Can anyone recommend an editor to do this? Basically I want to combine pieces of two videos but keep the original encoding in the output. I know you can edit a single file with Avidemux and not change the original encoding but that's not adequate in my case.
Thanks for any help.
+ Reply to Thread
Results 1 to 8 of 8
Did you mean composite multiple tracks and layers ? If so , it's not possible without re-encoding
If you meant end to end, ie. appending videos as-is in a single track, but +/- multiple audio tracks then it's possible
I may have misstated my requirement; I just mean two videos where each has just one video and audio track. For example, insert a short clip in the middle of a long one. I already use Blender for this kind of stuff but why can't it be combined without re-encoding if both files were encoded identically? It's basically just splitting and joining multiple times which could probably be done using a cumbersome script.
Basically you're asking to append 3 videos -
This is discussed frequently in other threads.
The short version is you need to cut on keyframes ( GOP boundaries) then join them. If you need frame accuracy (if where you want to cut does not occur on a keyframe), then you need to re-encode the few frames around a cutsite up to the next keyframe - this is known as "smart rendering" . Some programs that can do this are videoredo, tmpgenc smart renderer, solveigmm video splitter
What type of video ?
There are NLE's that can do this, mix/match and only render frames that involve effects/ transitions , other frames are stream copied or passed through untouched - but it's more source format dependent
e.g most professional NLE' s can do it with I-frame formats such as DV, DNxHR, DNxHD, AVC-Intra. Vegas can do it with some times of MPEG2 including long GOP XDCam . FCP/FCPX can do it with prores, same with the mac version of Premiere.
If you're using generic long GOP AVC or HEVC, none will proper have smart rendering support , or they will be very buggy. e.g. PowerDirector claims to be able to do it with SRVT, but it's very buggy and frames get glitched or mixed up
why can't it be combined without re-encoding if both files were encoded identically?
1. Segment Must start with an 'I' frame and end with a 'P' frame if a segment is joined at the end.
2. Timing at the end must be consistent, i.e. The last few frames of 0.04ms for 25fps should have 0.04ms between each otherwise a glitch is introduced
as 0.04 - 0.04 - 0.08. Videos like these are caused by many programs (FFMpeg or FFMpeg based) that cut at a 'B' frame and introduce a 'P' frame with
the duration time equal to the missing 'B' frames in the original before the actual original ending 'P' frame.
[Attachment 54920 - Click to enlarge]
[Attachment 54921 - Click to enlarge]
3. Audio time of all segments must be almost equal to the video or a 'Freeze' frame with be shown at the join point.
[Attachment 54922 - Click to enlarge]