Master Feed vs. Dish Network Shootout

[pendragon]

There has been a fair amount of banter back and forth about the relative merits of 4DTV master feeds compared to pizza dish HD-lite. I actually did some direct comparisons a few months ago with the raw transport streams, but after satisfying myself with the results, I tossed the files of the losers. I'm going to try to find some of the salient examples on backup tapes, but that will take some time. But I wanted to provide an example to start, patched together as best as I can.

A few months ago I collected a number of cases where I had recorded a master 4DTV or DVB-S feed (by subscription or ITC) at the exact same time as I recorded Dish Network's re-encoded version via that subscription. Since then I have dropped most of my DN premium channels and sadly the 4DTV HD feeds from HBO and STARZ are no longer available. Until I find the old files, this makes any comparison indirect and potentially less valid. But it's still better than nothing.

I haven't gotten around to canceling my $0.01 per year subscription to Cinemax on DN, so I did a quick comparison of what was showing this afternoon with a list of my master feed recordings and found "Fatal Attraction". This isn't the sharpest movie around, and my master feed is a little low on the bit-rate, but I guess beggars can't be choosers.

I recorded both the DN SD and HD versions on 5MAX through my R5000s. I use similar R5000s for Shaw and for 4DTV in the past. For DVB-S and DVB-S2 I simply use a PC-based tuner. By capturing the original transport streams, I can make meaningful comparisons without any extra processing that might bias the results. To ensure this is clear: the master feed was not what DN used as a source for the uplink. But it was probably similar.

So here we go:

Master feed was 1920x1080i, MPEG-2 at about 10 Mbps.
Dish Network 5MAX HD was 1440x1080i, H.264 at about 4 Mbps
Dish Network 5MAX SD was 544x480i, MPEG-2 at about 1.7 Mbps

For this example I have grabbed one frame from each. This is a very static shot, which will not show artifacts typical of a scene with a lot of action. When I get some more time I'll grab an example of the latter. I compressed the raw images with Photoshop to JPEG at the 10 "Maximum" quality setting. I'll leave the comparisons mostly for you to make, but I'll point out that the Dish Network HD version isn't that bad when compared to the master feed. However the Dish Network SD version is pretty awful. The SD is full-frame, rather than letterboxed or pan & scan.

 

[pendragon]

For a little more fun, I enlarged the right lower corner of each frame described in the original post. This makes it pretty easy to do a comparison of each flavor.

What I would really like to find is a master feed in SD, but until then I came up with an alternative. I took the HD master feed and downrezzed it to SD at a fairly high bit rate (6 Mbps) with a good MPEG-2 encoder. I then grabbed a frame from that and enlarged the corner in the same manner. This likely represents as good as one will ever get with a SD master feed.

So see for yourself. I personally would want something better than the master feed, but I can live with it. I could probably even stand the DN HD version if the whole movie looked like this. It's certainly beats the pants off the SD master simulation and the DN SD version. I will try to find a scene more representative of a lot of action to test the DN encoders on that.

 

[pendragon]

The file names of the attachments in my first two posts are self-descriptive, but in case you don't download them, here is the legend, right to left:

First post:

1. Master feed HD
2. Dish Network HD
3. Dish Network SD

Second post:

1. Master feed HD zoomed
2. Dish Network HD zoomed
3. Dish Network SD zoomed
4. Simulated master feed SD zoomed (from master feed HD)

 

[Comptech]

I would be happy to record a TS stream of a master SD feed for you tomprrow to compare. I would do it tonight but it is time for bed, need to be at work by 5AM.But I still defend TVpro that alot of it has to do with how your TV upscales the video.Before I moved I had both dish HD and 4DTV and I will tell you from personal experience that the mater Discovery feed on the 4DTV looked better than the Dish HD Discovery.Again both myself and TVpro have Sharp aqous TV's so maybe that has alot to do with it.But on the other hand how do you explain why the anolog feeds using a whole transponer look so good also? It all boils down to bandwith, and the cables and DSS's want us to think what they offer is true HD,when a simple look at a TNN hd 422 NBA feed will show you true HD.I guess what I am trying to say that is a SD master can look as good if not better than a HD compressed feed,we are just taught that this is what HD should look like by the big cable and DSS companys.

 

[pendragon]

I would be happy to record a TS stream of a master SD feed for you tomprrow to compare.

If you're going to record something, it would be best to coordinate. For example if I have a HD master feed of a movie and can record DN's HD and SD uplinks at the same time as you record as SD master, we're pretty much set. Otherwise I may still have a complete set on backup tapes, but it will take some digging as I had no intention of keeping them.

But I still defend TVpro that alot of it has to do with how your TV upscales the video.Before I moved I had both dish HD and 4DTV and I will tell you from personal experience that the mater Discovery feed on the 4DTV looked better than the Dish HD Discovery.Again both myself and TVpro have Sharp aqous TV's so maybe that has alot to do with it. But on the other hand how do you explain why the anolog feeds using a whole transponer look so good also?

Do I interpret correctly that you say that 4DTV Discovery SD looks better than DN Discovery HD? I have never done a comparison of those. But a lot depends on the source material. There is certainly a fair amount of HD where little care was invested. It may be 1080i, but it has no more resolution than 480i.

I have some professional scalers that can make SD look pretty good, but it's partly a trick to the mind. Edge enhancement and color games can fool a lot of people into thinking they are seeing a higher-resolution picture, when they are not. If one has a trained eye, you spot the hocus-pocus and recognize that rescaling cannot add any more resolution than is there. If you look at the zoomed photos, one cannot read much of anything on the milk carton in the simulated SD picture. Upscaling this might make the edges 'look' sharper, but you won't be able to make the type legible. The master and Dish HD images are MUCH better. All of the above applies to analog, like it or not.

There is a technique called Super Resolution that can use multiple frames to truly enhance the resolution of a SD video. I've not seen an example in product or in labs that gets it anything close to HD, however. Last time I checked only Toshiba offered a consumer product capable of doing this. I did some quick and dirty tests and while it was better than a traditional scaler, it wasn't impressive.

It all boils down to bandwith, and the cables and DSS's want us to think what they offer is true HD,when a simple look at a TNN hd 422 NBA feed will show you true HD.I guess what I am trying to say that is a SD master can look as good if not better than a HD compressed feed,we are just taught that this is what HD should look like by the big cable and DSS companys.

I'm not saying the cable or sat providers are giving us anything close to real HD. I have a fair number of master feeds at 30+ Mbps that are outrageously great. But I believe people are fooling themselves into thinking 4DTV SD masters are only a small step down. Sorry, they only have 1/6 the resolution of real HD. The cable and sat providers may not offer a 6X improvement in resolution with their HD, but I have presented evidence that it's more than 1X. If you can post evidence to the contrary, I'd love to see it.

 

[tvropro]

Here are my contributions, all self explanatory, these are from ts streams:

HD Lite The Big Lie...

C Band vs DSS HD Lite Full Size Images

The C Band HD Master Page

There are some SD masters here but taken with my camera from the tv screen:

TvroPro's Photo Gallery

I have some other SD 4DTV masters taken with my camera from the AQUOS on my tripod. I will look through them and post some later.

In SD's defense 480i may have less pixels than HD lite, but a master is a master. If the bandwidth & bitrate are there it's clean. You can have 1080p that looks like crap. My local ABC OTA affiliate uses 720p they are running 2 HD channels and one SD channel. The HD looks terrible. Especially when they go to a wide shot. Faces in the distance are blocks out of place. If that's HD I'd rather watch something I was DXing on analog back in the day.

 

[tvropro]

But on the other hand how do you explain why the anolog feeds using a whole transponer look so good also? It all boils down to bandwith, and the cables and DSS's want us to think what they offer is true HD,when a simple look at a TNN hd 422 NBA feed will show you true HD.I guess what I am trying to say that is a SD master can look as good if not better than a HD compressed feed,we are just taught that this is what HD should look like by the big cable and DSS companys.

It does boil down to bandwidth. Analog is receiving 27 to 36 mhz of bandwidth for one channel. It's also in real time unlike digital and is not processed.

Disney's SD master is running 6.14mbps per channel on G-14 according to Photoman in another post. Now posted here Dish HD lite is running 4 mbps. Hell 4 is a minimum IMHO for SD. How the hell can they put HD efficiently in that space? Something has to give. I heard all the stories about Mpeg 4 vs Mpeg 2 etc but still something had to give. Dish network also uses things such as edge enhancement to make you believe your getting a sharper image. The real lack of bandwidth rears it's ugly head when the action starts moving on Dish HD. When it starves and looses information it becomes digital artifacts plain and simple. Mpeg 4 handles them differently than Mpeg 2. It actually blurs the artifacts to try to mask them.

I want my Bandwidth! :rant:

 

[tvropro]

SD 4DTV Master pictures

Here are some 4DTV (hidden) SD masters from the AQUOS screen taken with my camera. I even blew up one image to show the lack of artifacts around the images and clarity of wording etc. In short lots of information (bandwidth) there. BTW these are all taken with Smart stretch on so you can see how well the Sharp handles the up conversion to 1080p from 480i. The 4DTV is hooked up using a Acoustic Research Pro Series 2, S video cable 3 foot long.

Open them up to full size to get to the nitty gritty.

 

[tdti1]

Well of course 4dtv SD is not as good as pizza HD, 4dtv is good for SD thats it. I use the 4d to move the dish most of the time for my dvb HD

 

[B.J.]

I've been reading this discussion with interest, mainly because I have observed lots of cases where good SD looks pretty good compared to HD, dependent upon a lot of different factors, and I've always been curious about just what causes a lot of the things I've observed. Personally, given the things I have access to to watch, I prefer HD over SD in almost all cases, although there are times that I've turned off the real low quality HD to watch the SD equivalent.

I've also made hundreds of still frame captures over the years, trying to compare different sources, but unfortunately have deleted most of them. Just a couple days ago, I captured some frames from a ~11mbps DN signal, and just noticed something about my capture that I also noticed in the captures in the post below:

There has been a fair amount of banter back and forth about the relative merits of 4DTV master feeds compared to pizza dish HD-lite. ....
....

So here we go:

Master feed was 1920x1080i, MPEG-2 at about 10 Mbps.
Dish Network 5MAX HD was 1440x1080i, H.264 at about 4 Mbps
Dish Network 5MAX SD was 544x480i, MPEG-2 at about 1.7 Mbps

For this example I have grabbed one frame from each. This is a very static shot, which will not show artifacts typical of a scene with a lot of action. When I get some more time I'll grab an example of the latter. I compressed the raw images with Photoshop to JPEG at the 10 "Maximum" quality setting. I'll leave the comparisons mostly for you to make, ...
....

What I was curious about, re the above, and my own DN captures, is that even though the DN video is 1440x1080i, the frames I captured came out as 1920x1080 images. On the other hand, when I did a thumnail capture in TSREADER, they came out as 480x360, ie 1/3 the 1440x1080, indicating that the images actually are available there in the original 1440x1080i, unfortunately TSREADER won't give me a full frame capture, just thumbnails.

I've been making captures by making recordings, then using VideoReDo to capture frames, but as mentioned, it seems to scale up the raw images into the 1920x1080i, instead of giving the raw image. I know that in most cases, this is probably desirable, as it gives a picture that isn't artificially stretched, however when trying to compare the quality of the image before the TV scales it to fit it's screen, I think I'd rather see the raw 1440x1080 image rather than trusting VRD to do it's scaling.
I thought that perhaps VLC might give me true raw images, as I had noticed a setting to turn off scaling and turn off some feature where it adjusted for image resolutions that were different from 1920x1080i, but turning those things off didn't change things, I still got the snapshots in 1920x1080.

Anyway, the question here, is..... first of all, why is this happening? I assumed that that the mpeg4 stream must indicate both the actual size of the image plus the intended screen aspect ratio, so that TVs and video programs do their scaling up to the intended 1920x1080 size, however in TSREADER, these particular recordings I made had no mention of the aspect ratio nor was there any mention of 1920 at all, so I don't know why the screen captures are coming out in 1920 instead of 1440. Is there any easy way to generate raw images that haven't been scaled up to the 1920x1080i?

In any event, I think the high (for them) bitrate DN HD doesn't look bad at all, although clearly not as good as good network HD. At first, I was not at all impressed with MPEG4 video, but apparently that was mainly because the only way I had to view it was via VLC on a slow computer. But viewed via PopCornHour or Azbox going through HDMI to a 1080p TV, the high bitrate DN MPEG4 stuff looks pretty good to me. Unfortunately most of their HD is really LOW bitrate. I sure wouldn't watch an NFL game on one of their local channels. I tried that last fall, and it was TERRIBLE quality. It's OK when there is no motion, but even the slightest motion blurs out the whole screen.

 

[pendragon]

Here are my contributions, all self explanatory, these are from ts streams:

HD Lite The Big Lie...

C Band vs DSS HD Lite Full Size Images

The C Band HD Master Page

Yes, HD master feeds are better than pizza dish HD-lite. I think we can agree on that. I would gladly pay a premium for getting the HD masters, but they are all but gone from 4DTV. Nothing we can do about that.

On FTA I can get a lot of sports feeds ITC and of course the OTA networks. Anything else is hit or miss. However I'm more interested in movies and the fine arts, and other than PBS on 125W, there's not much else I can depend on for a HD master. Not that the 4DTV HD masters were that good. Even FTA PBS HD generally looks better than 4DTV HBO HD did at the end, and far better than 4DTV STARZ HD was.

My point is that SD masters, while they offer the pinnacle of the art for SD, are still SD. Upscaling them and stretching them does not make them HD anymore than DN/DTV HD-lite. My first thesis is that HD-lite has progressed to the point where it is demonstrably closer to the master HD feed than a best-case SD feed could EVER be on at least static scenes. I hope to tackle the comparison of high action scenes at a later time in this thread.

There are some SD masters here but taken with my camera from the tv screen:

TvroPro's Photo Gallery

I have some other SD 4DTV masters taken with my camera from the AQUOS on my tripod. I will look through them and post some later.

I'm not saying SD masters can't look great. Heck, I've even worked with analog NTSC in the 1970s that looked better than some of these shots. But none of these are impressive when compared to either real HD or HD-lite. But you haven't done that.

 

[Scott Greczkowski]

So here we go:

Master feed was 1920x1080i, MPEG-2 at about 10 Mbps.
Dish Network 5MAX HD was 1440x1080i, H.264 at about 4 Mbps
Dish Network 5MAX SD was 544x480i, MPEG-2 at about 1.7 Mbps

For this example I have grabbed one frame from each. This is a very static shot, which will not show artifacts typical of a scene with a lot of action. When I get some more time I'll grab an example of the latter. I compressed the raw images with Photoshop to JPEG at the 10 "Maximum" quality setting. I'll leave the comparisons mostly for you to make, but I'll point out that the Dish Network HD version isn't that bad when compared to the master feed. However the Dish Network SD version is pretty awful. The SD is full-frame, rather than letterboxed or pan & scan.

Some things to note...

1) Dish's HD version is MPEG4, so while only 4Mbps it is much more efficiant than MPEG2. This is why you noted that the DISh quality was good on the HD.

Now as far as the SD feed goes, Dish gets that from Showtimes SD feed. Showtime is the one who downreses the content for SD, so its not DISH doing the conversion to SD. With that said though, yes the SD channel does look like crap, whic is why we all want HD.

 

[pendragon]

Disney's SD master is running 6.14mbps per channel on G-14 according to Photoman in another post.

That's why I did my best-case SD master simulation at a 6 Mbps rate. All that bandwidth didn't make the type on the milk carton readable. I'm saying best-case because I used the finest MPEG-2 encoder I had lying around, which is two-pass and definitely not real-time. From what I've seen on 4DTV when I was subscribed, normal SD masters are not as well encoded.

Now posted here Dish HD lite is running 4 mbps. Hell 4 is a minimum IMHO for SD. How the hell can they put HD efficiently in that space? Something has to give. I heard all the stories about Mpeg 4 vs Mpeg 2 etc but still something had to give. Dish network also uses things such as edge enhancement to make you believe your getting a sharper image.

In an earlier thread you suggested that what really matter is how the video looks, not what the numbers say. In deference to you I have focused on pictures, not on rates. While the DN HD frame is not quite as good as the master HD, I see more detail in it and can read far more of the type on the milk carton than in the best-case simulated SD master.

I don't consider H.264 the penultimate video encoder any more than I did MPEG-2 in its heyday. The definition of MPEG-2 incorporated the technology limitations of the time, which if you wanted to make a real-time encoder, you had to be realistic about the amount of processing that could be done. Less processing means the compression is not as efficient as in H.264. Conceived many years later, the H.264 spec assumed far more processing was available than for MPEG-2, something that is very apparent if you run excellent examples of both as software encoders on the same machine. H.264 is far more adaptable to what is going on in the frames than MPEG-2, meaning it can focus its bits more effectively.

Because I did not want to get carried away on bit-rate studies, I have not addressed some of the finer points in the original posts. But the quoted rates in the first post are simply over the entire movie, not what was used by the encoder for the particular frame. That is much more interesting, but much harder to evaluate and understand.

The simulated SD master for the sequence of frames I converted was set to a CBR of 6 Mbps to match the Disney rate. That's probably a lot more than was needed for a static scene. I only did a quick check of the master feed bit rate, and while it was variable across the movie, it didn't vary that much. DN seems to allow their encoders more flexibility in adjusting rates, which means they can provide higher rates for frames that need more. So the 4 Mbps is simply an average, not the maximum that can be used. On Shaw, I have seen cases where the encoder went crazy handing out bits for an action scene, while turning the rate way down for some static frames. How does one analyze or compare this?

The real lack of bandwidth rears it's ugly head when the action starts moving on Dish HD. When it starves and looses information it becomes digital artifacts plain and simple. Mpeg 4 handles them differently than Mpeg 2. It actually blurs the artifacts to try to mask them.

Both H.264 and MPEG-2 break the picture into macroblocks as the action begins to stress the available bit rate. Because of its vintage technology, MPEG-2 uses cruder increments and cannot adapt the macroblocks as well to the parts of the scene that are showing the most change. H.264 has some capabilities that help mask this, that do not exist in MPEG-2. The net effect of either is that the resolution of the image is reduced in action shots. This happens to correspond to the physiology of the human eye and brain - we do not perceive the same detail with motion as we do in a static image.

Problems start to arise for any codec when it simply does not have enough bits to 'fool' the eye any more. That's when we start to see dramatic artifacts. H.264 can stay in the game with about half the bit rate of MPEG-2, but it can't do much better. However the days of CBR encoding for DBS distribution are long over. These providers for better or worse mux many channels together into one transport stream. Unless every channel is running high-bit rate action scenes, the static scene channels will have extra bits to give away to those requiring more. I have noted that DN has made good progress in this regard over the past years. This can do a good job of ameliorating an apparent low average bit-rate.

Other than morbid technical curiosity with regard to video quality, I have zero interest in sports and waste little time in that regard. In my possibly ignorant view, it seems the DBS providers bit-starve the locals, RSNs and other temporary channels employed to carry sports events, at least compared to certain of the other HD channels. This probably makes economic sense to them, but I can see why it incenses those who care about sports. However I don't see this as too problematic because a lot of sports feeds are available FTA ITC at some of the highest bit-rates known to human-kind. If you care about sports there is always FTA. The same is much harder to say with regard to premium movies and the fine arts.

I hope to cover some comparisons between master feeds and HD-lite for high motion scenes once I come up with a straightforward method of presenting it.

 

[pendragon]

Here are some 4DTV (hidden) SD masters from the AQUOS screen taken with my camera. I even blew up one image to show the lack of artifacts around the images and clarity of wording etc. In short lots of information (bandwidth) there. BTW these are all taken with Smart stretch on so you can see how well the Sharp handles the up conversion to 1080p from 480i. The 4DTV is hooked up using a Acoustic Research Pro Series 2, S video cable 3 foot long.

Putting aside my concerns about using a camera pointed at a screen to evaluate the quality of a feed, I will note that these screen shots look just like decent SD. Upscaling them may smooth out the pixels, but it doesn't enhance the resolution. Your scaler is decent, but it is adding a number of upscaling artifacts that are as disturbing to me as edge-enhancement and motion artifacts. I also do not care for the distorted linearity of the so-called 'smart stretching'. I like my human bodies to stay the same shape as the camera pans across the screen, or have type in subtitles and credits not appear as it came out the FX shop. Some people are more sensitive to this than others.

 

[pendragon]

What I was curious about, re the above, and my own DN captures, is that even though the DN video is 1440x1080i, the frames I captured came out as 1920x1080 images. On the other hand, when I did a thumnail capture in TSREADER, they came out as 480x360, ie 1/3 the 1440x1080, indicating that the images actually are available there in the original 1440x1080i, unfortunately TSREADER won't give me a full frame capture, just thumbnails.

DN and many others improve the compressibility of their video by downrezzing it somewhat. When they take a 1920x1080 frame from a master feed and turn it into 1440x1080, they are no longer using square pixels. meaning the resolution per unit measure is less in the horizontal dimension than vertical. My memory escapes me, but I vaguely recall the human eye/brain may be less sensitive in this manner and this is why anamorphic techniques have been used in the film world for so long.

When the video is played back from say a DN source, the decoder must rescale the video back into the square pixels on the screen, but that is pretty much child's play in one dimension. DN does this in their receivers, but only after the TS are decoded. For my tests I used a codec that outputs the raw, unscaled frames from the TS as uncompressed images. I took these into Photoshop and scaled them to the proper aspect ratios, which is why the pixel dimensions come out as ideal.

By the way, HD is simple compared to SD. For many historical and technical issues, there is almost no such thing as a square pixel in the capture stage for SD. Everything gets scaled and/or cropped. There is an interesting piece about this on the web:

A Quick Guide to Digital Video Resolution and Aspect Ratio Conversions

 

[Scott Greczkowski]

DN and many others improve the compressibility of their video by downrezzing it somewhat. When they take a 1920x1080 frame from a master feed and turn it into 1440x1080, they are no longer using square pixels. meaning the resolution per unit measure is less in the horizontal dimension than vertical.

This was and is true to a point, however all the new HD launched recently on DISH on the 72.7 satellite is full 1920x1080.

I think they got the encoders (finally) to where they should be.

Remember these live MPEG4 encoders were basically made for DISH and DIRECTV, they were not an off the shelf item when they started moving to MPEG4. The MPEG4 encoders have progressed very nicely over the past few years.

 

[tvropro]

It would be better for the purists to forget about compression all together, and all the processing and go one for one, like with analog. All the years I had C band, analog was the cats meow because of it's amount of bandwidth. I know these days it's all about the $$$ numbers with the bean counters and thats where it all suffers.

Unfortunately not making the master HD feeds for the cable channels available for us stinks. Those that need that fair have to bow to the dirty 18" providers or cable to get a beat down versions. The true meaning of eye candy HD suffers though. Thats what HD was all about wasn't it? To be a perfect High Definition copy.

The LSD & cable providers shouldn't hype it up though for something it really isn't. But then again there's a sucker born every minute.

 

[pendragon]

This was and is true to a point, however all the new HD launched recently on DISH on the 72.7 satellite is full 1920x1080.

Very interesting. I'm actually set up for both 61.5/72.7/77 and 110/119/129 on independent switches, but we've been using the WA for the past year or so because the downlinks have been better with the newer birds on that side. I should move one of my R5000 receivers over to the EA and compare the two. So many things to do, so little time.

 

[pendragon]

It would be better for the purists to forget about compression all together, and all the processing and go one for one, like with analog. All the years I had C band, analog was the cats meow because of it's amount of bandwidth. I know these days it's all about the $$$ numbers with the bean counters and thats where it all suffers.

The only problem is uncompressed 1080i video requires about 1.5 Gbps. It would take the whole C-band or Ku-band payload of one bird to provide a single HD channel. Fortunately most of this information is redundant and one can cut this back by a fair fraction even with lossless compression. However even lossless compression isn't economically viable for satellite distribution.

Be careful what you ask for. Fiber can do lossless compression and even uncompressed HD. There have been some recent demonstrations of this, perhaps the past Super Bowl if I recall correctly. Fiber to the home would be great, but if it's only used for backhauls, things are going to get worse before they get better.

 

[pendragon]

Better source comparison

Grabbed some more video off DN last night from a movie where I have a slightly improved 4DTV master feed for comparison. The movie transfer also appears better than for the original post:

1. C-band 4DTV HD, 1920x1080i, MPEG-2, 13.4 Mbps
2. Dish Network 5MAX HD, 1440x1080i, H.264, 4.6 Mbps
3. Dish Network 5MAX SD, 544x480i, MPEG-2, 1.9 Mbps

I chose a frame from the title sequence because it's easy to compare the sources. Rather than bother with a MPEG-2 conversion to simulate a master SD feed, I chose an even more optimistic case: direct conversion in Photoshop from the C-band HD master down to SD. This is extremely optimistic for what one might see on SD master, because the effective bit rate would be around 220 Mpbs as each frame is essentially uncompressed. I figure this way no one can accuse me of biasing the results to make SD feeds look bad.

Following that I did an ideal upscaling of the 'SD master' to HD. Again, this is unrealistic, but represents about the best one could ever hope for when watching a SD feed on an upscaling screen. No stretching was done.

4. C-band 4DTV HD downscaled to simulated SD feed, 640x480, uncompressed
5. Simulated SD feed upscaled to HD, 1920x1080, uncompressed

Here are the frames, hopefully in order as shown above. In the next post I'll zoom in and show how they compare.