Hi. I would love to know how you obtain from a video sequence the segment sizes that you input to your application. Do you have a pipeline to directly obtain that from the dashed representations, or do you use some tool like FFmpeg on the different encodings of the video and pre-generate all the txt files for different representations?
My naive pipeline is to encode videos corresponding to different representation into mp4 files and then run ffprobe on them to get the frame size information in bytes using a command like:
ffprobe Test.mp4 -show_frames | grep pkt_size > framesizes.csv
Then depending on the frame rate and length of a segment, one knows how many frames are in a segment and adds the size of that many frames to get the size of a segment.
However, I thought you may have a more straightforward pipeline and would appreciate any hints. Thanks!
Hi. I would love to know how you obtain from a video sequence the segment sizes that you input to your application. Do you have a pipeline to directly obtain that from the dashed representations, or do you use some tool like FFmpeg on the different encodings of the video and pre-generate all the txt files for different representations?
My naive pipeline is to encode videos corresponding to different representation into mp4 files and then run ffprobe on them to get the frame size information in bytes using a command like:
ffprobe Test.mp4 -show_frames | grep pkt_size > framesizes.csvThen depending on the frame rate and length of a segment, one knows how many frames are in a segment and adds the size of that many frames to get the size of a segment.
However, I thought you may have a more straightforward pipeline and would appreciate any hints. Thanks!