Increase the density of keyframes

[Stasik0]

Hi, I try to cut some words out of some speeches. The problem is, that the video fragments become very short and do not always start with a keyframe, which causes ugly artefacts. How can I fix this problem?

[Khaver]

Convert the whole video to a lossless format first then edit that. When done, convert the edited version to any format you want.

[guns1inger]

Or export the audio and edit is separately in an audio editor, where frame have no influence. Once done, import the audio to replace the original.

[lordsmurf]

Or export the audio and edit is separately in an audio editor, where frame have no influence. Once done, import the audio to replace the original.

Yeah, do this.

The difficulty, however, will be dependent on the format of video you have.

[Stasik0]

Well, exporting only audio is not an alternative. I do need audio and video parts.

What lossless format should I take? MPEG1? What audio, wav. I use mencoder.

Best Regards

[manono]

Well, exporting only audio is not an alternative. I do need audio and video parts.

Did you not read the part about importing the audio after the edit? guns1inger's suggestion is far and away the best so far.

[Stasik0]

Yes, but my problems is with the video artefacts, not audio. Audio Track is ok.

[guns1inger]

Perhaps you can more clearly explain what you are trying to do. If you are trying to gram the video of a single word, then you will have to re-encode the video if not cutting on keyframes, unless you use either uncompressed video (very large files) or a format that uses intra-frame compression only (DV being the most compact example).

AVI Demux should be able to do what you need. If you are removing short snippets and keeping the rest, you can use it's built-in smart encoding to reduce the amount of re-encoding you have to do.

Recent builds of virtualdub also support smart encoding, although the range of formats that virtualdub can handle is less than AVI Demux.

[Stasik0]

That is not precise what i want to do.

I want to cut a speech of a politician into many short clips (one clip for each word).

The problem is that in theses very short monies, the key-frames are not on the beginning.

These clips are often shorter than 1 sec. I'd like to do it from command line under linux. ffmpeg or mencoder for example.

Best Ragards

[guns1inger]

The whole command-line under linux thing is kinda important . . . . . . just saying.

You can get avisynth working under wine. That is probably your best solution. Frame accurate, script driven, feed to your encoder to encode. You may have to write some pearl or similar to create your avisynth scripts for you.

[Stasik0]

will it place keyframe to the beginngin of the part which is cut off?

[guns1inger]

Keyframes are created when you encode. You will get one for the first frame, and subsequent keyframes will be created depending on the interval you select when you configure your encoder, and the codec being use. For example, DV has a keyframe every frame.