Implementation plan: back
Roadblocks and Challenges:
- JSVM implementation challenges:
- Rewriter configure file details are not available. Currently only 3 layer CGS encoding is working. Still it fails at various locations in the decoding/rewriting phases
- GoP based file data decomposition is not working. Decoder not able to recognise raw GoP file segments. Probably an implementation issue but hard to pursue due to lack of documentation
- Cannot use opensvcdecoder. Does not support rewriting. Corresponded with the developer Mickael. opensvcdecoder is also written in a crytic way which is hard to understand without documentation.
Work items
- Install OpenStack at loacalhost
- Create VM, Install Darwin Streaming Server or some other open source DASH server
- User sends request with capability (Howto find user capability?)
- Document opensvcdecoder? Currently it supports bit stream extraction in command line
- what about workflows?
- h.264 AVC multiple resolutions
- SVC to AVC rewriting: First needs to do a layer extraction using the jsvm tools
- How to split an SVC stream into single/multiple GoP small files?
- [Later] SVC layer extraction
- [other?]
Operational structure
- Openstack is up an running
- One front instance is up and running: includes the streaming server or h.264 plugin enabled webserver
- One storage instance is running. contains all the video
- front receives request for video, also included is capability info
- based on capability either:
- start streamming AVC
- initiate a distributed stream rewrite opration
- create writer vms [how many? depends on the video length?]
- chop and send gops
- recieve and put together rewritten gops
- start streaming AVC
Testing
- A simple html webpage with different resolution/quality video links
- Onclick plays in embedded player: flowplayer?
- Apple mobile devices can probably play it too.
References and Links
- h.264 plugin for web servers? [1]
- open content with audio and subs etc
