Question

3

Using ffmpeg to cut up video at multiple points

rated 0 times [ 3] [ 0] / answers: 1 / hits: 1701 / 2 Years ago, mon, july 11, 2022, 1:24:56

I have a video that is a football game. I would like to cut the time before the game, during half time and finally the end of the game.

Total Clip Length 1h 51m or 111m

Ideally the cuts and time are as follows:

+------------+-------------+---------------------+

| Start Time | Finish Time |    Clip Duration    |

+------------+-------------+---------------------+

| 1:30       |   47:30     |  47:30              |

| 53:00      |   100:00    |  47                 |

+------------+-------------+---------------------+

However my video finishes at

ffmpeg -ss 00:01:30 -i Tripod_Camera.mp4 -t 00:47:30 -ss 00:53:00 -t 00:47:00 -c copy VideoClip.mp4

Answers

Only authorized users can answer the question. Please sign in first, or register a free account.

riffnkful

Add To Favorites

Follow

Total Points: 390

Total Questions: 123

Total Answers: 110

Location: Puerto Rico

Member since Sat, Mar 13, 2021

3 Years ago

riffnkful questions

1 Internet issues after upgrading to 22.10

Thu, Nov 17, 22, 04:24, 1 Year ago

1 Problem with upgrading packages Ubuntu 20.04

Mon, Nov 28, 22, 01:40, 1 Year ago

1 Is there any way I can play Minecraft: Bedrock Edition) for free on Ubuntu?

Mon, Dec 20, 21, 10:33, 2 Years ago

1 Reduce GRUB's contdown time on Ubuntu 20.04 LTS

Sun, Feb 5, 23, 05:06, 1 Year ago

1 Why are there so many OpenBLAS packages and which one would yield fastest results?

Sun, Aug 28, 22, 14:16, 2 Years ago

View All

answered 2 Years ago rontablis · Accepted Answer

Make input.txt containing the timestamps (in seconds) to cut:

file 'input.mp4'

inpoint 90

outpoint 2850

file 'input.mp4'

inpoint 3180

outpoint 6000

Run ffmpeg to concatenate with the concat demuxer:

ffmpeg -f concat -i input.txt -c copy output.mp4

This will stream copy, so no re-encoding occurs meaning the whole process will be fast and quality will be preserved. However, cuts will be made on keyframes, so it may not be accurate enough. If greater accuracy is needed you'll have to use a much slower method such as using the (a)trim, (a)setpts, and concat filters which requires re-encoding.

If you get A/V desync then get timestamps of keyframes, use keyframe timestamps as your inpoint, but add 0.001 to each inpoint timestamp in input.txt.