In order of preference:

1. If the content you wish to download is film-based, I’d take the sat or cable provider’s 1080p24 feed miles ahead of a 1080i60 broadcast. In that case it’s not *just* marketing. Since their source D5 or HDCAM tape is most assuredly a 1080p23.98 tape, I’d like fewer conversions along the way to my display if possible. It’ll likely already be compressed to within an inch of its life, I’d like to not deal with 2:3 pulldown along the way if possible.

2. If the content was shot at 1080p30 (probably unlikely) and is available for 1080p download, then 1080p is the preferred format. However, interlacing 1080p30 to 1080i60 for broadcast and then our devices deinterlacing it *back* to 1080p30 should be seamless by most processors, but conversions are fallible. Having said that, in this scenario I’m less-inclined to purchase or download one format over the other. While I trust a broadcaster’s deinterlacer more than I do mine (Pixelworks & HQV), if I’m considering VOD content in the first place, then PQ isn’t my #1 priority - convenience is.

3. If the content was shot at 1080i60 and is available for 1080p download, I’m less-inclined to purchase or download one over the other. Given the choice of getting my content deinterlaced by a professional piece of hardware versus trusting consumer-grade hardware to do it just as well, I’ll take the pro everytime in a technical discussion, but for the reasons stated in the last paragraph - it’s about convenience.

My 2 cents is that it’s not *just* marketing. There are techincal reasons why 1080p is preferrable to 1080i in two out of three of the cases I mentioned, but as many of us have also mentioned, for over-compressed, VOD services, they may be made irrelevant.

First, the only place you’ll find a completely uncompressed HD stream is in production. You’ll never find it hitting a consumer source or display.

Second, further to the first point, compression has tremendous impact on final quality which is why “HD” downloads or “HD” cable is notably inferior to HD physical media. Compressed bitrates of DLC is typically less than half (7-10 Mb/s) what you would find on an average Blu-ray (18-24 Mb/s). Even considering the additional bandwidth of high-res audio, the bitrate differences for video are consequential.

Lastly (unaccounted for thought), there are definitely instances where convenience overrides quality but the reverse can also be true. Wedding Crashers in high bitrate 1080p has less allure than Baraka for instance.

The bandwidth savings is pretty simple: in a completely uncompressed stream:
720P = 1280*720*60 frames a second ~ 55.3 mbits/sec
1080i = 1920*540*60 (or 1920*1080 * 30) ~ 62.2 mbits/sec
1080p/24 = 1920*1080*24 ~49.8 m/bits or 20% less than 62.2m/bits for 1080i if you assume the compressed stream keeps a similar relationship to the raw.

*To simplify the explanation I didn’t include bits for color depth, if you want real “raw” bits rates multiple the above by 8.

Second thought is this is another example of how downloading or near timing content is surplanting the need for physical media like bluray. Yes Bluray can have a better picture due to it’s higher bit rate capability but: 95%+ of folks don’t care and convienience * cost = King. (Although if you are reading this you uare probably of the 5% that do care…). If I can get a great looking movie for $6 and it’s there when I want it…. Is it worth the trip to the store? Is it worth ordering and waiting?

The “big deal” about having an integer-multiple of 24Hz is to eliminate “judder” in the picture. Judder is a timing disconnect between the source and display framerates, and it shows up as “jerky” motion. Televisions touting 120Hz modes do so because it is the lowest framerate that is both an integer multiple of 24 and 60Hz (we will ignore fractions here, like 59.94…).

“And a question for Bill: Where and/or how does the 20% bandwidth savings come in for a video provider if they switch to transmitting 1080/24p material via 1080i to transmitting it natively? I thought the two were virtually the same bandwidth. I can see their processing overhead going down, but don’t see the bandwidth savings.”

The difference in bandwidth would come from eliminating the redundant information in the encoded video stream. The reason this might happen is due to potential differences in how a broadcaster would encode (really: transcode) its content. Many broadcast encoders in the past have been single-pass, constant bit-rate (CBR) designs emphasizing reliability and speed over absolute image quality and efficiency. Conversely, packaged media have used multi-pass variable bit rate (VBR) encoders for years. The difference between these is that a VBR approach, which requires more than one pass through the data, is a LOT more efficient (better quality output) at a given bitrate than a comparable single-pass encoder. Even where you have a relatively efficient encoder, broadcasters often have to pad data with nulls in order to make a data stream conform to meet specifications expected by some piece of gear in the chain (e.g., if you strip the nulls out of an ATSC stream, you can often save half of the resulting file size — at least with some of my local broadcasters!).

So, where did the fairly specific 20% number come from? It came from what ought to be a “best case” scenario for bandwidth savings from DirecTV or Dish. Simplistically, a 24p data stream is 20% smaller than a 30i data stream. However, whether any of this is fully realized has to do with how the data is handled (or mangled) during transmission, storage and playback. Hence my proposed experiment. If a 1080i VOD movie underwent a clean telecine process that is preserved through to the display, then one ought to be able to tell by forcing a “film” deinterlacing mode. If the movie has deinterlacing issues (e.g., loses sync, breaks up, etc.), then there is a pretty good chance that the provider’s “standard” delivery process for VOD has to force a conversion to “video” somewhere along the way, and in that case, a 24p delivery format more than likely saves them bandwidth.