CHAPTER 23
Optical Disc

DVD, Blu-Ray, and ROM

Introduction

This chapter is about video delivered on optical discs, particularly DVD and Blu-ray disc (BD), and also general principles that apply to Video CD, CD- and DVD-ROM, and other optical media–based usage.

While discs may seem somewhat atavistic in the Internet era, they’re still a huge portion of content delivery. And Hollywood studios sequence their releases, so there’s a particular “DVD window” (which includes Blu-ray) that comes directly after theatrical release but before content is made available for video-on-demand (VOD)/pay-per-view (PPV), then premium cable, and lastly broadcast.

Since DVD makes up such a huge portion of Hollywood revenue (although it has shrunk some in recent years), it gets a lot of focus from the studios, who make a lot more per disc than they do per viewing in the VOD/PPV or later windows.

The combination of big markets, and hence big budgets, plus the interoperable interactivity on DVD and Blu-ray means that disc get lion’s share of complex authoring. So discs are where we see scene-by-scene compression tuning, director’s commentaries, multiple languages for audio and captions, nicely designed menus, and a panoply of extras.

From a compression perspective, DVD and Blu-ray have very precise specs, which makes things a lot more straightfoward than web encoding. However, they also have high expectations, which can make those limitations chafe in some cases.

I hope we’ll see the “DVD window” evolve into more of a “premium window” so interactivity (and high budgets!) can have their bits liberated from their prison of spinning polycarbonate. This is a project I’ve been personally working on for a while now, and 2010 should see a number of downloadable movies with all the features of the DVD version.

But for the time being, DVD and Blu-ray are where we see premium interactive design and innovations, and the budgets that can afford those efforts.

Characteristics of Disc Playback

The physical differences between Internet delivery and discs have an impact on how we compress. With discs, we typically get a lot of (but not limitless) capacity with quite high peak bitrates. However, due to the relatively slow rotational speed of optical discs, GOPS and buffers are smaller than we use for network-distributed media in order to deliver decent random access. And with discs, we have a precise maximum number of bits we can use, without the “maybe we can get away with 1500 Kbps” gambles of web video.

The combination of fixed maximum capacity and fixed maximum bitrate means that there are two basic strategies for rate control on discs, depending on capacity of the disc and duration of the content:

However, particularly with DVD, there’s a big difference between discs that are stamped in a factory (called “replicated”) and those burned one-at-a-time in a drive (called “duplicated”), forcing a tradeoff between breadth of compatibility and peak bitrate. I’ll talk later in this chapter about how that works in practice.

The interactive layers of DVD and Blu-ray are well specified, which make them more feasible to use than typical PC media players, although still behind Flash and Silverlight. Today, DVD’s interactive technology may seem nearly as primitive as an abacus, but a decade of refinement of authoring technologies targeting that allow for quite rich experiences.

Blu-ray has very deep interactivity support via Java (BD-J), although we still haven’t seen easy-to-use interactive authoring tools for BD-J like we have for DVD.

DVD

DVDs were probably the first place most people saw digital video without palpable artifacts. It was definitely a quality jump over broadcast, cable, and particularly VHS of its era. And despite the predictions of some film buffs, it could look a lot better than the Laserdisc. Laserdisc was the best ever composite consumer format, but DVD was component, could do higher bitrate 5.1 sound, and had much better interactivity.

DVD Tech Specs

DVDs can be either single- or dual-sided, and each side can be either one or two layers. The second layer has somewhat lower capacity than the first.

Most DVDs are the classic 12 cm diameter, like CD and Blu-ray. A smaller 8 cm size is also available. However, these small discs don’t work reliably in disc-load drives like those in laptops. While cute, they’re no cheaper to make than 12 cm discs, and I recommend against them for compatibility reasons. Table 23.1 shows the various DVD types and their capacities.

Table 23.1 DVD Types and Capacity.

Name	Sides	Layers	Diameter	Capacity
DVD-1	1	1	8 cm	1.46 GB
DVD-2	1	2	8 cm	2.66 GB
DVD-3	2	1/1	8 cm	2.92 GB
DVD-4	2	2/2	8 cm	5.32 GB
DVD-5	1	1	12 cm	4.70 GB
DVD-9	1	2	12 cm	8.54 GB
DVD-10	2	1/1	12 cm	9.40 GB
DVD-14	2	2/1	12 cm	13.24 GB
DVD-18	2	2/2	12 cm	17.08 GB

DVD introduced the UDF (Universal Disc Format) file system, an updated version of which is used with Blu-ray. UDF is very handy for ROM discs as well, as it is well-supported by all operating systems.

One nice physical property of DVD is that each side is symmetrical, with the reflective surface in the center of the disc. This contrasts with CD, where the reflective surface was at the bottom of the plastic, next to the label, and Blu-ray, where the reflective surface is under just 0.1 mm of protective coating. This makes DVDs easier to manufacture and quicker to cool, allowing more discs an hour through a machine and a cheaper per-disc price than either CD or BD.

MPEG-2 for DVD

DVD video uses a tight subset of MPEG-2, with each disc being either NTSC or PAL. DVD is very capable of producing the highest quality of any SD consumer video format. The spec maximum bitrate is 9.8 Mbps for video and audio and 10 Mbps total for audio, video, and subtitles. In practice, experts recommend using a peak of at most 9.6 Mbps for replicated discs. DVD supports VBR, and high-end discs almost always use 2-pass VBR to produce optimal quality within the disc capacity.

While the encoded video on nearly all DVDs is 720x width, other modes are supported for 4:3 content (720x is required for 16:9). MPEG-1 bitstreams encoded to Video CD spec are also supported for compatibility with existing content. But even if 320 × 240/288 encoding is appropriate (like trying to cram six hours of talking-head video on a disc), MPEG-2’s greater efficiency and support for VBR make that a better choice. Legal DVD encoding modes are shown in Table 23.2.

Table 23.2 Legal DVD Encoding Modes.

Format	Standard	Width	Height	Aspect Ratio
MPEG-1	NTSC	352	240	4:3
MPEG-2	NTSC	720	480	4:3 or 16:9
MPEG-2	NTSC	704	480	4:3
MPEG-2	NTSC	352	480	4:3
MPEG-2	NTSC	352	240	4:3
MPEG-1	PAL	352	288	4:3
MPEG-2	PAL	720	576	4:3 or 16:9
MPEG-2	PAL	704	576	4:3
MPEG-2	PAL	352	576	4:3
MPEG-2	PAL	352	288	4:3

Many encoders for DVD only support the 720x modes.

Quality expectations for DVD are quite high, especially compared to other MPEG-2 applications. Typically, any obvious blocking or ringing artifacts are considered unacceptable. “Good enough” data rates with professional encoders for 24p/25p sources are around 4–5 Mbps with peaks up around 8 Mbps. MPEG-2 VBR encoders may provide a minimum bitrate, used to make sure that the VBR algorithm doesn’t strip easy scenes of too many bits. Interlaced DVDs can be either top or bottom field first. I recommend setting the output field order to the same as the source to avoid unnecessary conversions. Interlaced content requires roughly 20 percent higher bitrate than progressive.

The maximum GOP length of DVD is 18 frames for NTSC and 15 for PAL, with the defaults being 15 and 12 respectively. This makes for quick random access, but a risk of rapid keyframe strobing at lower bitrates.

Aspect Ratio

MPEG-2 for DVD supports both 4:3 and 16:9, and the player will convert as needed for the attached display (assuming correct configuration).

16:9 is still a taller ratio than most films. Because of this, even 16:9-encoded MPEG-2 for DVDs will have some letterboxing, although much less than the same content in 4:3. 16:9 can use 33 percent more active pixels than 4:3 with wide-screen content—especially important when upscaled on computers or HD displays.

You can gain a slight increase in compression efficiency by aligning the letterbox edges with macroblock or block boundaries. The very sharp edge between the letterbox and the film content can cause noticeable ringing; MPEG-2 is more prone to this than more recent codecs. See Table 6.1 for some typical combinations.

Progressive DVD

It is a lingering tragedy of the age that MPEG-2 on DVD doesn’t support MPEG-2’s progressive sequence mode; only interlaced sequences are used, so there’s no way to do “real” 24p encoding. However, interlaced sequences do support both progressive and interlaced frames. For NTSC discs, this can be used to encode the 3:2 telecine pattern in a way that’s easy for players to reverse (see page 115 for a more thorough discussion of 3:2 pulldown and inverse telecine).

24p DVD uses a MPEG-2 sequence with repeat_field tags. These let the encoder not encode the duplicate field in the “3” phase of “3:2 pulldown.” This doesn’t need to be handled manually—the encoder itself should correctly tag the output if the source is 24p.

Then, on playback a “progressive” DVD player will see the pattern of repeat_field tags and reassemble the original 24p content.

PAL compressionists have it easier; 24p source is sped up to 25p and encoded as a series of progressive frames. Furthermore, they get 20 percent more pixels a frame (576 instead of 480).

NTSC 30p content can also be encoded as 30p easily, by just encoding it as a series of progressive frames. This is often a better choice for 60p sources without a lot of high motion (25p can be better for 50i for the same reason). For sports, going from 60p or 50p to 30i or 25 can deliver much smoother motion, but it always feels shameful and cruel to rend innocent progressive frames into interlaced fields.

With interlaced displays are on their way out, progressive discs look a lot better than interlaced on a progressive screen, particularly for PC or HD playback.

Multi-Angle DVD

One much-hyped feature of DVD is its ability to present multiple camera angles that can be selected and viewed on the fly. It sounds cooler in theory than it is in practice. Multi-angle is complex to author, and even when done perfectly, the speed of angle transitions can vary quite a bit depending on the player. To have seamless multi-angle switching, the GOPs in each stream must align perfectly. So you’ll normally need to compress with:

Unlike multiple audio streams, which come right out of video top bandwidth, each angle gets its own bandwidth. In theory, each can be VBR, but that makes quite complicated; many products require multi-angle tracks to be CBR. Needless to say, five 8 Mbps CBR angles eat up disc capacity very quickly.

Multi-angle is an interesting technology, but not much used outside of the adult industry (I’ll leave that proof as an exercise for the reader). Thus it’s not supported by many tools outside of the high end.

Multi-angle is one of those finicky things you want to have successfully authored before you commit to a project using it.

DVD Audio

DVD’s practical mux limit is 9.6 Mbps is for video plus all audio tracks. All DVD audio codecs are CBR, so every bit allocated to audio comes out of your video peak bitrate. You therefore need to be judicious in how many audio tracks you include and how many bits to give them, particularly with harder-to-encode interlaced video.

The PCM, Dolby Digital, and DTS codecs are covered in more detail along with MPEG-2 on page 173.

PCM (mandatory)

PCM audio is mandatory on DVD players, but I strongly recommend against it. Even something as simple as stereo audio eats up 1.5 Mbps of your precious bandwidth. Dolby Digital encodes much more efficiently.

Historically, PCM audio was mainly seen in consumer-authored discs created with tools that didn’t license AC-3 encoders from Dolby. However, any DVD authoring tool worth using today includes at least stereo AC-3.

PCM support does offer some impressive numbers, but they’re largely specsmanship for anything outside of the rare audio-only “Audio DVD” format, particularly 96 KHz. Supported options are:

Dolby Digital (mandatory)

Dolby Digital (AC-3) is really the default codec for DVD; most of the discs in the world use AC-3 and nothing else.

DVD has a pretty typical implementation, supporting up to 448 Kbps (less than the codec’s 640 Kbps max as supported on Blu-ray, but an upgrade from Laserdisc’s 384 Kbps). Good defaults are 224 Kbps for stereo and 384 Kbps or 448 Kbps for 5.1 mixes. If there’s only a single track and the project is audiocentric, you can max out stereo at 448 Kbps; that’s still more than 1 Mbps saved versus PCM.

But even with AC-3’s better compression, be careful when providing a lot of audio tracks. One 448 Kbps track isn’t a problem, but add English, Spanish, and French at 448 Kbps plus a commentary track at 224 Kbps and your peak bitrate is 1568 Kbps lower.

I normally only encode the original language at 448 Kbps. Since dubbed audio generally sounds lousy anyway, I’ll reduce bitrate on those if needed to keep video quality up; audiophiles will be listening to the original language track anyway. And a commentary track can go down to 128–196 Kbps as long as it’s mainly speech and doesn’t mix in too much of the soundtrack.

Note that DVD only plays back one audio track at a time. If you want to present commentary with the soundtrack in the background, you’ll need to mix all of that together before compression.

DTS (optional)

The DTS audio codec is optional on DVD. This means any DVD using DTS also needs to include PCM or AC-3.

While DTS, particularly in its higher 1536 Kbps mode, can be more transparent than AC-3 at 448 Kbps, that’s a big price in bandwidth budget, particularly with an AC-3 fallback stream still required. Most DTS on DVD uses the lower 768 Kbps mode, which isn’t consistently better than 448 Kbps AC-3.

Due to DTS being an optional codec, for most content the reduction in video bitrate and capacity isn’t worth the theoretical advantages. One obvious exception would be musiccentric content, like concert videos. In those cases, a premium audio experience at the top DTS data rate may be well worth it.

MPEG-2 Layer II (mythical)

MPEG-2 Layer II is often listed at a DVD audio codec, but it’s almost never seen in the wild, and isn’t mandatory in NTSC players. AC-3 is more efficient than Layer II, and Layer II isn’t universally supported by receivers, so there’s no reason to use it.

DVD Interactivity

That DVDs are as interactive as they are is a testament to more than a decade of hard work on the part of authoring tool vendors much more so than to the underlying technology, of which there really isn’t much. I doubt you can buy a wristwatch that doesn’t have more processing power than the first DVD players.

In the end, DVD can play a single video and a single audio track at a time. Subtitles are just 4-bit indexed color TIFF files, as are most other overlay graphics. Total available memory for programming is 256 bits. Not Mbits. Not Kbits. Just bits: a set of sixteen 16-bit numbers. And programming them can feel like using an abacus wearing mittens.

But wow, some impressive experiences have been built on top of that simple base. A key innovation was the use of video for menus, so the very simple 4-bit TIFF overlays aren’t needed.

DVD Mastering

VOB files

DVD video doesn’t actually use MPEG-2 files, but incorporates the MPEG-2 video and audio information into Video Object (VOB) files, multiplexing (muxing) the video, audio, and subtitles together. These individually are limited to 1 GB in size, so if you look at the disc’s file structure, you’ll see a series of these for movie-length content.

Replication

A replicated disc is mass-produced from a “glass master,” as opposed to the one-at-a-time discs burned in duplication. The replicator creates the glass master from a provided disc image. Traditionally, this was provided on a DLT tape, but most replicators now also support DVD-R masters.

DVD replication has become dirt-cheap and broadly available with the maturity of the technology. Short runs of 300–500 discs can be purchased for under $1,000, with the cost per disc dropping as volume goes up (that’s just for the disc, without printing or packaging).

The replicator is responsible for collecting a $0.04 MPEG-2 license fee per disc, paid to MPEG-LA.

While dual-layer replication was initially finicky and expensive, it’s now only a slight premium over single-layer, and actually cheaper per GB.

Beyond being cheaper as discs runs get into the thousands, replicated discs are also more reliable, particularly in older players, and more particularly with higher peak mux rates and dual layers.

Duplication

Duplication ranges from a laptop burn of a single copy of a disc to professional-grade automated burning of hundreds of discs by a replication company.

For the usual political–technical reasons, there’s two camps of writable DVD discs: the “dash” family originated by Pioneer and the official writable format of the DVD Association (call it a “minus” at your peril!) and the “plus” family later introduced by Philips and Sony.

There are those in the industry who still take this split very personally, but there are pragmatic reasons to pick a particular format for a particular application. The good news is that PC DVD-ROM drives of today support reading and writing all the formats well. It’s dedicated DVD video players, particularly older ones, where compatibility concerns creep in.

DVD-R

DVD-R was the first writable disc format. The original version was called “DVD-R for Authoring” and clocked in at only 3.95 GB. I worked with Intel on the first DVD encoded on a PC instead of the refrigerator-sized real-time encoders of the day. We burned test discs on one of the first three DVD-R drives in the United States. The blanks were $40 each, took several hours to burn and verify, and failed about half the time. When burning a disc we’d have everyone clear off that floor of the building to minimize any vibrations that might increase the risk of a bad burn.

Fortunately, this was soon replaced with today’s “DVD-R for General,” which everyone calls just DVD-R today. With further fortune, it has the same 4.7 GB capacity of a DVD-5.

DVD-R is the most reliable burnable disc for DVD players, particularly older ones that predate the later writable formats. Even so, some early players have trouble with higher bitrates and Open GOP; my first-generation Sony 700 DVD player only liked Mitsui Gold DVD-R discs, and would spit out anything else. If content can handle it, I generally recommend a maximum mux rate of 6.5 Mbps with DVD-R discs, as this improves reliability.

There is a dual-layer variant called DVD-R DL. It is much less compatible with older players. Quite a bit more expensive, and burns much slower than single-layer. I stay away from it for DVD unless it’s really the only way to get the movie on the disc, and if I think the audience is mainly going to have newer players or be using PCs.

DVD-RW

DVD-RW is the rewritable version of DVD-R. Thus, you can add to or rewrite the disc after an initial write. This is handy for PC use, but doesn’t offer much for DVD authoring, since the disc is mastered in one pass anyway. Given how cheap DVD-R discs are, and that DVD-R burns at a higher speed, DVD-RW doesn’t have a clear niche for DVD video.

DVD+R

DVD+R was only introduced in 2002, and was more focused on replacing CD-R discs for PC use than competing with DVD-R in movie recording. Thus the “+” formats are better suited to PC file storage, but have lower compatibility with older players. Since they don’t have any advantages when it comes to burning movie discs, I don’t recommend DVD R for DVD.

DVD+R has a dual-layer variant called DVD + RW DL. Even less compatible for DVD video.

DVD+RW

DVD+RW, as you figured out from the title, is the “+” rewritable variant. +RW predates +R, and was the impetus for the format. It was originally launched in 1997 as a 2.8 GB disc (big for its time), but quickly abandoned until 2001 when a 4.7 GB version emerged. Its biggest difference from DVD-RW is support for random access writes instead of DVD-RW’s track-based writes, making it much faster and easier to make a lot of little changes to a disc.

Thus, it’s the best “floppy replacement” of the disc types. As USB thumb drives are quickly taking over that market, however, the DVD + RW is seeing rapidly declining use.

DVD-RAM

DVD-RAM is an early cartridge-based rewritable disc format. While the oldest of the bunch, it’s the least compatible (not even including the cartridge thing) of the writable discs. I used it for backups and client delivery many moons ago, but its low capacity makes it ill-suited for anything today.

Blu-ray

Introduction

Blu-ray was originally conceived of as a HD version of the DVD recorders then very popular in Japan, used for off-air recording like VHS once was. Thus it was designed as a writable format from the first instance, avoiding DVD’s painful experience trying to retrofit a writable format into a replicated disc technology. Blu-ray eventually evolved into a packaged media format once they had replication and DRM figured out.

Blu-ray launched into a fierce format war with the rival HD DVD format. Each format delivered similar experiences but with their own strengths and weaknesses. There was a lot of sound and fury from corporations and enthusiasts, since both sides felt HD discs couldn’t take off until the format war was resolved, and pushed hard for a swift victory. In the end, HD DVD was defeated in the boardroom, not the lab.

It remains to be seen whether Blu-ray is the last great optical format or the first casualty of the HD Internet.

For the readers of this book, the biggest challenges in Blu-ray may come from the expense and complexity of authoring—in particular, the requirement that all replicated discs require AACS encryption.

Blu-ray certainly isn’t going to provide the “optical floppy” and backup roles that CD-ROM and to a lesser degree DVD-ROM did. Hard drives are already much cheaper than Blu-ray per GB, flash memory is approaching that point quickly, and adoption of BD-ROM drives in PCs is much slower than it was for CD- and DVD-ROM. The success of Blu-ray largely hinges on its success for video content.

Blu-Ray Tech Specs

Similar to DVD, Blu-ray discs can be either single- or dual-layered, and single or dual-sided, in either 12 cm or 8 cm size. Unlike DVD, both layers have the same data density of 25 GB (so dual-layer is 50 GB). So far all movie discs have been 12 cm single-sided, with a mix of BD-25 and BD-50 capacity. The 8 cm discs have been used mainly in camcorders, but haven’t really caught on and appear to be fading out in favor of flash memory storage.

Red-laser media can also be used for Blu-ray content, with those discs being called either

BD-5 or BD-9; these use the DVD physical formats with the BD logical format. Some of the first generation Blu-ray players weren’t able to play those with their original firmware; it’s unclear if a meaningful number of those are still in use.

Player profiles

While Blu-ray is lauded for its interactive features, not all players support those, and none of the first generation. There have been three profiles of video players so far. In the rush to compete with HD DVD, Blu-ray launched with a simplified “Grace Period” Profile 1.0 with limited features. This has since been replaced by Profile 1.1 (a.k.a. “Final Standard” or “Bonus View”) and Profile 2.0 “BD-Live” players (see Table 23.3).

Table 23.3 Blu-Ray Profiles.

	Profile 1.0	Profile 1.1	Profile 2.0
Internet connection	None	None	Required
Local storage	Optional	256 MiB min	1024 MiB min
Secondary video decoder	Optional	Required	Required
Secondary audio decoder	Optional	Required	Required

There is also a BD-Audio profile (3.0, although it has fewer features than 1.0) for audio-only disc. There has been little interest in BD-Audio so far, unsurprising given the staggering lack of customer interest in the previous DVD-Audio and SuperAudio CD formats.

The different levels of functionality can make authoring complex, as the disc has to be able to hide or expose features based on what the player can handle.

Additionally, Profile 1.0 players without a network connection have no easy way to get firmware updates, and so may have issues with current discs. Increasingly, publishers aren’t bothering to test compatibility with older firmware in older players, instead offering instructions for how to find current firmware. This requires a user to download a disc image on a PC and burn a DVD-R.

The most popular and capable Blu-ray player so far has been the PS3, sold at a loss by Sony, hoping to make it back on game sales. The PS3’s Cell processor, which hasn’t proved anything special for games per se, is an excellent media processor able to handle full BD playback, including dual-channel decoders, in software. The PS3 has had continual firmware updates keeping even first-generation models competitive with the latest high-end dedicated Blu-ray players.

Blu-Ray Video Codecs

Blu-ray has three mandatory video codecs: MPEG-2, VC-1, and H.264. The initial wave of Blu-ray discs used MPEG-2, and didn’t deliver great quality, although this was as much due to poor mastering and QA as to poor compression. It turned out that Sony’s compression lab was using expensive, high-end Sony HD CRT displays with professional calibration. Properly calibrated with uniform gamma, CRT produces a softer image than a 1080p LCD, and its much lower black level hides blocking-in-black issues that are painfully revealed on LCDs. After the poor reviews of their first round of titles, Sony bought a bunch of consumer displays to use in their QA process.

Today, most commercial Blu-ray discs use either H.264 or VC-1, providing much more efficient encoding. While an early goal for Blu-ray was to use MPEG-2 to record ATSC or DVB broadcasts for later playback, hard drive storage has become much cheaper than Blu-ray discs per GB. And so PVRs or Media Center–style recording dominates most markets.

At this point in the market, there are a number of compression tools with “Blu-ray” settings that aren’t fully compliant with the Blu-ray spec; the output files won’t mux correctly or pass validation. Generally the high-end tools are fine, but make sure others have been able to use lower-end tools for real Blu-ray production before you spend a long time authoring.

Blu-ray supports a big array of video formats, spanning SD to HD (Table 23.4). In a welcome innovation, it supports real 24.000 fps as a frame rate, as well as the more traditional 23.976.

Table 23.4 Supported Video Formats and Specs for Blu-Ray.

Resolution	Frame rates	Aspect ratio	Codec
1920 × 1080	50i59.94i	16:9	Any
1920 × 1080	23.976p24.000p	16:9	Any
1440 × 1080	50i59.94i	16:9	H.264 and VC-1 only
1440 × 1080	23.976p24.000p	16:9	H.264 and VC-1 only
1280 × 720	50p59.94p	16:9	Any
1280 × 720	23.976p24p	16:9	Any
720 × 480	59.94i	4:3/16:9	Any
720 × 576	50i	4:3/16:9	Any

You can get a lot of SD video onto a Blu-ray with H.264. At 1 Mbps VBR (an aggressive H.264 High Profile bitrate for SD interlaced, but not insanely so), about 100 hours including low-bitrate audio can fit on a BD-50.

For typical use, even BD-25 is ample for long-form content. BD supports a maximum mux rate of 54 Mbps and a maximum video bitrate of 40 Mbps—nice to have for MPEG-2 but overkill for VC-1 and H.264.

I think DVD-9 could be a fine medium for quite a lot of content. That’s about 8 Mbps for two hours of video, which is within reach of 1080p24 with good encoders.

Blu-Ray Audio

Blu-ray has a confusing selection of audio codecs, many of them optional. With optional codecs, there’s no requirement for players to include a decoder, and so the user may need a receiver to decode that bitstream. And since the player can’t do any mixing in that case, using an outboard decoder means that any sound effects or audio commentary stored in a separate audio track can’t be heard.

So, Blu-ray audio really should be decoded and mixed in-player, and then passed out along with the decoded and composited video frames. The best and easiest option is to simply have HDMI carry out synced uncompressed video and audio. Having six discreet analog audio outputs also works, albeit with a much greater tripping hazard. In cases where HDMI audio isn’t an option, a receiver without HDMI 1.1, a fallback to TOSLink output may be required. In that case, 5.1 audio would be re-encoded from the mixed version to AC-3 or DTS.

PCM (mandatory)

Straight-up uncompressed audio is as simple as it gets. While the bitrates are high, they’re generally fine for simple titles; even 7.1 24-bit 48 KHz can fit into the 14 Mbps between the 40 Mbps max video bitrate and the 54 Mbps max mux rate. Many early BD titles included both 5.1 PCM and 640 Kbps AC-3.

Blu-ray’s PCM goes up to 7.1 channels, 24-bit per channel, and 96 KHz. I remain of the belief that anything beyond a well mastered 48 KHz 20-bit is mainly useful for annoying bats, but don’t discount the placebo power of big numbers, and the marketing value thereof.

The downside to PCM comes when trying to pack a whole lot of stuff onto a single disc, particularly a red-laser BD.

Dolby Digital (mandatory)

Dolby Digital is a mandatory codec, and is included on most discs. BD’s AC-3 goes up to 640 Kbps maximum, compared to the DVD and ATSC maximum of 448 Kbps. Unless you’re targeting DVD-5/9, the full 640 Kbps should be easy to fit.

DTS-HD High-Resolution Audio (mandatory)

One big change from DVD is that DTS has become a mandatory codec, after being optional on DVD. This means a disc can use only DTS, without an AC-3 or PCM fallback.

While the higher bitrate 1.5 Mbps flavor of DTS was rarely used on DVD due to the high bitrate, it’s much more feasible on Blu-ray media. I can’t see using the half-rate mode of DTS over 640 Kbps AC-3; AC-3’s greater efficiency should give it better quality if there’s any apparent difference.

The HD extension to DTS allows up to 96 KHz 24-bit (again, mainly useful for annoying bats) at up to 6 Mbps. DTS-HD is a good intermediate for content where 640 Kbps AC-3 might not be enough, but lossless is too expensive in bits.

Dolby Digital Plus (optional)

Dolby Digital Plus was a great codec on HD DVD, but is much less applicable in its Blu-ray implementation. BD DD+ is merely an enhancement layer to add two extra channels to take a 5.1 AC-3 stream to 7.1 audio. Given the dearth of 7.1 audio sources in general, DD+ on BD is unlikely to see wide use.

DD+ can use up to 1 Mbps extra, but since it’ll only contain two channels compared to the 640 Kbps used for 5.1, in practice it’ll be much less.

Dolby TrueHD (optional)

Dolby TrueHD is Dolby’s lossless codec. In its Blu-ray implementation, it’s really two different bitstreams muxed together—a normal AC-3 track and the lossless track, using the Meridian Lossless Prediction algorithm from the old DVD-Audio format. Thus, TrueHD is always backward-compatible to old players, although no bits are saved through this mechanism.

TrueHD on Blu-ray supports up 24-bit, with up to 7.1 with 48 KHz, and up to 96 KHz with stereo only.

DTS Master Audio (optional)

DTS Master Audio (DTS-MA) is the DTS lossless codec. Like TrueHD, it includes a backward-compatible core. Unlike TrueHD, the lossless layer is encoded as the difference between source and the core encode, making for a more efficient encode.

DTS-MA goes up to 24-bit and 96 KHz for stereo and 48 KHz up to 7.1. The bitrate can go up to peaks of 24.5 Mbps.

Blu-Ray Interactivity

HDMV

Initial Blu-ray titles used the HDMV interactivity layer, which provided a DVD-like authoring environment. It’s largely being displaced by the much more capable BD-J.

HDMV may become memorable as the last major media technology to use indexed color graphics; HDMV image overlays were 256-indexed color palette PNG.

BD-J

BD-J is the “real” Blu-ray interactivity layer, and increasingly the only one supported by authoring tools.

BD-J is an implementation of Java that runs on BD players, and working with it directly is software engineering, not menu design. Vendors are building design layers that sit on top of BD-J to provide richer design environments that don’t require coding, like they did with DVD.

Blu-Ray Mastering

Replication

Blu-ray struggled for a while with BD50 replication, which frustrated many proponents, since that offered a hypeable capacity advantage over HD DVD’s 30 GB max.

AACS

Perhaps the most frustrating aspect of Blu-ray is the mandatory requirement for AACS encryption on replicated discs (red laser too), which includes some pretty stiff fees. And even these fees are a big reduction from what they were until the summer of 2009 (disclosure: thanks in part to lobbying from the Interactive Digital Media Association, of which I’m a board member).

This adds to the cost and complexity of using replicated BD discs; coupled with the high costs of BD-R media, it makes business models based around short-run titles (think wedding videos) a lot less economically feasible than with DVD (Table 23.5). DVD-R is the only feasible choice today for short-run Blu-ray content in many cases.

Table 23.5 Mandatory AACS Fees.

Fee Category	Original	Current	Note
AACS Content Provider License	$3,000 one-time	$500/year	$3000 eliminates yearly fee
AACS Title Key Certificate	$1,300/master	$500/master	Per master, not per title
AACS per-disc Royalty	$0.04/disc	$0.04/disc	Uncapped

Duplication

Duplication on Blu-ray has a couple of big advantages over DVD.

First, since Blu-ray was designed as a writable format, all players can handle all BD-R discs, both 25 GB and 50 GB.

And more importantly, AACS isn’t required, which makes low runs a lot more cost-effective.

But there’s a big downside; BD-R blanks are a lot more expensive than BD replication. We’ll hopefully see prices fall over time as we did with CD-R and DVD-R.

DVD-R

Blu-ray works fine on red-laser burnable media, including DVD-9. And with VC-1 and H.264, you can actually get a couple of hours of pretty good-looking video on a single disc.

In a compressionist showoff move, I once did a two-hour 1080p movie on DVD-9 with MPEG-2, using 8.5 Mbps ABR. It wasn’t “Blu-ray quality” but held its own against a lot of broadcast ATSC MPEG-2. Modern VC-1 and H.264 implementations can do very nicely at these bitrates at 1080p24.

If you stick to peak bitrates of 15 Mbps or less, you can use a two-second GOP, which can help efficiency on a small disc. Also, while Level 4.1 H.264 requires four slices in the Blu-ray spec, lower bitrates are Level 4.0 and so can use single-slice encoding for a little better efficiency.

BD-R

BD-R is the write-once flavor of BD. It was originally very expensive, but BD25 blanks have dropped to about $4 in volume as this book was written (still more expensive per GB than DVD-R). Dual-layer are also available and have good compatibility, but are about four to five times as expensive.

BD-RE

BD-RE is the rewritable flavor of BD. It’s less than twice the cost of BD-R, so it can be useful for test discs. It’s available in BD25 and (quite expensive) BD50.

There had been high hopes for BD-RE to be the next generation floppy/CD-RW/DVD RW sneakernet/home backup format, but this seems unlikely unless prices drop dramatically and Blu-ray drives become much more common in PCs.