I received a press release email last night from MatrixStream, a company that specializes in IPTV solutions and located just a few blocks away from my home (I might pay them a visit to do a feature article on them). Maxistream has developed streaming servers, h.264 real-time encoders, client 1080p hardware, software clients — the real IPTV deal. I happened to think of HD IPTV quite a bit in the last few weeks. I think that’s one of the sectors where business will be growing a lot in the next 5 years or so.
In association with MaxiStream, a new company called myTVPal was formed and went public this week. It currently runs a free trial that let’s you access 700 TV channels from around the world. About 20 of these channels are HD (720p, 1080p) or EDTV (480p). You have to give them your personal details to download their player and login with it (a bit extreme IMO). Anyways, I did so, and here is what I found out so far:
Their h.264 encoding is aggressive on their HD channels. It feels like a VGA video resized to fit the screen (along with the visual artifacts that comes with such resizing) rather than real HD. However, as a tech person I understand their technical limitations (bandwidth costs money). The good thing is that playback starts almost immediately, it’s very impressive to see buffering going so fast for HD content! What I really didn’t like though is that their HD channels are just not smooth on their software player. I tried their player on my hyperthreaded P4 3.0 GHz with 1 GB of RAM and a 800 KB/sec Comcast cable connection. There are dropped frames with 720p playback and almost no frames rendering with 1080p content. The dropped frames don’t even make sense to be dropped because my PC does not max out its two CPUs (it goes up to 44% of CPU usage on 720p) and there is no bandwidth problem on my side. The dropped frames look more like a bug on the software player rather than out-of-processing-power or bandwidth problem. The problem does not exist on EDTV channels though, where the CPU uses just 15%. For 1080p content I get 1 frame per 20 seconds or so (much slower than how QuickTime performs), and yet, my CPUs never go above 60% of usage. This is something that Maxistream and MyPalmTV must fix on their software player before they go live with this whole idea.
Anyways, the point is that this can be done right. The technology exists and it’s called h.264. Personally, I don’t need 720p or 1080p resolutions, which take away so much CPU power. What I want instead is IPTV channels with 480p EDTV/HD resolution (848×480) that have a less aggressive encoding. A properly encoded EDTV signal is way better than a poorly encoded HD signal, and while the bandwidth consumption might be similar at the end, CPU on the client machine won’t go off the roof. And this has value for the customer who might want to browse the internet or reply to some email on his second monitor while watching IPTV on the background of his first monitor (that would be my user case in fact).
If I had 200 million US dollars (minimum required) I would create a new studio production company in Tijuana, Mexico — just a 2.5 hours drive south to LA. I would employ Mexicans to handle most of the technical aspects of the work and talented but unknown (mostly) American actors who would commute a few times per month between LA and Tijuana to shoot original, high-quality IPTV series. To shoot a single episode today of a good TV show you need between 800,000 and 3 million US dollars. In Mexico, you can do well with $300,000-500,000 per episode if your actors are not demanding and you manage to employ talented and creative people. Monday to Thursday are the big days regarding TV viewing, and primetime is between 8 and 11 PM. This means 4 days with 3 shows per day equal 12 original TV shows per week which would collectively cost about $5 million altogether (and maybe two half-hour sitcoms for Sunday at 8 PM). Commercials would be served to the viewers like on normal TV, but with the ability to serve different ones depending on the IP address of the viewer (with clickable links for each advertiser’s site).
Then, distribution would be via the internet on a variety of devices and resolutions. The company must develop and offer their content any way they can, and this means standard h.264 streaming clients (preferably open source and QuickTime should also work) for:
EDTV 848×480: Windows 9x/XP/Vista, OSX, Linux x86/PPC, FreeBSD, Sony PS3, XBox360, Wii.
480×272: Sony PSP (a deal with Sony via firmware upgrade), Archos 604 Wifi (via firmware upgrade) and desktops with lesser bandwidth or bigger CPU constraints.
320×240: all 4 Symbian UI versions, WinMobile, PalmOS, Linux/S60/Blackberry phones with Real Player streaming feed, QVGA cellphones that can stream MP4.
176×144: Cellphones with 176×220 resolution and MP4 streaming capability. Maybe a Java client too for best cellphone compatibility.
Web Client: 320×240 Flash player, viewable on the studio’s website front page (auto-connect and playback to try to hook viewers immediately upon visiting the site for info).
The 3 original shows of the day must rerun 3 times a day. Once when it’s primetime for Japan/Australia, second time when it’s primetime for mid-Europe & South Africa and once for Canada & USA’s Mid-west. This way, you capture most of the Internet (tech-oriented) users who actually have the bandwidth and the means to view your content. Each primetime timeslot holds two shows per year, with 18 episodes each. That’s 18+18=36 weeks, plus 12 weeks for the summer that it will mostly be hosting reruns, plus 4 holiday weeks with reruns (these weeks are low-traffic so it doesn’t make sense to broadcast new episodes). Overall, that’s 18+18+12+4=52 weeks.
The rest of a typical broadcasting day you can have tech-related shows (e.g. hardware, game reviews), live magazines, scientific documentaries licensed from Discovery or Science Channel, some global news twice a day and older classic TV shows that come for cheap to fill up the schedule.
Now that I’ve said all that, let me also note that I don’t have $200 million for such a startup. Some ideas require money before you can realize them and no venture capital has so much money to give away — even if they might wanted to.