Okay, here's another comparison: Dish Network SD Letterbox vs. DIRECTV SD Letterbox.
More than any other content, these two sat providers destroy the details in widescreen or letterbox programming on SD channels. The same kind of bad filtering/encoding is used on SD source video with large, easily detectable black areas as is used for full-screen SD content. They should be detecting the black areas, recognizing that those will compress down to almost nothing, and minimizing filtering in the non-black areas. This would boost the picture quality because reducing filtering allows more details to reach the encoder, and the encoder can spend the bits it would have used on the normally full part of the frame (which is pure black in this case) on better encoding of the middle of the frame. They don't appear to be doing this for SD channels, the result of which is that widescreen/letterbox content on those channels can be almost unwatchable on high-res displays, especially for 2.35:1 films. A somewhat related problem exists with the HD channels on Dish in that bits are excessively over-allocated to pillarboxed, upscaled SD content stealing bits away from channels airing real HD content (full-screen or letterboxed) that could make far better use of those bits.
Channels including IFC, TCM, Sundance, and FMC regularly air letterboxed documentaries and films. Only one has an HD equivalent (IFC) and neither of the US satellite providers currently carry it, so the "you should watch the HD version instead" argument can't be made here. I have no doubt that the SD feeds of these channels look excellent based on viewing experiences reported by FiOS TV customers, but unfortunately I don't have any samples from feeds or FiOS for a three-way comparison at this time.
Note that the previous comparisons were of different airings from different channels and/or different times. This comparison is from a single airing of the movie 1941 on TCM last year.
First up is Dish Network. The resolution was 544x480 and the average bitrate was 1.45 Mbps. This bitrate is notably lower than the previous samples because the pure black areas compress down to nothing. The portion of bits that would otherwise have been used for the remainder of a "full" SD frame are simply given up rather than reallocated to where it's needed in the middle of the frame (to give it more detail and such). The image looks very noisy all over, including areas that obviously lack detail and should be smooth such as walls. In the Dish vs. Bell comparison it was demonstrated that Dish inserts noise prior to encoding where no detail or noise existed in the source material, so the same appears to be true here also. A small explosion from a gunshot is starting in the background and there is a bit more blockiness in and around it as a result. There really shouldn't be a shortage of bits available on-demand to prevent blockiness since so many are being tossed aside due to the black areas, so this shows that the encoder and/or load-balancing system don't work well. This also shows that their filtering methods don't always (or ever?) actually prevent the artifacts they're supposed to at below-average bitrates.
Dish Network (544x480 scaled to 640x480 for proper aspect ratio):
Dish Network (Crop of relevant picture data at 1.5x zoom):
Second is DIRECTV. The resolution was 480x480 and the average bitrate was 1.74 Mbps. This bitrate, combined with the difference in resolution, should effectively provide DIRECTV with effective 33% more bandwidth. The raw bitrate difference alone is 20%, so DIRECTV's picture quality can't be worse than Dish's, right? Well, yes and no. The picture quality can be worse because DIRECTV outdid Dish in terms of softening the picture. The picture quality can be better for at least two reasons. First, the area around the explosion is definitely a tad blocky, but the blocks are softened so they aren't nearly as noticeable as in the Dish sample. Second, if you take a closer look at the DIRECTV sample, you'll notice that everything is what is appears to be: softened forms of the original source material. In the Dish sample, details and edges sometimes look like what they are, but often look like blotches (because they are blotches) that aren't necessarily where or what they should be.
DIRECTV (480x480 scaled to 640x480 for proper aspect ratio):
DIRECTV (Crop of relevant picture data at 1.5x zoom):
One note on the JPEG images: similarly to the previous comparison (Dish vs. Bell), the samples from Dish were less compressible when making the JPEG files by 14% for the uncropped frames and 12% for the cropped frames. This suggests that they contain more detail, more noise, or both that the DIRECTV sample. We know for sure that the Dish images contain lots of noise, so if by some chance there are also any well-defined (non-blotch) details in the image, the amount of it limiting compressibility relative to the DIRECTV images can't be very high.
I don't think it can really be said definitively that SD quality is better on Dish or DIRECTV. Dish offers a noisy, noticably pixelated, soft picture that places blotches of color where details would otherwise be, and then places blotches everywhere else too. DIRECTV offers a softer but less pixelated picture without introducing all kinds of noise. What can be said definitively is that DIRECTV's picture should have looked a noticeable amount better than Dish's, especially in terms of sharpness, however they too are very clearly using filtering and/or encoding methods that go far beyond what is necessary to improve compressibility. They're both not good compared to what they should be, especially given the resources of both companies, but quality-control has never been a high priority I suppose.
If I were to guess, I would say DIRECTV's SD should look better on large screens due to less noticeable pixelation and the fact that the soft, clean image should scale better than the blotchy noise posing as detail on Dish. Dish probably looks better than DIRECTV on smaller screens, especially CRTs. Neither will really look good on any normal size TV, but it should at least be somewhat possible to choose a kind of TV that does the least to exaggerate the different picture quality deficiencies these two satellite providers serve up.
Not sure if I have any more comparisons to do after this one. Will have to wait and see.