W3C home > Mailing lists > Public > public-tt@w3.org > April 2014

IMSC Issues / Change Proposals

From: John Birch <John.Birch@screensystems.tv>
Date: Wed, 2 Apr 2014 10:43:18 +0000
To: Nigel Megitt <nigel.megitt@bbc.co.uk>, Pierre-Anthony Lemieux <pal@sandflow.com>, Michael Dolan <mdolan@newtbt.com>
CC: "Public-Tt (public-tt@w3.org)" <public-tt@w3.org>
Message-ID: <0981DC6F684DE44FBE8E2602C456E8AB014F6FE3C2@SS-IP-EXMB-01.screensystems.tv>
Dear Nigel, All,

I would like to formally request consideration of 3 issues with respect to the IMSC publication. I have separated my concerns with the current document into 3 issues because, following informal discussions with both Nigel and Pierre I think it is easier to deal with them as separate proposals for changes. Before delving into the specific issues I would like to say that I do believe that a (TTML) distribution profile specified within IMSC, if implemented in processors, will encourage widespread adoption of TTML.

I have raised these issues within an email because, as an 'Invited expert', I do not have access to the online issue tracker. I believe the most appropriate approach is therefore to raise them 'through the chair', such that, if they are considered relevant, they might be formally presented on my behalf within the TTWG. I hope that these issue might be considered in the context of future work on IMSC. As a proposer of changes to IMSC, I would be happy to contribute to the process of drafting proposed changes to the IMSC document.

Issue 1: The applicability with respect to the production and distribution chain needs to be made clear.

I do not believe that IMSC represents the requirements of production and archive. More importantly, it is not clear from the IMSC draft whether the guidance in the document applies only to distribution forms (e.g. streamed and downloaded content), or more widely to the assets used to create that delivered content. Consequently IMSC has a potentially severe implications on the nature of subtitle and caption *mezzanine* files. There is currently I believe quite a high likelihood that the IMSC 'limits' might be naively assumed as an acceptable 'mezzanine level' for TTML subtitle and caption content. IMHO that would be a disaster as it would destroy all possibilities of 'smart' subtitle / caption conversion where a single rich archive serves as a source for many different forms of subtitle presentation at distribution.

I do not believe that examining the requirements of a decoder, a manufacturer's economics, archives of existing *distribution format* subtitle files, or examining current or proposed delivery mechanisms, provides the best foundation for understanding or supporting the requirements of the content owners, distributors or authors. As an analogy, it's rather like saying that there is no need to shoot in 4K because *we* only need 640 x 480 at the display. However I do believe all of those previously mentioned resources are valuable sources of what is current convention and what might typically be the requirements of current provision in traditional broadcast, and so what might be required if you wish to *replicate current subtitle provision* on the Internet.

But a huge assumption is being made if it is suggested that current conventions automatically map to all Internet subtitling, or that current display conventions are the minimum requirements at authoring or mezzanine / archive level. The traditional linear broadcast infrastructure has naturally tended to create a close correlation between the authoring and distribution forms used in subtitling (e.g. STL files are essentially a Teletext encoding, and SCC files are basically CEA 608). But the near one to one relationship between the authoring and target audience / distribution form does not generally hold for subtitling on the Internet. Internet subtitles will need to be authored without the comfort of a convenient set of constraints (effectively imposed by the distribution mechanism), instead there is a real requirement for a (hopefully) automated mapping between a single rich subtitle mezzanine file and many different forms of Internet subtitle distribution file, where the differences in distribution will be significantly greater than seen previously in the linear broadcast world.

I am very concerned that IMSC in its current form will (perhaps inadvertently) promulgate a viewpoint of 'simplistic subtitling or captioning' worldwide in the new landscape of Internet Subtitling and Captioning. This I feel would be much to the detriment of ultimately every participant in the captioning or subtitling chain (including the viewers!). Consequently I think the TTWG needs to fairly quickly demonstrate that it acknowledges a wider perspective, and to clearly show the *relevance or applicability* of the standards (profiles) it does choose to define within that landscape.

>From my perspective the current IMSC document should sit alongside the SDP-US and other standards as a (*very good*) definition of some of the intrinsic aspects of traditional subtitling and captioning that are not covered by those other standards, E.g. subtitle presentation rates, subtitle text characteristics and layout conventions. In effect it does support extremely well the migration of blu-ray / dvd subtitling conventions to the Internet!

Issue 2: It needs to be clear whether IMSC is a processor or a document specification - at the moment it appears to be both simultaneously.

This issue is related to the previous one, but is perhaps also a reflection of the lack of context with the IMSC document itself (Issue 3). I believe IMSC should be careful not to present as a document specification, but should be far more concerned with the processing of, and performance requirements for TTML documents that are used for subtitling. As a performance model for defining expectations it is a very good framework, although I believe that at the moment the 'model' has a greater emphasis in the document than the 'magic numbers'! My personal preference would be that the 'magic numbers' should be introduced first, with a following explanation of how those numbers relate to the model.

In addition, I feel that aspects of the current IMSC document that basically reflect CFF-TT restrictions should be removed, as I feel this limits the general application of the IMSC model to other standards. Clearly this separation may be difficult in some respects since some of the CFF-TT limitations are reflected in the model itself... but my personal preference would be to see IMSC develop into a generic performance model that is broadly applicable, rather than being compromised by being tied to a specific single implementation.

Issue 3: The source of addressed requirements is absent.

Rather than saying anything about the domain of subtitling and captioning, the IMSC document immediately launches into setting constraints (actually without even any explanation as to why they might be considered necessary!). The constraints themselves are clearly based upon a limited subset of the necessary requirements for subtitling and captioning generally, but there is no text in the document about how the requirements were gathered, or how the conclusions were drawn from them. If the IMSC was a document with the stated intention (within a more constrained Title) of describing what limits might be acceptable for Internet distribution of prepared subtitling or captioning provision *for long form content* this would be OK, but I am unconvinced that IMSC is currently appropriate for a wider scope that also necessarily includes live or short form content (e.g. talk / news /game shows and fast paced reality genres).

Whilst it may be out of scope of the TTWG, a document that describes the landscape of Internet Subtitle and Caption provision and the challenges faced is, IMHO, *a necessary context* for IMSC. Setting the current IMSC document within a larger more encompassing work, where the IMSC is a description of acceptable boundaries for certain types of program delivery and subtitling / captioning requirements would be more useful, and less potentially confusing . That does of course imply quite a much larger task than perhaps was envisaged. Part of this suggested context might also cover the concept that the IMSC might be considered as the lowest baseline, and that it would be hoped that most (many) implementations exceed the requirements of IMSC? And that perhaps in future the expectation is that the 'goal-line' might move as technology improves?

This is what is not present in IMSC - a distinction between what is possible technically, what is commercially palatable, and what might 'ethically' and reasonably be expected. One personal concern is that IMSC currently gives an unfounded impression of a technical justification of 'acceptability of a lowest common denominator service' for all Internet subtitling provision *because of an absence of this contextual prose*. IMSC should make it clear that the performance boundary in service presentation that is effectively created by IMSC is nothing to do with technical capability, but is wholly to do with other considerations (which are still valid). These considerations being amongst others; tradition, economics and practicality. Of course it is a delicate task to convey the concept of 'minimal acceptability' appropriately. A strategy might be to incorporate a strong recommendation within IMSC for implementations to considerably exceed the IMSC defined performance criteria.

With best regards and with the greatest respect,

John Birch
John Birch | Strategic Partnerships Manager | Screen
Main Line : +44 1473 831700 | Ext : 2208 | Direct Dial : +44 1473 834532
Mobile : +44 7919 558380 | Fax : +44 1473 830078
John.Birch@screensystems.tv<mailto:John.Birch@screensystems.tv> | www.screensystems.tv<http://www.screensystems.tv> | https://twitter.com/screensystems

Visit us at
NAB, Las Vegas Convention Center, 7-10 April, stand N5821

P Before printing, think about the environment

This message may contain confidential and/or privileged information. If you are not the intended recipient you must not use, copy, disclose or take any action based on this message or any information herein. If you have received this message in error, please advise the sender immediately by reply e-mail and delete this message. Thank you for your cooperation. Screen Subtitling Systems Ltd. Registered in England No. 2596832. Registered Office: The Old Rectory, Claydon Church Lane, Claydon, Ipswich, Suffolk, IP6 0EQ
Received on Wednesday, 2 April 2014 10:43:51 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 5 October 2017 18:24:14 UTC