[FFmpeg-user] Why was patch declined: Modified force_key_frames option to accept frame numbers?

Ted Park kumowoon1025 at gmail.com
Sun Feb 17 03:54:56 EET 2019


> I calculated the I-Frame positions from a already segmented video with a
> different resolution.
> When I disabled scenedetection and placed them manually while encoding
> using force_key_frame it worked!
> BUT, I miss 8 I-Frames because they are not packed into seperate chunks.
Okay, but I frames might not be keyframes. And scenecut might place I frames that are not idr frames. Maybe that’s the reason for the differences.

> BUT, I miss 8 I-Frames because they are not packed into seperate chunks.
> If I now insert all I-Frames (obtained from the video with different
> resolution), the alignment is displaced again. :(

When you say you insert I-frames, you mean you are forcing keyframes right? Are you also telling the segmenter to split at those timestamps?

> Is there a way to select the I-Frame where the video gets split into chunks?
Depends on the muxer you are using.

> Is there a limit for the minimum filesize for a chunk?
Probably, depends on the format.

> Can I split on each I-frame?
You can split anywhere, but the results might not be very useful depending on what you plan to do with it.


More information about the ffmpeg-user mailing list