Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Video Eyeball Processor and Review
Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
veyepar - Video Eyeball Processor and Review Video metadata manager - given file names of raw footage, construct cut list (with human intervention), transcode, upload, announce. based on DebConf system authors: Carl Karsten <email@example.com> Install and usage: 1. install dependancies: INSTALL.sh This installs all of the python modules in a virtualenv sandbox using virtualenvwrapper. So before running any python apps, you must activate the virtualenv with: workon veyepar. This has been added to ~/.bashrc so it should be setup. Then you must run everything with python <script name>.py Running ./<script>.py won't work. Using virtualenv makes develoment easier and reduces the amount of things that require root/sudo privs. 1.1 create the database INSTAL.sh calls rt.sh (run tests) which will create the db and an admin username: test, password abc. For production you should use a secure set of credentials so delete veyepar.db and run: veyepar/dj$ python ./manage.py syncdb This will prompt for an admin username and password. 2. create a location, client, show, and episode records in the db browse to http://localhost:8080 log in a. click "Locations" b. enter Location Name, hit save. c. click "Clients" d. enter Client Name, hit save. f. enter Show Name, select Location(s) from step b, hit Save. g. click on show link If you have many Episodes it might be worth the effort to import locaion and episodes from a json or .csv file using addeps.py. Otherwise, continue to use the browser: h. enter Episode Name, start/duration - hit Save. i. repeat for each Episode 3. veyepar.cfg: set client and show to the client and show slugs. review settings, pal/ntsc is one of the more important ones. 4. create the dirs for the .dv files: veyepar/dj/scripts$ python mkdirs.py # Makes the dir tree to put files into 4.1. Put in place the show's title.svg: mkdirs.py (previous step) drops a copy in the show dir: ~/Videos/veyepar/test_client/test_show/bling$ inkscape title.svg show stats: shows longest title and name mk_titles.py - makes title screens for all videos and uploads to datacenter Do some sanity checks: browse to the show page: check "overlapping episodes" - there shouldn't be any. show stats: grid of days by room. should look like your show schedule: crude grid of events in a room and time slot. review titles meeting announcement: meant for usergroup meetings, doesn't scale well to a large event. add_to_richard.py = creates landing page for videos email_title.py - emails url of title image and landing page to presenters for review. recording sheets: pdf to be printed and put next to each recoreding station runing DVswitch. Use DVswitch to record talks. Note the actual start/end time on the recoding sheet. 4.2. put the dv files into the dirs Use your mad file copy skilz to rsync, mv, whatever. 5. for all dv files, add records into the db: veyepar/dj/scripts$ python adddv.py Adds the .dv filenames to the raw files table 6.0 collect dv metadata 6.1 start/end veyepar/dj/scripts$ python tsdv.py - timestamp dv Gets start from one of: the file name (assumes hh_mm_ss.dv format) There is code to read the time stamp from the .dv file frames, but that doesn't work well at all. 6.2 ocr/thumbnail (don't bother, not used for anything any more) use ocr to find some text, which makes for a good thumnail. veyepar/dj/scripts$ python mkthumbs.py - find thumbs, ocr untill we find some text, saves png. 7. Associte dv files with Episodes. (This now happens automatically when you browse to an Episode that does not have a cutlist. I am leaving this here because their may be a performance hit for larger 100+ Episode shows.) Create cutlist records for each intersection of a dv file and episodes start/end time and location. veyepar/dj/scripts$ python assocdv.py: # creates cutlist items for dv files that might belong to an episode 8. make low quality versions of all the dv files so you can preview each file in a browser using <video> (not necessary, skip if you are rushing.) veyepar/dj/scripts$ python dv2ogv.py:# makes .ogv for all dv in a show 9. Post production editing 9.1 for each episode, look at what .dv files have been assigned to it. browse http://0.0.0.0:8080/main/E/1 (for a fresh system, else find your show) DV files that might belong to the talk are marked to be included. un mark [x]Apply the ones that should not be included. If there are ones missing, .. get help. use "add raw file" at bottom of page. or use fixdv.py to fix batches, or use django's admin UI to add one by one. If a dv needs junk trimmed from the start or end, put the time offset in "Cut List - Start/End" 1 or 0:0:1 means trim 1 second from start. 1:2:3 means the clip ends 1 hour 2 min 3 seconds from the start of the file, so in this case 1h 2m 3 seconds would be used. Set Episode.state to Encode (2) 10. encode files to upload enc.py:# encodes to ogg, flv, m4v, webm, mpeg adds titles too. 11. upload push.py - sends copy to private cloud stoage. (currently disabled) post_yt.py - sends to youtube (and archive and rax cuz eww, terible hack.) post_blip.py - posts to blip.tv, post_vimeo.py - posts to vimeo (not used in years, suffers from bit rot) pw.py holds user/passwords, not checked in, make one from pw_samp.py add_to_richard.py - adds video url's to landing page 12. visually review: richard, youtube, archive, title slide, encoding quality. If good, manualy move state to email. Once satified that everything is in place and no longer needs every video to be reviewed, run review_1.py to move evertyhing to next step. email_url.py - emails presenter the richard URL again, this time encouraging them to advertise the URL to their social network, or report back any problmes they see. 13. If needed, wait for presenter to reply. for instance, if they never singed a release. Move state to public when it is ready for public consumption. Either using the web UI for each one, or review_2.py for all. mk_public.py - changes state from draft/hidden to public in youtube and richard 12. tweet tweet.py HACKING ON THE CODE There are no useful docs for setting up a development environment. If you want to hack on this: Ubuntu LTS, likely any ubuntu/debian in the last 3 years. Build the encoder binary: https://github.com/CarlFK/veyepar/blob/master/setup/melt/mkmlt.sh Setup a box for production using the steps at the top of this file. The test suite is often in a broken state, tyically it will error. If it completes but shows tests failing, that is typically a real problem that should be addressed asap.