With the current state of DVD production tools, spec compliance variance by player manufacturers, certain realities at replication plants and the lack of analysis and verification capabilities of the vast majority of DVD production enterprises, among other reasons; there is no way to tell the client the Image being sent into the replicator is free of error or potential issues and ready to go, without extra measures.
Certain myths, among them that it's OK to submit a client's master for replication on DVD-R, will be dispensed with here. Various other important misconceptions will be brought to the industry's attention; for instance, that most authoring houses can "verify" or "compare" the DDP Image on the the DLTs and restore their contents to hard drive for checking. While those proofing practices are useful and recommended; by themselves, are not enough to meet professional standards in Image verification before replication submission.
The combination of shortcomings in authoring tools, the complexity of the replication process, and the reality that most authors are, rightly, not at the nuclear engineer level with years of experience when it comes to 'DVD-Video' and a plethora of other specifications, and are also without verification tools, creates an unwitting collaboration of factors that push the percentage up on the possibility that something will go wrong.
Therefore, I propose that whoever is taking responsibility for the replication run (the person or entity backing the check for the replicator), be educated on the advisability of further proofing, playability testing, Image analysis and verification of the DVD master and the check disc manufactured from it. A standard orientation would include briefing the client on the the less than perfect state of the performance of our industry's production tools, medium and playback devices, and contain professional disclosure of the authoring studio's inability to perform fully in house, all the checks required.
In fact, it's only when the client has been given the opportunity to accept or decline the various options of quality assurance (QA) and professional verification, from minimally approving a check disc for scrutiny, all the way to full scale verification of the DVD Image both 'before and after', that credence is given to the spirit of any waiver releasing the authoring studio from liability at the manufacturing plant.
In other words, clients need to be told that their project will be submitted to the replicator with mainly a hope nothing bad will happen, unless they approve extra, high-end industry standard measures.
It is intended that the definition of DVD Verification itself, will be expanded by the question asked in this report. It's not only running a battery of tests on the Image, and playability testing, etc., but also a thorough going over of the implementation of known and new 'best practices', and avoidance of myriad potential gotcha's of the DVD's programming strategy and asset domain placement (authoring), and a gaggle of other issues under the authoring studios control, including preparing and submitting the Image in a manner that 'encourages' the replication process to be successful. Without approaching full DVD Verification under the extended definition outlined here, a gash exists in the authoring studio's Quality Control regimen.
This article attempts a brief circumspection of the land faced by DVD producers and their clients, in such a manner as the answer to 'Why Verify the DVD' materializes; for when all is accounted for, it is possible to consistently get through replication with DVDs that play as intended in everything.
Replicators do not Verify
Since the DVD-Video Format's beginning, it became immediately apparent and has proved up to this minute, that the 20 or so various file, data, stream and other specifications which DVD-Video encompasses are open to interpretation by authoring tool providers, player manufacturers and decoding software programming teams, and others. All of the misinterpretations by these various camps need to be known and accounted for as part of any competent DVD proofing and verification strategy. But it is important to also know; For the most part, DVD disc manufacturers have left themselves almost completely out of the loop in gaining years of experience on how the DVD-Video file system and associated specifications are supposed to work.
The DVD-Video file system is one of many "file systems" that are presented to replicators manufacturing CDs and DVDs. Cursory checks are automated into the process pertaining to the particular type of Image (CD, DVD and their different formats) to be transferred to the Glass Master, Only a few of the file system specifications that make up a DVD, including the DVD-Video file system, are checked at all by replicators.
Ken Byrd, of DVDAustin acquits the DVD Replication industry in this manner, in a post on Tully's DVDList:
"With incredibly rare exceptions, replicators care not about your authoring software, your workflow, your 'safe authoring', or your 'best practices'. These are your responsibilities and yours alone. Replicators make stampers and sell polycarbonate. Lots of it. From a technical standpoint, you keep your end of the bargain, they keep theirs. Most replicators don't know a PGC from a BLT - and they shouldn't. That isn't their job....
Verification, compliance, and usability are all responsibilities of the authoring house. If I haven't made it clear, let me hammer the point home one more time: Replicators are only responsible for the physical integrity of the media they create. If they're coasters, they should be coasters with low jitter and good reflectivity."
Replicators are almost wholly dependent on Eclipse Data Technologies and Doug Carson Associates (DCA) for Image checking software.
If the replicator finds a problem with the Image, as reported by Eclipse ImageAnalysis, they can only reject it. Though 'good' replicators will work with the client and the legendary Eclipse Tech support team to help figure out the issues, they can only try to identify the problem for the authoring studio.
But with only a few parameters of the DVD-Video specification checked during the Image analysis process, the boundaries of these helpful technical support sessions may not always extend to a solution.
Since replicators don't check the DVD-Video file system for spec compliance, nor for the dozens of issues that can occur that are not against the spec per se, but yet can hinder or even cripple a DVD's performance, none the less; An Image that gets past a replicator's initial checks is still not in the clear. If there are further problems still lurking on the DVD Image, the authoring studio and their client can only hope that playability testing and other testing of the check discs catches any remaining issues before the disc's release into the market.
Barry Braverman in his "Defensive DVD Authoring", describes the 'joy' (not) experienced when a DVD with issues is prevented from being distributed onto retail shelves:
"You know what I'm talking about. Your check disc comes back from the replicator and there are problems. Maybe one or two menu buttons mysteriously don't work or the disc plays back in fits and stumbles. Perhaps the navigation is exhibiting erratic tendencies, illogically executing post-commands or end jumps. Of course the replicator blames YOU and insists that you (or your client) pay for a new glass master to the tune of $1000-$2000."
The only way a replicator can help a sometimes exasperated client on a deadline at this juncture (or, if the Image is rejected outright), is to point them to either their own authoring services, or an affiliated authoring studio. These referrals are, at times, preceded by the replicator telling the client the authoring system their authoring studio is using has had lots of problems. It is hard to blame a replicator responding in this manner to a client, who asks them what can be done when unexplained halts in the job arise; as remember, the client is paying the replicator, and the replicator doesn't want to lose the client, either. The only way to fix a problem DVD Image includes re-pre-mastering as part of the procedure, and sometimes the deadline results in the perceived need to switch authoring studios when the job is held up at the plant.
Hopefully, the authoring studio's relationship to the client is strong enough to withstand such a redirection attempt when there's a hold up due to errors or warnings. But a solution will have to be found fast. Important Note: When engaged, TFDVD Research Labs moves to not only quickly find out what the problems are and fix them, but works to help preserve the Authoring Studio, client relationship.
The DVD Image Analysis Process is Not Standardized
The Rules for what to do with any of the over 500 situations Eclipse ImageAnalysis reports on, is left up to the discretion of the Replicator.
Which DVD Image parameters reported on, rules for what to reject in an Image, vary from plant to plant; and even if the Eclipse ImageAnalysis reports have been read, in some cases; as there are more than a few replicators out there who are going to replicate what the authoring studio sends them, no matter what, even if they do 'read' the report .
Rarely does a replicator just run an Eclipse Image Analysis report on an Image. These Analysis reports are generated, automatically, in the context of other functions the EclipseSuite of tools do for the replicator; like transfer an Image off of tape or DVD-R, or, when the Eclipse controlled encoder is mastering the Image to the glass master. And during the transfer of the Image off of tape or disc, no replicator that we know of will set any of the rules violations to abort the process (all the Replicator's rules settings in an Eclipse ImageAnalysis are listed as part of the report, and we've never seen an 'abort', other than when an operator aborts the process, say due to a DVD-R with an Image that can't be read, in the dozens of replicator Eclipse reports we've seen). This means that Images with errors, that should really be aborts, are ending up on Replicator's servers in line for mastering. There are many replicators who will let the client know if something questionable shows up on an Eclipse report, and that is good; but a bullet-proof DVD Verification and proofing strategy does not allow for problems with the Image at the plant, in the first place. Only replicator manufacturing error can attempt to "do the strategy in"; and then the high-end process is prepared for even that possibility (see below).
Image Analysis Software's Efforts to "Keep up"
What has been found to cause playability problems, etc., are added slowly to Image analysis software over the years, and many replicators are still using several years old versions. TFDVD Research Labs received a circa 2001 version 1.4 Image Analysis report from a replicator a few months ago (we're up to version 5.1 right now). It wasn't until March 2005 that Eclipse ImageAnalysis added checking for the layer break for DVD-Video specification compliance on DVD-9's, for instance.
Before this addition, hundreds of thousands of DVDs were formatted by DVD Studio Pro® (from version 1's release, until 1.52; a period of almost two years) with incorrect layer break settings on certain project types, and could not reliably play video through the layer break. Yet, these discs sailed through replication and in many cases were distributed into the market.
Gear Pro Mastering Edition, for years, has been a DVD-9 accident waiting to happen since the beginning of it's offering the capability, with no way to target and prepare a cell for the layer break from directly within the program. It didn't help that their support web site at first had instructions for formatting DVD-9's, that if followed, guaranteed two blown glass masters (the Layer break cell setting would be off by the number of sectors contained in the menu domain of the VTS the layer break occurred in, according to their instructions; also, there was no accounting for how to properly flag the LB cell). Gear released Version 7 with stated CSS copy protection formatting capability that didn't work, either (Here's the first page of an Eclipse ImageAnalysis report of a Gear Pro Mastering Edition formatted CSS Image, with tens of thousands of warnings!).
A competent DVD Proofing and Verification strategy never assumes the authoring tool or formatting software outputs a valid Image. They must pass tests before going to replication.
There are still many replicators out there that do not check dual layer Images to make sure the layer break occurs on a cell (first entity on layer 1). Here's a recent letter an Indie film maker wrote to us in desperation, with four blown glass masters (two DVD-9 attempts) at two different plants (output by Encore DVD, and Gear Pro Mastering Edition, respectively), with several months of project suspension.
We received the two discs, one from the thousand run, and one check disc from the second attempt. The Eclipse Image Analysis report for the Encore formatted project (shown after the letter in the above link) not only show the layer break did not occur legally on a cell at all, but it's lead-in control data stated it had more data on layer 0 than is allowed by specification! The Gear formatted project just made the layer break illegally in the data (not on a cell). Why didn't these two replicators catch these mistakes before mastering?! (see above and below).
Many replicators will not even compare a check disc off the line with the original Image submitted on the DLT (or DVD-R). More cautious replicators on larger runs will compare a check disc (first off the line) to the original Image with Eclipse ImageVerify. And a few replicators even have several DVD players in a rack that they use to make sure what they just stamped will play, right before continuing the run.
Replicators do not test their discs for physical errors
The main enemy of the replicator (well, besides all these unchecked Images being submitted to them) is time. Replicators can only survive if they have massive uninterrupted throughput of Disc production. They don't even have time to properly test their output with CATS and other physical testing. They rely, instead, on what's called Bit-for-Bit Compare, checking that the image on the replica is identical to Image submitted. Unfortunately, this is a faulty test because they use better high-speed optical drives than are found in DVD players, and regularly results in entire runs of discs getting out to market with all kinds of playback issues. Here's an article going over the issue and how TFDVD Research Labs has combated this reality at the plant:
Breakthrough 'TFDVD Research' on DVD Replication Problems
DVD Manufacturing is not Standardized
It would take many pages to begin to attempt to address the variance in equipment, procedural workflow, accepted submission formats, materials, quality control, and other issues which affect the mastering, bonding, packaging and other parameters that effect the viability of the manufactured discs.
Basically, there are defined 'Physical Specifications', but it's left completely up to the manufacturing concern how to design and engineer the process which create the discs to conform to them. And there are old, new, better and not the best ways to go about it. Two pages in an updated version of the 'Physical Specifications for DVD-ROM' put out by the DVD Forum, go into ways that disc manufacturers are making mistakes with discs 'in the market place', and are told 'should be avoided'!
In about half the replication trouble-shooting cases brought to TFDVD Research Labs in the past year, where authoring studios were having problems with the discs after the replication run, it proved to be the replicators fault - an article in itself. Though we had not been involved in the pre-mastering or testing of these particular discs, beforehand, these studios came to us after days, and in several instances, months of project limbo.
When There Are Problems at the Plant
It is more than useful to have extensive documentation certifying the Image, when something goes wrong during the replication process.
Replication can be a labyrinth
Most replication facilities these days do not create the 'glass masters' on site. And of those that do advertise in-house stamper manufacturing, most have to send out for this service with Images that require dual layers and/or CSS and Macrovision. Add into the mix that a significant volume of replication is handled by replication brokers, who have a stable of replication facilities they're contracted with, each with their own manufacturing process, and it makes it prudent that the Image being sent to run this gauntlet be a certified known entity. If a problem arises with the manufactured discs, the documentation replicators respect, i.e. an Eclipse ImageCopy or ImageAnalysis report (included along with the DLTs), might be the only thing to fall back on for proof that what was submitted was perfect, as it can sometimes be impossible to retrace the chain to communicate with who really handled the Image.
Check Discs Are Not Optional
It is imperative that at least 10 check discs be ordered and several tested, before the final run is commenced. It is not enough to get one or two check discs, as often the first several off the replication line from the stamper will contain imperfections and errors, which is normal. 10 discs, which will not cost the replicator a minute more in time or material over one or two check discs, allows a more representative sampling of the run. TFDVD Research Labs asks for the majority to run through our routines; the authoring studio and client can divvy up a couple so they can view the production, now finally on a replica.
If there's a suspected manufacturing glitch discovered during the check disc phase, again, prior documentation certifying the Image submitted, is crucial to save the needed time to meet a deadline, as it can take days and even weeks to retrieve the original master tapes to run comparison tests. For even if an obvious manufacturing error (and not so obvious; as erratic navigation can be caused by unreadable sectors in the IFO files where the routing commands are) is found on ALL the check discs, it doesn't necessarily mean the replicator will readily agree to redo the glass master (not to mention redoing an entire run) , due to various other factors that can (and will) be cited by the replicator that could possibly cause the playback imperfection (illegal Mpeg 2 data spike streams, etc.). By the time everything is sorted out and expense undertaken to bring in a third party, etc., the deadline is long gone. Prior certification of the Image, and having the same master Image sent into the plant via DLT on hand to compare the check disc to, can greatly speed up any troubleshooting process. TFDVD Research Labs provides this service with our Premiere Pre-mastering Process, should the need arise; and of course we're available to troubleshoot jobs we don't pre-check.
Technically, the authoring studio and client have every right to expect perfectly manufactured discs, even if they don't request a check disc. But ordering a check disc is still necessary, similar to inspecting a print proof before running thousands of copies of a magazine, etc.
For all pre-mastering jobs, TFDVD Research Labs maintains for an appropriate time, an exact copy of the Image submitted to the plant on DLT to compare the check discs against with Eclipse ImageVerify (It wouldn't do to 'pre-master twice' to two different sets of tape, for instance). This is the bare minimum required for hope of protection against possibly stubborn replicators and very expensive job delays, when the check discs have suspected manufacturing problems.
Basically, if check discs are not ordered, the client and the replicator are taking an unnecessary risk; as both could use one last proofing of their work.
TF's DVD Research
Some Milestones on
Trai's Quest to Master DVD
(not in chronological order)
Trai Forrester begins to learn the replication process first hand, taping the manufacture of a DVD from start to finish, at Deluxe Digital in the Spring of 2000.
Trai gets his start in DVD in April 1999; selling, training and supporting customers on Sonic Creator and Spruce Maestro DVD Authoring systems, up and down the East Coast (Boston to Virginia) for The Lerro Corporation
Trai attends Sonic's first factory training session on Sonic DVD Creator at Video Symphony, in Burbank CA, Spring 1999 (Taught by John Broddeus).
Trai sits down with Spruce rep Mason Peffer at The Lerro Corporation, where they both begin to learn the Spruce Maestro DVD Authoring system Spring, 1999.
Apple's purchase of Spruce 3 years later would have Trai well prepared for clients and TFDVD.com members, with the release of DVD Studio Pro 2 which is based on the old Maestro, in April, 2003.
Trai interviews Jim Taylor who would later become Sonic Solutions Chief DVD Technology Officer, at the DVD Pro Conference, Summer of 2000, for his 'DVD from A to Z' DVD project.
Trai meets Tim Edwards (finally in person) of Denon Digital, summer of 2005. Tim's custom software used at Denon first identified DVD Studio Pro© was incorrectly setting the layer break, back in July 2002. This collaboration with Tim lead to Trai's ability to develop and test the work around which became known as "TFDVD-9 Maker". This procedure was the only way to guarantee a spec compliant DVD-9 in DVD Studio Pro, for almost two years afterwards!
Hundreds of authors worldwide, most of them members at TFDVD.com, used the TFDVD-9 Maker method at dozens of replication facilities. It was a cumbersome process, but performed flawlessly.
Trai's DVD Studio Pro support site, founded in late 2001, identified the bugs and created the workarounds to insure successful projects with DVD Studio Pro. Over 1,100 professionals stopped by for a time in its 5 year run. Currently, the site is being transformed into the
DVD Producers 'Guild
For a nostalgic tour of some of the DVD Industry's raging DVD authoring issues many of which are still relevant, enter "TFDVD.com" in the Internet Wayback Machine. TFDVD.com and it's members have had many "DVD victories" over the years. The biggest achievement, was the development of TFDVDEdit Mastering Edition (now DVDAfterEdit), created and tested, initially, entirely by TFDVD.com members!
"I'm beginning to understand. All DVD Studio Pro roads lead to Trai" - A Poster on 2-Pop, Spring, 2002.
Unsolicited comments from TFDVD.com members, during 2002/2003
Trai's studies of the DVD VIDEO_TS Folder will lead to a transformation of professional DVD authoring.
In the Fall of 2001, Trai starts his research into the files DVD authoring systems output, with the help of a couple of programs. IFOEdit and VOBEdit were initially crucial to Trai's ability to solve major and minor issues with DVD Studio Pro, and other authoring systems. Although he has never condoned the main purpose the IFOEdit program was developed for, Trai gives credit where credit's due, none the less. Here's Trai's acknowledgement in a post announcing his own DVD Forum licensed VIDEO_TS folder editor, TFDVDEdit 2, halfway down the thread, (Trai is the first person to use his real name on Doom9! :-)
Trai Forrester works with Gustav Karlsson, of Gear Software, August - November, 2002
The TFDVD-9 Maker process to fix DVD Studio Pro's raging DVD-9 formatting bug (for non CSS projects), required a third party formatting application. Gear Pro Mastering Edition beta was the only show in town at the time. Trai worked with Gustav to help get their program with a method so it could properly make the layer break, as well.
Trai Forrester's 'New Constellation Technologies' becomes Licensed with Eclipse ImageAnalysis, as part of the verification of the TFDVD-9 Maker Process, September, 2002.
Trai kept the development and deployment of the TFDVD-9 Maker process (60 page illustrated PDF) under strict secrecy at TFDVD.com where support was also provided. Not a single user blew a glass master. The process was rigorously tested with Eclipse, MEI, Interra Surveyor, and even run through Intellikey Labs' top DVD playability package: 100% successful in over 300 players!
TFDVD-9 Maker would soon lead into Trai leading the development of the world's only professional 'post DVD authoring system', now known as "DVDAfterEdit".
Trai's DVD Consulting practice takes off with the DVD specification knowledge gained inside the VIDEO_TS folder.
Example: In the fall of 2002, one of the world's largest manufacturers of high-end DVD and music entertainment systems contacts Trai to troubleshoot why their pristine surround tracks that were destined to be used in their thousands of retail locations, weren't sounding right. Trai quickly finds it's not their hardware, but a DVD Studio Pro bug with how it designated it's Dolby surround attributes for the audio stream! Trai is able to quickly fix the setting right in the VIDEO_TS folder.
Trai personally causes a substantial 'migration' of Mac DVD Studio Pro users over to Windows to get Mpeg 2 encoding with ProCoder. due to Apple's initial horrible Mpeg encoding offered with the release of DVD Studio Pro 1.0. Trai had approached Canopus for help in the Spring of 2001, and by Fall of the same year, he was on the ProCoder Alpha then, beta team.
It is said: "The Teacher teaches that, which he most needs to know."
Trai 'talking DVD', on the DVD Demystified Second Edition DVD. Shot in early 2000.
Trai and Adrian Ramsier of All4DVD, take aim at the Pioneer's booth (whose entire booth floor were plasma screens) at Infocomm below, shooting a segment for Trai's 'DVD from A to Z - A DVD-Video Production Primer', in Summer of 2000. Trai hopes that one day, the project will get done.
Trai finds his background in video production is key to helping himself, members and clients understand the DVD Production workflow.