[Webinar] Learn how to a build a cloud-first strategyRegister Now

  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 745
  • Last Modified:

ffmpeg accurate timing with video/audio extraction

I am trying to extract very small pieces of a .mp4 file using ffmpeg.

I am using ffmpeg at the command line for this.

I am using win32 static build, build on 3/14/2014.
(Does ffmpeg for linux work better than ffmpeg for windows?)

An example extract command is -
ffmpeg -ss 20.05 -i robert2.mp4 -t 0.1 -c copy output2.mp4

In this example I am trying to start at a position of 20.05 seconds and extract .1 seconds of video.

the problem is that ffmpeg seems to be aligning itself to certain boundaries.

So it doesn't seem to matter whether I tell it to start at 20.0 seconds or 20.5 seconds, it still seems to start at the same point.

I have also tried using frames -
ffmpeg -i robert2.mp4 -ss 14.6 -frames:v 20 output3.mp4

Using frames seems to be very accurate, but the audio and the video don't seem to sync properly.

Does anyone know how to get ffmpeg to give me very accurate small extractions?

I need to go down to < .1 second resolution.  Also, I need the audio to properly extract and sync as well.

Or is there a better command line tool for this?

I am willing to pay for a  tool (within reasonable cost) as long as I can use it at the command line and/or with scripting and as long as it is accurate and efficient.

Thanks, Kevin.
  • 2
1 Solution
Dan CraciunIT ConsultantCommented:
AFAIK, on encoding you setup key frames. Usually at 10-15 frames interval.
When you tell ffmpeg to give you something at some point in time, it will start from the nearest key frame, cause that's the first full image it has.

I think you'll need to reencode the movie and setup every key as a key frame.

The definition from Final Cut manual:
Key Frames: Key frames are available if your selected codec uses temporal compression. Most video frames in close proximity have a high percentage of visual redundancy. Compression key frames reduce data rate by only storing complete images at regular intervals or when abrupt visual shifts occur. The remaining frames only store information about the change, or delta, between themselves and the key frame. Increasing the number of frames between key frames increases the amount of compression and makes the final file size smaller.

kevinvw1Author Commented:
Yes!  It was the key frames.  I re-encoded using a key frame interval of 2 -

ffmpeg -i robert2.mp4 -g 2 robert3.mp4

The video went from 2 megs to 14 megs, but now I can accurately edit it on 2 frame intervals which is about 1/15 of second.

Thanks a million!

Dan CraciunIT ConsultantCommented:
Glad I could help!

Featured Post

Veeam and MySQL: How to Perform Backup & Recovery

MySQL and the MariaDB variant are among the most used databases in Linux environments, and many critical applications support their data on them. Watch this recorded webinar to find out how Veeam Backup & Replication allows you to get consistent backups of MySQL databases.

  • 2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now