be the case for many customers in a few years. That will force them to make some sort of change or lose a number of customers.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: HDTV Programming
This thread is locked  
AVS › AVS Forum › HDTV › HDTV Programming › Comcast HD Quality Reduction: Details, Screenshots