I'm telling ya, SD looks worse on purpose so we buy more HDTV sets

ehren

SatelliteGuys Pro
Original poster
Apr 8, 2006
1,440
0
Wisconsin
If I may quote the late George Carlin doing the voice of the hippie VW bus in the movie Cars...

"its a conspiracy man'!!!

The Anaheim/Calgary game on NHL Center Ice today from Flames PPV was the best SD PQ i've seen lately. Then you try and watch CSN-Bay Area SD and CSN-New England SD and it's like taking your contacts out!

Trying to watch the Blackhawks/Canucks game on WGN right now in SD is terrible as well. I don't care if I bring up topics that nobody will probably care about anyways but this SD quality makes HD-Lite look so good doesn't it!
 
If I may quote the late George Carlin doing the voice of the hippie VW bus in the movie Cars...

"its a conspiracy man'!!!

The Anaheim/Calgary game on NHL Center Ice today from Flames PPV was the best SD PQ i've seen lately. Then you try and watch CSN-Bay Area SD and CSN-New England SD and it's like taking your contacts out!

Trying to watch the Blackhawks/Canucks game on WGN right now in SD is terrible as well. I don't care if I bring up topics that nobody will probably care about anyways but this SD quality makes HD-Lite look so good doesn't it!

SD has, somewhat unfairly, been given a bad name. SD can actually be very watchable--if it's done right. The widescreen SD episodes of COPS on Fox is a good example. Most of the time, however, they compress the crap out of SD. On a modestly-sized SD television set it doesn't really matter too much, but when you blow it up onto a big, hi-rez HDTV screen all the flaws come out.

IMO, resolution is generally over-emphasized as an indicator of picture quality. I just had some people over to watch "I Am Legend" on my 108-inch (9 foot!) projector screen. My projector is a Sony Bravia VPL-AW15, which is a 720P model, and the movie was recorded from one of the HBO HD channels. The picture was absolutely razor sharp with incredible quality. One of my friends commented "Wow...I can't tell that it's only 720P." It wasn't the resolution that made the picture beautiful...it was bitrate. Over-compression and macro-blocking look just as bad in 1080 as they do in 720.
 
Warning: Giant post ahead. Mods, please split if necessary. :)

ehren: I absolutely agree that both DIRECTV and Dish do unnecessary damage to the picture quality on SD channels. Identical MPEG-2 content with comparable bitrates from other sources looks like DVD quality.

I'm sure the reason you gave has become part of their motivation behind intentional quality reduction. It can't be their only motivation however because it can be proven that some channels consistently have a worse picture than other channels, regardless of the current bitrate, and many of the worst-quality channels have no HD-equivalent to upgrade to. Side-by-side comparisons show that on Dish, the Encore channels are much worse than the Starz channels even though those originate from the same source and generally have bitrates (from Dish) within 10-15% of one another. This is backwards in terms of your theory since HD-equivalents of the already decent-quality Starz channels are available, but only the main Encore channel has an HD version.

It is believed by some that all of this quality-reduction is necessary due to bandwidth constraints. Numerous methods of pre-filtering exist that are quite effective at improving compressibility of video, so there is some truth behind this idea. In the early days of the sat providers running out of bandwidth, there probably weren't much higher-quality MPEG-2 encoders available than when the services first launched, so they probably invested in lots of real-time pre-filtering equipment instead. The pre-filtering equipment, like the MPEG-2 encoders, was probably not of the highest quality either though. It blurs the video (on DIRECTV and Dish Network) and distorts the video (at least on Dish Network), and the amount of filtering now being used is completely overkill. After a few years of use, it must have become clear that the filtering equipment was not meeting their needs properly, which should have resulted in better equipment for filtering and/or encoding. Upgrades to the MPEG-2 SD-related equipment should have been done periodically as significantly more advanced equipment became available, not entirely unlike the switch from MPEG-2 to H.264 for HD channels since MPEG-2 HD channels have very high bandwidth requirements in order to retain quality. If such upgrades were done, they most definitely weren't done properly.

Good pre-filtering methods are supposed to make minimal visible changes that maximize compressibility without lots of blocking and other artifacts. Bad or grossly overused methods reduce DVD-quality source material to sub-VHS quality without making any appreciable improvement to compressibility. The latter seems to be the case for the sat providers in the US because SD picture quality steadily declines over time as more national, local, international, and HD channels are added. Spot beams and new birds help to add lots more channels, but that added capacity is promptly gobbled up by new channels, which leaves the previously existing channels in the same cramped quarters as before. At least on Dish, the pre-filtering is now being done in such a poor manner that I have often found pixelation to be worse on channels that have more visible pre-filtering than on those with much less visible pre-filtering, all bitrates being approximately equal, and all content compared being from the same sources.

All of this leads me to believe that some or all of the following things are happening with SD channels on Dish and DIRECTV:
- Some or all of the original, low-quality (by today's standards) MPEG-2 encoders are still in-use, requiring the use of at least some form of highly aggressive/destructive pre-filtering method.
- Some or all of the original, low-quality pre-filtering equipment/software is still in-use, causing the picture quality to be terrible even if the MPEG-2 encoders have been upgraded. Older filters are probably less configurable and efficient than newer ones are, so as long as they are in-use, the amount of damage they do to picture quality may be difficult or impossible to control.
- The older MPEG-2 encoders and/or the pre-filtering devices may have been replaced, but are also of poor quality (cheap) or have been configured improperly to maximize picture quality given known bandwidth constraints.
- Pre-filter configurations are static and are supposed to try to minimize pixelation in the event of significant, temporary bandwidth drops (due to increased bandwidth requirements from another channel in the same mux). Such dramatic spikes seldom occur, so the excess filtering causes video quality to suffer horribly the large majority of the time as a precaution against an unlikely scenario. Additionally, this unlikely scenario is supposed to be prevented already by the load balancing system that controls the bitrates for each channel in a mux at any given time. More bits are supposed to be allocated to the channel with the most complex picture data in the mux at any given moment, with the number of available bits scaled based on the number of channels in the mux, the priority of a given channel relative to all others in the mux, and a maximum allowed bitrate.
- They get some kind of discount from the programming providers because the content, once degraded, can't possibly be worth full price. This may sound far-fetched, but consider the quality-degrading copy-protection schemes that are supposed to become more common in the future, and then it doesn't seem so crazy. Such a discount would be considered substantial enough to degrade the video in order to get it, but not so substantial as to pass any savings onto the customers, and instead allow the prices for exactly the same packages of channels, with few or no channel additions, and with more commercials than ever before, to have risen by 70-80% in the past 10 years.
- They just don't have a clue what they are doing, they believe the picture quality is actually good (as they enjoy claiming, occasionally even in courtrooms) and ultimately don't care because they are so delusional that they don't really believe the population is slowly but steadily upgrading to high-definition TVs with digital inputs that allow the bad SD picture quality that had been masked by NTSC CRTs for so many years to finally show through.

I have other theories, but this seems like more than enough for the moment. Basically, if any of that stuff really is true and the US sat providers have no plans to do anything about it, they can look forward to being buried alive by the likes of FiOS and U-Verse. Once the FTTH TV services are more widespread, and DOCSIS 3.0 becomes dominant for cable to lessen constraints on their bandwidth, DIRECTV and Dish are going to have one hell of a time trying to actively compete if they refuse to improve the quality of their service. It has been reported (maybe by Scott?) in another thread that the H.264 SD on Dish is WORSE than the MPEG-2 SD, so that's a pretty clear indication that Dish's current priorities lie solely in investing in cheap, low-grade systems for handling video compression rather than planning for the future. When used well at the bitrates Dish is currently using for SD (~1000-1500kbps), H.264 makes it easier than ever to make SD video look great with far fewer bits than MPEG-2. With the new Eastern Arc's SD looking nasty and their general inability to satisfy their customers lately, it is no wonder that Dish is losing rather than gaining. They can use the economy as a scapegoat all they like, but the reality is that due to the economic shifts, people are MORE likely to engage in stay-at-home activities like watching TV, not less, because it is more affordable overall. They're simply in the wrong business to be playing a blame game in which anyone or anything is allowed be a contestant except for themselves.
 
Edgar_in_Indy: The East/West major network feeds are given a disproportionately high amount of bandwidth relative to most of the other channels on Dish, so those should at least look decent. Spotbeams vary of course, so I can't say for sure what the case is there.

As for resolution being overemphasized, you're very right. If you watch an HD channel on Dish (1440x1080i) or DIRECTV (1280x1080i), and it hasn't been filtered a lot, it can look very nice on an HD display. The same can often be said of an upscaled DVD (720x480i/p) depending on the transfer, bitrate, encoder, and of course any filtering that may have been used. Unlike with HD channels however, the SD channels on Dish (544x480i) and DIRECTV (480x480i) never look nearly as good as DVD-quality even though the SD-sat vs. SD-DVD encoding resolution difference is comparable to the HD-sat vs. Blu-Ray/HD-DVD resolution scenario. Only transfer quality is beyond the control of sat providers. The encodings they receive are of high quality and have higher-than-absolutely-necessary bitrates, full resolution (704x480i?), and are not filtered poorly. It is the sat providers that strip the bulk of the quality from the video they receive before relaying it to us, the customers, which is what makes it appropriate to direct the complaints at them.

I will try to capture some frames relating to all of this tomorrow. Quite enough fun commentary for this evening. Night! :)
 
Dishnetworks Mpeg 4 SD channels are 480x480 now as well for Eastern Arc customer. D* SD PQ used to be better then E*, But I can't say that anymore. But its still close.
 
Dish MPEG-2 SD vs. Dish MPEG-2 SD

Ok, here's a pair of comparison shots. This is a comparison of two MPEG-2 SD channels on Dish. Transfers are the same, and the source feeds used by Dish should be from the same satellite. The shots are PNG images made from BMP exports from MPEG-VCR of progressive frames in digital captures (transport streams). PNGs aren't ideal, but I didn't want to post big BMPs or JPG that could introduce substantial artifacts.

The first one is from Starz In Black (119W) with an average bitrate of 1.99Mbps. It shows a fair amount of detail in the face, but is also quite a soft image for the resolution used (544x480).



The second one is from Encore Action (110W) with an average bitrate of 1.71MBps. The Starz airing had just 16% more bits allocated to it on average, but the Encore airing consistently used a lot more pre-filtering regardless of variations in the bitrate meant to accommodate increases in details and action. In this image, the face details are very washed out. This Encore image is what you might expect to see if your vision isn't that great, and the Starz image is what you see when you put your glasses on.



Notice that the Encore channel was carrying analog closed-caption data (static at the top of the image) while the Starz channel wasn't. Also, all of the motion on the Encore channel has to battle its way around the stationary logo, which makes encoding more difficult. So, not only has the Starz channel been granted more bits and given less filtering, but it is guaranteed to have less noise on every single frame that is encoded. By all means, more bits and far more selective filtering should be applied to the Encore channel, while the Starz channel could probably afford to give up some of its bits and not suffer thanks to lower noise levels.

Oh, and don't forget that Encore Action is an ACTION channel. Ya know, the kind of channel that would benefit a great deal more than most other channels from having a higher quality picture (for special effects) and more bits allocated to it (for instantly handling explosions, fast motion, rapid back-to-back scene changes, and other elements consistently present in action films, rather than it taking a second or two to MAYBE have the encoder detect the action taking place and start to allocate more bits, but more than likely make this change far too late to prevent one scene after another from being quite sloppy).

This is just one example of how resources are being allocated improperly and filtering is not being configured well by Dish. In this case, the two affected channels are on separate satellites, so they can't be in the same mux to exchange bits directly with one another. However, Dish moves channels pretty regularly (at least more often than DIRECTV), so exchanging bits to more effectively balance the load across the entire platform isn't an unreasonable thing for them to do. As mentioned previously though, the bitrate isn't the only factor involved here.

The next comparison post should be foreign SD vs. Dish SD. :)
 
Dish is probably being careful to not let the MPEG4 Eastern Arc picture quality be any better than the MPEG2 Western Arc so new customers won't demand an Eastern Arc installation or worse, millions of existing customers demanding to be switched over. Hopefully, over time, they will up the picture quality since so many people have large screen sets now.
 
Last edited:
Dish is probably being careful to not let the MPEG4 Eastern Arc picture quality be any better than the MPEG2 Western Arc so customers won't demand an Eastern Arc installation. Hopefully, over time, they will up the picture quality since so many people have large screen sets now.

Sounds like a happy ending. While they have no excuse for not improving the quality of MPEG-4/H.264 SD channels in the future, they have already invested time and money in resources that intentionally reduce picture quality. I suppose it depends on what the cost of those resources is versus the benefits of preventing interest in EA during the early rollout period.

I've had DIRECTV since '99 and Dish on-and-off since 2000, and had been hoping to switch to Dish exclusively due to EA, but decided against it since there's not really an advantage currently. Trees block out 129W and cause interference to 119W as soon as the leaves come in, so if Dish's plan is to prevent Western Arc subs from switching to EA (which I know they hadn't been allowing anyway, regardless of what customers wanted), they're also succeeding in preventing DIRECTV subs from making the switch. D'oh!
 
Last edited:
Okay, here's a comparison that's a bit more fun: Canadian sat provider Bell TV vs. Dish Network.

The Bell sample I received has a resolution of 704x480 and an average bitrate of 3.05Mbps. The Dish sample has a resolution of 544x480 and an average bitrate of 2.38Mbps. The Dish sample has been allocated 22% fewer bits for video than the Bell sample. However, the reduced resolution of the Dish sample effectively reduces the bit allocation difference to about 8% (assuming the transfer resolution is actually 640x480 and that it was stretched to 704x480 for transmission, which seems to be a common resolution for full-quality digital SD channel transmission in both the US and Canada).

Once again, a close-up of a face will be used since faces are loaded with all kinds of details to examine and compare. This time I have a reference picture (sort of) from an "HD" airing on HBO so that the effects of pre-filtering and/or compression artifacts can be distinguished from the original content in the video. I put HD in quotes because the HBO HD channel was airing the 4:3 pan-and-scan version of a film which they cropped in order to fill a 1.78:1 screen, so the SD versions actually have more of the frame visible. It does the job well enough for this comparison, so the "is that cropped SD on HBO HD?" thread will have to wait for another day. Note that all images are 99% quality JPEGs this time since the PNGs were giving me trouble with dithering. Also, all images have been scaled from their transmitted resolutions for display at their intended aspect ratios (Reference = 16:9, Bell and Dish = 4:3).

First up is the reference frame. As mentioned, it's cropped, so the top and bottom are missing in comparison to the shots below. Otherwise, it sets a pretty good standard for comparison. It shows how much or little detail there should be in the background (wall and guy drinking soda) and foreground (other guy's face).



Second is the Bell sample. Not much to say. Looks a tad soft, but so does the reference frame. Probably a property of the SD transfers distributed to the networks. It doesn't look blurred or distorted around the edges. The hue is a bit different, but that doesn't make the frame doesn't look bad by any means. Face details look quite alright for SD and the image fills the frame neatly (no black areas around the frame like the Starz and Encore samples from the previous comparison, which is good since those make compression more difficult, with or without closed caption static/data present).



Last is the Dish sample. First thing I notice is that the colors are washed out. The face looks unnaturally pale compared to the other two samples, and everything else looks a bit faded. Next I notice a typical result of Dish's pre-filtering: more noise, rather than less, all over the frame. Some of the noise appears to be pixelation, but a lot of it appears to have been coded into the video. The background is full of all kinds of little chunks of color where mostly smooth, out-of-focus objects (relative to the face) are supposed to be. Looks like there's splattered gunk everywhere. Yuck.
Focusing on the face, at first glance it may appear that some of the wrinkles and such are more clearly defined in the Dish sample than in the others because they are darker. Closer inspection reveals that the tiny facial features in the Dish sample are actually heavily pixelated and noisy. They have used some method to try to emphasize the areas where details originally were by replacing the actual details with blotches. Using a 2-3x zoom on the Bell sample, the face basically just looks like a zoomed in face. Using a 2-3x zoom on the Dish sample clearly reveals that the face is very blocky and pixelated. The large amount of noise creates some illusion of detail, but it is the same kind of illusion created by film grain or light static on a TV. When blown up, shown on a high-resolution display, or both, the deception becomes far more apparent. The method they are using to pre-filter the video is almost definitely based on the assumption that their viewers are all using low-res CRT TVs since those hide the noise (fake detail) fairly well. Especially considering that they're peddling HD services and providing price advantages to customers who sign up for both HD and SD service rather than only HD, you would think they might have the tiniest clue what kind of televisions an ever-increasing number of their customers have, and what their SD video looks like on those TV sets. Apparently not...



Here's a 3x upscaled section of the face from each sample (Reference, Bell SD, and Dish SD in that order). Note that I have adjusted the hue on the Reference sample to make it easier to compare to the Bell sample.

Reference:


Bell:


Dish:


Considering that I don't ever want to hear someone claim that the SD "isn't that bad" on Dish and the fact that 2.38Mbps just for video is above-average on Dish, I want to show one more image to make a point. I want to put to rest the idea that SD picture quality has to be bad simply because what we're given is bad. So, what does the Bell frame look like if it is reduced to 544x480 and then expanded back to 640x480 (the display resolution based on the 4:3 aspect ratio, not its original 704x480 resolution)?



That's right, it still looks quite good, so the resolution change isn't the cause of Dish's terrible picture quality. Bitrate isn't the problem either because it was established from the beginning that there is only an 8% difference in effective bitrates between Dish and Bell. It all seems to come down to the wretched manner in which Dish is filtering the video prior to encoding it and probably also the use of low-quality MPEG-2 encoders. They are destroying the picture quality and probably making it more noisy and harder to compress rather than easier. In fact, when encoding all of the JPEG images for this post, the encoder had a more difficult time compressing the Dish images than the Bell images. That alone should certainly say something about the relationship between compressibility and picture quality.

Dish's goal at one point may have been to increase compressibility without creating problems, but if it was then they failed rather badly. I hate to say this, but the Americans really have something to learn from the Canadians in this case. It looks like they know how to do picture quality right. It's just too bad they aren't allowed to carry many channels from the US, otherwise I would be happy to make the switch to north-of-the-border TV service.


The next comparison will probably be DIRECTV SD letterbox vs. DISH SD letterbox. Fun, fun, fun. :)
 
Interesting read, to me the Bell SD feed looks better than the Dish HD channel

In some spots it does, others it doesn't. The "HD" frame is actually the one just before or after the frames from SD sources since it was the nearest progressive frame and it was a very low-motion scene. That difference could account for some minor variations in some details. Since the "HD" in this case is just upscaled SD, it should look about the same as good SD.
 
In response to OPs original thought, I don't believe low quality SD is being used to drive HDTV sales.

The reason I believe this?

The majority of HDTV owners are still watching SD on their HDTV, think they are actually watching HD just because they bought an HDTV, and have no idea that there is HD quality programming they need to subscribe to.
 
The majority of HDTV owners are still watching SD on their HDTV, think they are actually watching HD just because they bought an HDTV, and have no idea that there is HD quality programming they need to subscribe to.

A statement like this requires data to back it up. Do you have a source or is this just your opinion?
 
Well SD is SD and in the future will not matter due to we should be back to one format which is HD. So SD will at some point time will be gone once all the Networks get their HD Channel up and running. There will be no need for SD..
 
The majority of HDTV owners are still watching SD on their HDTV, think they are actually watching HD just because they bought an HDTV, and have no idea that there is HD quality programming they need to subscribe to.
A statement like this requires data to back it up. Do you have a source or is this just your opinion?
How about...“In the US, there are more than 39 million households with an installed HDTV set,” according to Mike Paxton, an In-Stat analyst. “However, only 22 million of those are HDTV households, meaning that 17 million US households with an installed HDTV set are not watching HD programming.”

Significant Gap Between HDTV Ownership and HD Programming Usage Among US Households
 
Well SD is SD and in the future will not matter due to we should be back to one format which is HD. So SD will at some point time will be gone once all the Networks get their HD Channel up and running. There will be no need for SD..

May I ask who you think is going to buy new HD receivers for everybody? We've seen lately how the coupon program for ATSC receivers didn't work like it was supposed to, and that didn't even affect the majority of the population. During the years prior to that mess and the latest (and probably final) defined switchover date, there were numerous delays in the switch due to lack of proper preparation.

Currently, cable and satellite providers sell/lease tons of SD-only STBs, almost definitely more than HD STBS, and it seems very unlikely that full swapouts of all such STBs could occur any time soon. If a swapout were to occur, it would have to begin with phasing out SD STBs altogether, across all cable and satellite platforms, including analog cable. I can't imagine such a process even beginning for at least 10-20 years from now.

Bandwidth is also a huge issue. Even with H.264, the satellite providers absolutely have no means of carrying an HD equivalent of every SD channel without having to launch several additional satellite each, replacing every dish for every subscriber, and still probably degrading the quality of the channels below that of high-quality SD (due to their established idea of "good" pre-filtering and encoding methods) just to carry everything.

In short, I expect SD to be around for a long time into our futures. I just hope Verizon isn't lying to us and that we'll be able to switch to FiOS TV by this October. I think I can hold out for that long without going blind from watching blocky, blurry TV.
 
I can't imagine such a process even beginning for at least 10-20 years from now.

The market and it's underlying philosophy has changed and continues to do so. The process has already begun and there is no telling what we may find in 10 to 20 years.
 
How about...“In the US, there are more than 39 million households with an installed HDTV set,” according to Mike Paxton, an In-Stat analyst. “However, only 22 million of those are HDTV households, meaning that 17 million US households with an installed HDTV set are not watching HD programming.”

Significant Gap Between HDTV Ownership and HD Programming Usage Among US Households

Thank you for the reference. The article states that of 39 million households with an installed HDTV set, 17 million (or 43.6%) are not watching HD programming while 22 million (56.4%) are.

While 43.6% is a very large percentage of HDTV owners without HD service, it does not constitute a majority, which was my point.
 
Thank you for the reference. The article states that of 39 million households with an installed HDTV set, 17 million (or 43.6%) are not watching HD programming while 22 million (56.4%) are.

While 43.6% is a very large percentage of HDTV owners without HD service, it does not constitute a majority, which was my point.

The year before it was above 65%, which is a majority if you are talking about just those who don't even subscribe to HDTV service.

I'm saying when you add those three categories I mentioned you get a solid majority who are still watching SD. It's a shame that so many are oblious to the fact that they are not even looking at HD but think they are.

Millions of hdtv owners and have no idea....
 
The market and it's underlying philosophy has changed and continues to do so. The process has already begun and there is no telling what we may find in 10 to 20 years.

I would have agreed with you a couple of years ago on this, but the rate of new HD channel deployment, especially on satellite platforms, seems to have dropped off greatly recently. Maybe there are some nice graphs around that show HD channels added per month, the percentage of SD-only vs. SD+HD channels existing (including those not offered by some cable/sat providers), and the percentage of SD-only vs. SD+HD channels actually offered to subscribers.

Dish's ability to increase the number of HD channels is quite limited right now, especially since they're trying to mirror all channels for both the Western and Eastern Arc systems. Add too many more, and HD picture quality will start to go south pretty fast. If by some chance Ceil II improves HD capacity, that will only involve Western Arc subs. Eastern Arc Capacity could be improved a bit by switching to 8PSK at 72W (16 xpndrs * 10Mbps/per xpndr = 160Mbps, or about 28 HD channels at the current bitrates) and possibly for some of the non-8PSK transponders at 61W, but there are quite a few more than 28 channels left that would need to upgraded to HD before they're all in HD.

And again, the cost of replacing all of the STBs and dishes would be absolutely immense. That could easily be the single largest source of delays among all of the other factors.
 
***

Users Who Are Viewing This Thread (Total: 1, Members: 0, Guests: 1)

Who Read This Thread (Total Members: 1)