On Mon, December 3, 2012 9:01 am, ioannis ioannis wrote:
Hi, Following George's comments, I have worked a bit during the weekend on my installation.
With the following configuration in apache2.conf I avoided using the relevant rewrite rules in .htaccess and all seems to work fine
<IfModule mod_actions.c> ScriptAlias /oggz-chop /usr/bin/oggz-chop.cgi Action application/ogg /oggz-chop.cgi Action video/ogg /oggz-chop.cgi Action audio/ogg /oggz-chop.cgi Action application/ogv /oggz-chop.cgi Action video/ogv /oggz-chop.cgi Action audio/ogv /oggz-chop.cgi </IfModule>
Yup, this one of the two other ways to trigger oggz-chop handing. I think AddHandler should work as well.
However there is an important difference in that the above example will send all Ogg multimedia through oggz-chop whereas the RewriteCond statement in .htaccess will only trigger oggz-chop if there is actually a query string present. Without a query string the media is delivered as a static file (w/ HTTP Range request support).
This difference affects at least three things: a) md5sum "correctness," b) load times, and c) client-side seeking performance.
For a) oggz-chop can and does rewrite some of the header metadata and can even omit "extra" Ogg streams listed in RFC3533. Thus, a file uploaded to the server will differ from the same one downloaded; consecutive downloads will also differ slightly.
For b) the current implementation of oggz-chop starts at byte 0 on the media and reads all bytes into memory until the specified time ranges are satisfied; once this is completed the media is then sent through Apache to the client. For ~1GB recordings read off of a non-SSD drive the delay can be around 8 seconds or so. A search algorithm that reads the first few Ogg packets, generates a bitrate, uses this to jump-to the approximate byte range of interest, and bumps forward/back as needed shouldn't be 'hard' to implement -- patches welcome?
On the plus side subsequent requests are read from memory (=temporarily fast access) and the Oggz codebase is mostly bullet-proof from a security perspective. The folks in #theora are friendly and highly knowledgeable.
For c) some browsers (e.g., Firefox) implement "fast" seeking based on byte range requests on raw media. I'm not entirely sure how well this interacts with oggz-chop other than to say that a well-trimmed video segment (i.e., a 10-min clip delivered w/ oggz-chop and a query string) sidesteps the need to do client-side seeking.
I also created two new pages in mediawiki to cater for the relevant input fields when tagging/annotating a video. Property:Bill & Property:Speech By
I created [[Property:Speech by]] with [[has type::page]] myself.
With the above modifications, I can add new videos and then edit them by adding tags/annotations and transcripts. A few more questions : A) I add a new video from special pages -> metavid add stream I edit the stream and add the video file, also typing in its length, i.e. 0:10:00
'Duration' listed in the 'Editing Stream:' view should be entered as total length in seconds, without punctuation.
However the video (and all videos imported) show a time length of 0:20:00. Any suggestions on how to make the video pages show the actual video length?
Append '&view=overview' (or, if using Short URLs, '?view=overview') to the URL to arrive at an "overview" of the meeting with the entire duration displayed. In this view transcripts are shown in logical "chunks" -- for example, the entire Q&A portion of a meeting would be be collapsed into a single block right after a "closing remarks" block.
To enable AJAX 'Date range' search functionality the 'date_start_time' property of the 'mv_streams' MySQL table should match the UNIX timestamp of the start of the meeting. The only way I know of entering this data is by editing the backend DB directly w/ phpMyAdmin.
B) It seems that I am not able to create a Sequence. From the demo video of metavid, It seems that by using a Sequence:ABCD you can create a sequence where videos can be edited in the timeline. How can I enable this?
Take a look at the following: http://openmeetings.org/w/js2/mwEmbed_rev115787/example_usage/
The Sequencer is more of a standalone component of mwEmbed at this point, I think?
C) In the new transcript part of editing, there seems to be an image of the person making the speech. In my case its always the missing person.jpg. How can I add my custom picture?
1) Create [[Template:Person]] with the following:
<div id="profile"><p>{{ #if: {Image:{{PAGENAME}}.jpg}| [[Image:{{PAGENAME}}.png|80px]] |}}<span>{{#external_value:title}}. {{PAGENAME}} </span></p></div>
2) Create/upload [[File:Firstname_Lastname.jpg]] and [[File:Firstname_Lastname.png]] with a 80×99 .jpg/.png images of the person. 3) In segment edit view, enter Speech by: -> Firsname Lastname such that the wikitext contains [[Speech by::Firstname Lastname]]. Do a hard page refresh as needed.
Cheers, George
I do have more questions, but I thinks the above are quite enough for now :)
Best Regards Ioannis
_______________________________________________MetaVid-l mailing list MetaVid-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/metavid-l