Hatter vs. Tab: The Codec Grudge Match [SPLIT]

[Tab.]

Well yeah, I remember discussions way before Matroska matured and it might have even been before Matroska split off from the MCF project about the advantages of file attachment abilities for codec installers and stuff. That may have been the prinicpal concept behind attachments.

Anyway, I don't think that would work for hardware players, they don't use software to decode the stuff, just dedicated hardware chips. But the team has already drawn up specs for hardware support, including different levels of support for video/audio codecs and subtitle formats.

[Tab.]

Btw, something I stumbled upon in the new A/V containers forum from a little while ago about possible but unlikely future official support for ogg used as it is as ogm.

Ogg Theora is using a Xiph internal way to identify the stream content and the codec that was used, similar to what OGM does to identify a Vorbis audio track in it.

OGM is using a Windows centric Video for Windows structure to identify codecs, mainly BITMAPINFOHEADER and also VIDEOINFOHEADER AFAIK. I doubt that Xiph will decide to use this method to identify non-Xiph video streams, but maybe they will make their constant promises become reality and adapt this way that Tobias has chosen for OGM, and will use it for the official Ogg Theora library also.

Another thing that differentiates OGM from Ogg Theora is the way the GRANULEPOS is used to describe to position in time ( similar to a timestamp ), as Monty has modified this for Theora to be able to identify keyframes from the GRANULEPOS, without the need to ask the codec where to find the next keyframe, hidden in all the Ogg pages. Adapting this method for non-Xiph codecs would make a lot of sense IMHO, but then automatically all old OGM files would never comply with the official Ogg container video storage solution, unless some special 'OGM mode' was put into the specs, allowing normal GRANULEPOS also.