Dear Genis, On 03/18/2013 02:47 PM, Genis Musulmanbekov wrote: > I installed ffmpeg 1.1.3 and compiled it with all options.
If 'mp4' encoding works but not 'webm', you might face the need to update the 'webm' encoding parameters that are used by Invenio when calling ffmpeg. The encoding parameters are defined in /opt/invenio/etc/bibencode /encoding_profiles.json: <http://invenio-software.org/repo/invenio/tree/modules/bibencode/etc/encoding_profiles.json> See for eg. a branch that update the parameters for ffmpeg v0.6 (instead of default-supported ffmpeg v0.7): <http://invenio-software.org/repo/personal/invenio-jerome/commit/?h=bibencode-ffmpeg-0.6-encoding-profile-parameters> By default BibEncode would use for eg: $ sudo -u www-data /usr/bin/ffmpeg -y -i /opt/invenio/var/tmp-shared/video_upload_1363616410_17082_tmposjXcd -acodec libvorbis -vcodec libvpx -ab 128000 -b 1000000 -s 854x480 -ac 2 -g 320 -qmax 63 /opt/invenio/var/data/files/g0/208/content.CERN-MOVIE-2011-192;webm Running ffmpeg as the Apache user might have an impact. It is also possible that 'webm' encoding is not triggered in your specific worklflow (it should be enabled in the case of the demo Atlantis video workflow. See also CFG_BIBENCODE_TEMPLATE_BATCH_SUBMISSION and parameter 'batch_template' of the WebSubmit function 'Video_Processing' in your submission workflow, in case it has been modified). <http://invenio-software.org/repo/invenio/tree/modules/bibencode/etc/batch_template_submission.json> You can also check the logs of the created BibEncode task, which details each encoding job (there are 7 jobs per BibEncode task in default Atlantis Demo Video setup, 2x webm, 2x mp4, 3x frames extractions). A detailed email titled "Error during BibEncode batch processing" should be sent to CFG_SITE_ADMIN_EMAIL in case of failure, and another one "Problem during the processing of your video" to the submitter. The BibEncode tasks created by the BibEncode daemon will be marked as "DONE" even in case of failure. > What could be the matter? Should we use ffmpeg 0.7? That would then probably work out of the box. Best regards -- Jerome Caffaro ** CERN Document Server ** <http://cds.cern.ch/>
