The first thing to recognize is that TiVo fixed most of the major gripes with their existing units. Their biggest problem is the cable companies themselves vis-a-vis CableLabs, and while I’ll not address anything having to do with them for now (there is a long list of gripes), I had a long list of things TiVo needed to fix in a draft blog post ready to hit the “Publish” button had they messed up. Lucky for them I’m scrapping that post! (well, recycling it into this post, got to be green!)

Upgraded Hardware. While the device is still limited to two tuners (the Moxi supports three, new cable cards will support up to six), the upgraded Broadcom Chip on the inside is a dual core 400MHz MIPS processor and 512MB of RAM with clustered multi-threading (portions of the core like the execution unit are partitioned to support more than one thread per core). So once they manage to optimize their interface they should be able to take advantage of the hardware, even if the 400MHz speed look rather slow.

New HD Interface. The Series 3 TiVo uses the ancient SD interface, while the new Series 4 models use the new Adobe Flash-based UI. While the old interface is leaps and bounds above the standard cable set-top box (STB), other set top box makers (DirecTV, Dish, etc) are quickly catching up, and non-broadcast STBs like the Boxee Box already provide an experience that is better. TiVo should be the far and away leader given the head start they had, but they haven’t kept up. The new UI still needs some (a lot) of polish (“My Shows” should go back to “Now Playing” considering it can contain non-TV show content) but they seem to have got out of the rut they were in.

Better integration with internet content. Whether its the latest episode of The Big Bang Theory or a new Tekzilla I want them all in one list, organized by show name. I want one screen that shows me all the content I can watch now, whether its recorded TV shows, internet TV shows, plus TV shows, movies, pictures and music from my home network. Everything in one place. While I wont be able to get the stuff from my home network, I’m hoping the UI addresses the centralization issue.

Apps. The new TiVo is supposed to have an API available for developers. Combined with the Bluetooth Remote/Keyboard I can see cool Facebook or Twitter notifications. We’ll see if TiVo opens it up to all comers. If so, they are definitely going to need some sort of App Store. It would be really neat though, to replicate some of the iPhone App Store successes on the TiVo.

What did the get wrong?

No DLNA. I wont mince words, this is a huge mistake. TiVo’s proprietary protocols for sharing recorded content needed to be dropped a long time ago in place of the DLNA standard. Part of this might be restrictions imposed on them in terms of getting video out, but at the very least, I should be allowed to stream audio and video into the TiVo from my Windows Home Server easily, and it would be nice if they supported video formats like MKV (MP4+AC3 or DTS) since its really only a container around codecs supported by the Broadcom decoder chip.

No Built-in Bluetooth. While I can understand selling the awesome slider remote for $80, not including the $10 Bluetooth chip inside the unit seems incredibly weak. If I already have a BT keyboard I could do without one in my remote (especially for $80). Allowing BT keyboards in the first place was a great idea, but allow people who already have the hardware to use it! Also, BT would be useful for talking to a TiVo iPhone/Android application to use my phone as an advanced remote control, again, meaning that I don’t need the remote and BT dongle, rather just the BT capability.

While I still think TiVo needs to strengthen their engineering department to make their product better (DLNA, TTG Mac client, etc), the Series 4 is a step in the right direction. Hopefully they can manage to produce a new box more often than every 3 years to keep up with the rate of change in consumer electronics and can manage to squeeze more out of the Series 4 hardware they’re going to start shipping soon.

Finally, one parting thought on comparing a Tivo to an iPhone.

I think its odd that I have no problem dropping $300 every year on an new iPhone plus $30 a month for data and yet still complaining about AT&T’s poor service. But everyone is griping about the TiVo’s price ($300) and monthly costs ($13 or $400 lifetime) and yet they still love their TiVo. It is incredible to me actually. Why does everyone have such a hard time justifying to themselves a $300 TiVo once every three years and the $12.95/mo. I might get more out of an iPhone, but I would presume more people spend more time in front of the TV than a phone (except for teenagers perhaps). The only possible reason I can think of is because the only people I hate more than my cell phone provider is the cable company for its annual price increases. That and it would cost me an extra $10/mo just to them to add a Series 4 TiVo to my house – $2/mo cable card fee PLUS $8/mo for “additional digital service outlet” which is a, pardon my language, bullshit charge hoisted on us by the cable companies and the hardware vendors.

Gear
Lenovo U1 (Engadget)
DLNA Server UI VIew (Engadget)
Ceton 4-tuner Cable Card (We Got Served)

Thoughts

The Lenovo U1 is an incredible product. Its a “regular” 11″ laptop with an Intel Core 2 Duo chip. But you can detach the display and its got a Snapdragon chip inside and it will transfer over the web browser (from Windows 7 to their Linux distro) and back when you undock and then redock the display slate. I’m interested to see how well the transition works in person.

I’m not quite sure where Engadget got its info on the DLNA specs, but if the cable companies are actually enthusiastic about it, it could stand to substantially remake the home entertainment experience with cable. The idea is very similar to the suggestion TiVo made in a FCC comment a few weeks ago. The basic premise is that your ethernet or wifi-connected would use DLNA to connect to the one cable box/omnibox in the house, and then view and even interact with the UI and get the video steaming the data back to the TV.

The Ceton cable card four tuner device is really awesome, but its very expensive – $399. The key to making this device worth while would be to turn the box into a in-house head-end where you can watch a bunch of channels live, or pull stuff up that was recorded, all from a cheap set top box (e.g. not a $200 Xbox, something closer to $129 or less). If Microsoft can pull it off it would be the light fantastic, sadly, I just don’t have a lot of confidence that they can do it. It just doesn’t seem to be in the cards lately for them.

Gear
Tegra 2 (Anandtech) – my predictions were right for the dual-core A9 CPU, though I was expecting a little more graphics performance
Ford Sync 4.0 (Engadget)
Palm Pre/Pixi Pro 16GB for Verizon Jan 25 (Engadget)
Intel Wireless Video – Laptop to TV (Anandtech) - which he was right to call the coolest thing at CES
Alienware M11X Laptop (Engadget) – I give Dell/Alienware huge props for being able to stick a Nvidia GT335M in that small of a case

Thoughts
The Tegra 2 is going to be great, assuming they can land some design wins. It was announced that the Boxee Box is going to be Tegra 2 powered, which is interesting because the Boxee Box will end up having very little power consumption. Its a great fit for cars, portable media devices, slates and tablets. I wish Nvidia well, they could use it right now.

The large number of new LED LCD TVs is great, and the key to the ultra-thin design isn’t thickness, rather the reduction in weight, materials and shipping costs (how many can you fit into a cargo container). What used to be a ton of raw materials, bolts, mounting kits and a 100 lb. LCD TV is now two screws and a 8 lb. LED LCD TV.

The Ford Sync 4.0 in-car system is a mixed bag…

• The Good: Sync API, leveraging existing app stores and not making your own, LCDs in the dashboard area next to the speedometer
• The Bad: Virtualizing just about every control the driver might want to use while driving onto a touch screen without any physical feedback
• The Ugly: Their UI designers, seriously Ford, just copy the iPhone UI design patterns and be done with it

Virtualizing all the controls onto a touch screen is an awful idea – its taking the dynamic user interface several steps too far. If I want to turn up or down the heater or AC in my car, I know exactly where the physical knob is to turn it and what buttons to push. On a touch screen with no physical feedback, I’m stuck glancing over at the screen to see if I’m on the right screen, and if my finger is in the right place, while I can turn the heat up or down in my car without looking and keeping my eyes on the road the entire time. And having to control this through a menu tree through the steering wheel inputs its too much time to change something like climate control (Menu 1 > Menu 2 > Menu 3 > Change temperature).

All I want Ford to do is fix the damn iPhone OS 3.0 issues with the current Sync system – the forever index problem and the auto-switch to bluetooth audio problem.

The reason why this is called day 0 is because traditionally, the day or two before the CES show floor opens (this year, its Thursday, Jan 7th) is filled with private, invite-only press expos that allow manufactures to attract the attention of the press.

Cool Stuff

Syabas PopBox (Engadget)
Boxee Box (Engadget)
Klipsch Light Speaker (Anandtech)

My Thoughts

CES 2010 might be the year of the media set top box. The best looking UI is the popbox, but I’d like to get my hands on both before I make a final purchasing decision. If I can use iTunes to manage all my media metadata (Name, actors, cover art, etc) and then these boxes can display it then that would provide for an excellent user experience.

Big Surprises So Far

Skype on HDTVs. I can definitely see this becoming standard. The only problem is that you have to buy a new TV. Also, PIP calling would be nice (continue to watch football/basketball game with video phone in the upper right corner).

Microsoft’s keynote-fail. I’d have a hard time thinking a 20 minute delay could ever happen to that other big tech company.

The idea lies in TiVo’s request to the FCC asking for a standard to be created for an in-home “gateway” from the cable (or satellite) provider to putting the audiovisual content that the user subscribes to onto their local network.

First is this underlying assumption – that many homes that the use of an antenna to capture OTA signals also has either cable or satellite on at least one TV. What keeps them from hooking up other TVs is the cost – both running the cables necessary and any needed set top box, plus additional fees from the cable or satellite company.

The goal of this solution is to reduce the number of homes in urban areas that need antennas and OTA television reception. As the number of homes in urban areas that use OTA reception shrinks, the cost of upgrading the OTA structure and those needing support in upgrading decreases as well. For rural areas, translators currently provide the necessary broadcast infrastructure and remained unchanged as the main need for additional infrastructure is located in very dense urban areas.

The use of the TiVo suggested “gateway” would allow all of the television and interactive services (on-demand, DVR) offered by your provider to be pushed over the local IP-based network in your home. Using 802.11n, you could push several HDTV streams at 12-15Mb/s (MPEG-4/AVC) over the network to computers or set top boxes for each TV.

The goal of the set top boxes would be to have them cost much cheaper than traditional set top boxes because they would lack the local storage necessary for a DVR (handled by the gateway or even further upstream in the cable plant) and tuners and the associated circuitry to decode analog signals sent over coax. The box would be equipped with antennas for wireless reception, hardwire gigabit ethernet, and HDMI and optical outputs. Inside, a decoder capable of decoding high bitrate MPEG4/AVC and the necessary chips to run the box. Eventually these boxes would be built into the TV.

One necessary feature that would be extremely useful would be the ability to lock out channels and features for the various set top boxes around the home. Because the devices are already hooked up to the in-home IP network through wires or wireless, people would be able to program these devices from their computer or perhaps even their iPhone to control what channels are available and viewable to children. This would be much more consumer friendly than some unfriendly on-screen menu.

Beyond the push to remove as many devices from relying on OTA broadcasts, the next step would be to once again change out the infrastructure that local television stations use to broadcast their equipment. The CEA/CTIA recommended the use of SFN (single frequency network) to allow for more, smaller, lower powered broadcast antennas. Beyond that, it would be advisable to switch from ATSC/8VSB coding using MPEG-2 as the codec to a LTE-based Multicast-Broadcast SFN (MBSFN) with MPEG4-AVC codec in a MPEG-2 TS (transport stream, much like many Blu-Ray discs). Using a 5MHz channel (with 0.5MHz guard bands on each side in the 6MHz channel) for LTE-based broadcasts should allow for 29Mb/s of bandwidth at 2/3 coding rate (1/3 used for forward error correction). In exchange for the hassle of more towers, the broadcasters are rewarded with more bandwidth and a more efficient codec (MPEG-4/AVC) which would allow for more advanced features like 3D HD, more sub-channels, etc. This could also allow broadcasters to use cell phone towers to co-locate their broadcast equipment, reducing total cost and maintenance (subcontracting the tower work out to contractors that manage all the towers in town for all the cell phone companies). The frequency savings would come from removing the restrictions on the adjacent channel rules, allowing TV stations to occupy consecutive 6MHz blocks instead of having empty 6MHz blocks between them.

While this is quite ambitious, and the transition of the existing OTA system is of questionable value, the goal is to free up frequencies for urban usage to alleviate bandwidth problems.

The details of the subscription service will be what makes or breaks the deal for consumers. You can break down the sucess or failure of the subscription service into three parts…

Ability to replace cable. Apple’s subscription service will have a very small niche if they can’t be looked at as a credible way to replace your cable box – those who never wanted cable in the first place and have fast internet connections. This means both having enough content as well as consumer-friendly features (DVR-like services, viewing older shows, etc). Having enough content means all four major networks, plus PBS, along with a few Viacom (MTV and Comedy Central), Discovery (Discovery Channel, TLC), and Scripps (HGTV, Food Network) channels. Also, users wont be very happy if the networks are missing shows – if they see the “ABC” name, they expect all the shows that are shown on ABC. If they don’t get all the shows they wont keep the service for very long.

Price. The supposed $30 price point (plus a up front price for a set top box) could be a huge way to threaten the cable industry. Most cable companies are charging $50-60 per month for 60-70 cable channels, and about $10 per 15-20 channel block after that (digital cable tiers). Apple would have a hard time matching the channel count if they’re going to be paying $2-4 per channel per month, plus pay for bandwidth, storage and overhead – they’re likely to have about 12-15 channels. If they manage to pick the right channels (the popular ones), they’ll stand a better chance to steal a swath of viewers from cable TV.

There are downsides however, the real cost of the plan would probably be $40/mo – the extra $10 comes from the discount most people get by having both cable TV and Internet services. Without cable, the price of internet services usually goes up $10/mo.

Live Events. Live broadcasts like your local newscast, American Idol and the Superbowl are something a lot of people couldn’t go without. Apple has two possible approaches when it comes to these situations – for major events they can stream it live, however that could consume a lot of bandwidth. Second is include an over-the-air tuner in a revised AppleTV device and come up with some sort of nice looking (possibly amplified) antenna for local TV reception.

***

There is significant demand, especially during the downturn, for families to cut costs. Personally, I don’t even watch that much cable TV, what I do watch is available over the air or I already have on DVD, and I certainly wouldn’t mind ditching the $85/mo I pay for cable TV, digital cable, cable card, and two TiVo subscriptions. I’m open to replacing that with a $30 (or even $45 for more channels) subscription plus paying up front for three set top boxes (which Apple does need to bring the price down on – $150 per box max).

(As to why I’m not bringing up a la carte cable as a solution to cheaper cable rates, read this)

In short, there are lots of ways for the studios or Apple to screw it up and kill consumer adoption. Apple has the spine to make sure that the studios do what is necessary to make it a success – but will the studios go along with it…

First, because of the practice of “bundling” by the content creators like Discovery, packaging The Discovery Channel along with TLC and several less popular channels (Military, Science), and offering cable companies a take it or leave it deal. If they don’t accept the less popular channels then they don’t get the very popular channels. Likewise, Disney does this with ESPN and all the ESPN family of networks – you get all of them whether you like it or not, and if you don’t, prepare to face the wrath of your own customers when you drop ESPN as they flee to other providers like DirecTV.

Second is because of the lack of competition and regulation. Disney/ESPN could offer the same deal and conditions to all television distributors and there is nothing anyone can do about it (other than manage to start a highly successful competitor to ESPN and manage to steal their mindshare). Even If the FCC or local regulating authority were to force a la carte on cable and satellite companies, ESPN and other large content creators can tell the cable company that they’re going to pay $2.50 per month per subscriber, regardless of whether or not the subscriber actually chooses to subscribe to ESPN or not. This means that regardless of what subscribers choose, on average they’ll end up paying the same price for less content.

Back in the late summer, we a few companies floating their digital-magazine-on-a-tablet format. The goal appears to be getting out in front of Apple, who has been rumored to be developing a tablet for many years now (basically all the way back to 2002, with the project being killed many times). Though, I would guess Apple already has a format written down someone, but they aren’t sharing it yet.

Next, we get a host of other companies attempting to develop sleek, stylish tablets. Nvidia showed a Tegra-based tablet that looked like a mock-up of something Apple would have come up with (a large screen and one “home” button). Other tablets (Notion Ink Smartpad, etc) are set to be unveiled and showed off at CES in early January.

Do I think Apple has anything to worry about on the tablet front? Not really – they’ve shown they can out innovate just about anyone. They’ve also got the tremendous mindshare of the iPod and iPhone. Finally is the App Store – getting developers to do some (hopefully) minor work to port any apps over to the tablet means they’ve got a huge lead on anyone else who tries to enter the market. An Intel Atom-based tablet is a pipe dream for the time being – the power consumption is too high and Windows 7 (required for multitouch support) will push the bill of materials up (faster Atom, 2GB of RAM, etc) in order to provide a good experience.

It wont be as smooth as a ride Apple had with the iPhone – it took three years from the initial announcement for someone to finally come out with a device that even came close to the iPhone (Droid). And Apple might get off to a slow start – they like to grow their products organically. The iPhone initially didn’t have App support, it even took many months for them to support saving web pages to icons on the home screen. But over the course of two years, the product grew up – gaining App support, MMS, Cut Copy Paste, landscape apps (Mail, Messages), video, etc., to finally a product that is mature and can start to explore areas that cell phones and even smart phones have never tread before (payment processing, and who knows what else).

The Improvements

The goal of Handbrake is to encode videos into the H.264 codec using the x264 engine, so we can start with those improvements. These improvements are done by a different group of developers (the x264 team) and the HB developers leverage x264 to create Handbrake. The first big boost is the instructions optimized for the Intel Nehalem architecture – up to 40% over a similarly clocked Penryn, and further improvements that have been subsequently committed to the codebase. Unfortunately I don’t have a Nehalem-based CPU to test them on (I’m waiting for motherboards that will have USB 3.0). Other non-performance code improvements have been made as well – one improvement is known as MB-Tree (macroblock tree). This optimization will track the progression of data from the current frame and how its used in future frames, so that data in the current frame that is wiped off screen very soon doesn’t get a whole lot of bits spent on it, and data that will be persisted for many frames to come will get more bits invested. Likewise weighted-prediction will increase the compression and quality of the video, but require extra processing and time.

The Handbrake software also has numerous improvements, most notably the addition of soft subtitles to allow people to turn on/off subtitles on their video instead of having them burned into the video frame. DVDs typically come with closed captioned text available, and you can choose to encode those into the video and turn them on and off through the video playback software if supported – VLC, iTunes, QuickTime X and the iPhone/iPod Touch all support it. I tested it on some raw high-def content (MPEG2 in a m2ts container and H.264 AVC in an m2ts container), it was able to handle converting it to a MP4 container (maintaining the 1080 resolution) as well as scaling it down to DVD resolution and iPhone resolution. Lastly, the video preview option will have HandBrake create a 5-60 second clip from anywhere in the video so you can check for things like proper deinterlacing as well as cropping without having to wait for the entire encode to finish (or just picking out one chapter).

Whats Been Cut

Much like the few months before your high school reunion, Handbrake developers cut the fat from the software by removing AVI container and Xvid and OGM codec support. As Handbrake is primarily for encoding video to H.264 in an MPEG-4 container, its appropriate to keep the focus on those aspects. It can still encode to MKV and supports features through MKV that aren’t supported using the MP4 container format (DTS audio passthrough).

Support was also removed for CQP (constant quantization parameter) encoding in lieu of CRF (constant rate factor). I could retype this entire page in a paragraph, but just go read it and understand that the HB developers made a sound decision. In one sentence, its another human perception vs computer analysis question that came down on the side of human perception. I would have liked to see some sort of feedback indicator on the CRF setting for novices – red for either too low or too high, yellow for the boundary area, and green for good RF values (18-22) – it was discussed at some point but not yet implemented.

What Isn’t Included

One of the top hits to my website from Google searches has to do with Handbrake and OpenCL or Grand Central technologies for the Mac. First, its inappropriate to even lay this at the feet of the Handbrake developer team because they aren’t in a position to implement them. As I mentioned above, the x264 team would be the ones responsible for implementing those technologies, and they haven’t shown much interest – they did post a Google Summer of Code opening for someone to write a Motion Estimator in CUDA (which would be restricted to Nvidia cards). The SoC is over now, and it doesn’t appear anything came of it so we’ll continue to wait. I believe it will eventually come, but it might not be for another year or two as the features of the underlying GPU hardware and languages like OpenCL and DirectCompute mature.

Another item not included is the support of WMA Pro as an input audio codec. The source library (ffmpeg) that would provide the decoding for WMA Pro had the feature added too late for the update to be included in 0.9.4. This means decoding of WMA Pro audio tracks wont be included in Handbrake until 0.9.5 (hopefully sooner than November 2010). Likewise, multi-threaded decoding is also not included (ffmpeg-mt). This isn’t quite needed yet, since the ffmpeg can currently decode input video faster in one thread than x264 can encode video in 8 threads, but as CPUs move towards 6 and 8 cores per CPU plus hyper-threading (12-16 threads) the need for multi-threaded decoding will increase – I would expect that by the time the next release hits sometime in 2010, ffmpeg-mt will be standard.

Benchmarks

Normal or iPhone presets at CRF: 20 (source: Juno DVD, all chapters; 1080i source from TiVo HD STB w/ deinterlace-slow)

Handbrake 0.9.4 Benchmark

I tested the 0.9.3 release against the 0.9.4 32- and 64-bit releases on a 4-core 2.66Ghz Penryn-based OSX computer. The source material for the iPhone and Normal tests were the Juno DVD (copied to HDD first), and the 1080i-source was an episode of The Big Bang Theory from my TiVo HD with the commercials removed. On the 1080i-source, the material was de-interlaced during the encode. You can see that the 64-bit version has better performance that the 32-bit version. The caveat is that the 64-bit version of VLC is (at the time of this post) not very stable. So if you do use the 64b version of HB, I’d recommend you pick another piece of software to remove the DVD encryption. The quality of the output was comparable for each version. The file size for the iPhone version was surprisingly larger (521MB vs 440MB), but the standard resolution version was smaller (929MB vs 1.1GB) and the 1080i version was much smaller in 0.9.4 (1.1GB vs 1.6GB).

There could be even larger gains for those who have Intel Core i5/i7-based CPUs, as x264 can take advantage of new SSE instructions in those CPUs – benchmarks from around the web indicate an additional boost of 10-25%. For those with the advanced chips, moving from 0.9.3 to 0.9.4 64-bit could result in performance increases between 40-50%.

Handbrake 0.9.4 is a much needed upgrade – the new features can really put a polish on the video for those who are in to ripping all their media to a home server, plus speed improvement from the x264 library will cut 20-30% or more off the time to encode video, which can be a while when you’re encoding in high definition. Get it, and if you can, get the Intel-64 bit version!

1. Increased competition (iPhone, Zune platforms) from non-traditional areas. I can play Rock Band on my Xbox 360 or my iPhone. Granted the experience is nowhere near the same, but its still something to worry about in terms of how gamers will allocate their time playing games – there are more platforms and more games being made but still only 24 hours in a day.

2. TV’s are stuck at 1080p/60Hz, while something like the Radeon 5000-series can draw 60fps at 1920×1080 in most games, let alone what will be available around the end of 2011 replacement time frame (able to render 3072×1920 at 60fps). CPUs will become more advanced but are likely to stay RISC (IBM’s PPC chips) and clock speed will go up. Until TVs move to 4K (4096×2560) and possibly faster frame rates available at input (120Hz) then the video output of the consoles is limited to current HD resolution.

3. Increased focus on margins earlier in the product cycle – Sony and MS wont want to sell the consoles for huge losses (e.g. PS3 initially cost about $800 and sold for $500-600). I would expect launch consoles to sell for $400-500 and be down to $250-300 within 18-24 months (one internal hardware refresh). The lack of cutting edge hardware might help in keeping prices down. What a $100 GPU can do in 2011 might be sufficient for the console, instead of a $200 GPU. And the same goes for the CPU. There is also something to be said about going smaller and more efficient in terms of power usage. Larger storage amounts should be trivial (2.5″ HDDs are now very cheap) and putting 2GB of storage on the device for an OS should be sufficient.

4. Advanced third-party fabs. Global Foundries will have bulk 28nm fabrication processes online in mid-2011 and could play a big role in not only producing the GPU for one or more of the consoles. There is even an outside chance of seeing Nintendo try and put the CPU and GPU on the same piece of silicon – something along the lines of the AMD Fusion chip but with a PPC or ARM-based core and a GPU that is capable of 720p/60 fps. This might trigger licensing issues going down the PPC road but that would ensure backwards compatibility.

5. The big feature will be physical interaction like we saw for the Wii. While this is novel, its not too terribly expensive (see the Wii). Compare this to the initial cost of Bluray, backwards compatibility and the Cell CPU for Sony, and HD graphics for both consoles. MS didn’t have any other high-cost features but the high cost of poor engineering and the Red Ring of Death means they’re eating costs they could have spent on making the consoles cheaper quicker. They might be more inclined to be a bit more conservative on power usage and heat and make sure the console is reliable.