search

hongzimao / pensieve

Neural Adaptive Video Streaming with Pensieve (SIGCOMM '17)
http://web.mit.edu/pensieve/
MIT License
517 stars 280 forks source link

Video encoding settings #92

Open bekiroguzhan opened 4 years ago

bekiroguzhan commented 4 years ago

Can you please give some details about what are the settings for the encoded videos?

What encoder was used? Is the encoding constant bitrate? Is there any buffer allowed while encoding? Is frame rate set to constant?

hongzimao commented 4 years ago

This project is about adapting bitrates at video streaming time -- we assume the video has already been encoded and the bitrate levels are known. This is the case for on-demand video streaming -- such as users watching videos on Youtube or Netflix.

The encoding is not constant bitrate. Each chunk of the video is encoded in different bitrates -- that's how we adapt the bitrate while the user is watching a video.

Buffer in our context refers to the playback buffer. That's a few seconds of videos that's already downloaded but not yet played (e.g., shown as the grey bar ahead of current video time in Youtube) I don't know the "buffer" you refer to when encoding a video.

Frame rate I believe is constant.

Hope these help!

bekiroguzhan commented 4 years ago

Hi Hongzi,

The question is for the original encoding settings for each source video before you convert them to DASH content(chunks). While encoding, you can set up constant bitrate but even in that case there can be some buffer so that the average quality is preserved for the selected bitrate. This can be set if you use ffmpeg encoder for h264 codec. You can also set no buffer but it may affect the quality of the encoded video. Can you please also let me know if the video's raw file (the original file not the dash content) is available at any public repository?

Best

On Thu, Mar 12, 2020 at 11:48 AM Hongzi Mao notifications@github.com wrote:

This project is about adapting bitrates at video streaming time -- we assume the video has already been encoded and the bitrate levels are known. This is the case for on-demand video streaming -- such as users watching videos on Youtube or Netflix.

The encoding is not constant bitrate. Each chunk of the video is encoded in different bitrates -- that's how we adapt the bitrate while the user is watching a video.

Buffer in our context refers to the playback buffer. That's a few seconds of videos that's already downloaded but not yet played (e.g., shown as the grey bar ahead of current video time in Youtube) I don't know the "buffer" you refer to when encoding a video.

Frame rate I believe is constant.

Hope these help!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/hongzimao/pensieve/issues/92#issuecomment-598262419, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADALXKVGQPMTVDJE2SLKCXLRHD76VANCNFSM4LEOQ6HQ .

hyperonex commented 4 years ago

@bekiroguzhan, I have seen your build issue of AMuSt-ndnSIM in the simulator issues page. I'm sorry to write to you here as the issue thread is closed there. I have resolved the building issue if you are still interested on that you can write to me using my email address achraf_gyahoo.fr. Regards. Hyperonex