v2.79b Short version:
I laid out a 10 minute video in the VSE. Everything in the VSE looks and sounds right and lined up properly on playback (av sync enabled). I render out video+audio under properties. On rendered video playback (used vlc player and ffplay) the audio after 2 minutes and 8 minutes is offset from where it was in the vse playback. To the point where the lipdub itself isn't lined up right. How do I fix this?
Render settings: h.264; avi; aac 384bitrate; 1080p; 5/0.21 fps; 100% percentscale
I set my project space's fps to match my video content before doing any editing. I've never touched the fps of the project space. The edits consisted of playing with the existing video frames and importing .wav files and lining those up to said frames. A proxy was used 25% before editing, no timecode was used. However I can duplicate this problem in another project that has a timecode for some reason. I've done a 30 minute project a similar way months ago and I don't remember audio sync of the rendered product being a problem. I did it in v2.79. I've updated to 2.79b version of blender since then (mbe part of the issue mbe not).
-maybe I just got a bad render and another will fix issue?
-I've tried hitting "update the animation cache" right before rendering (under scene->audio). I'm not sure exactly what or if this does anything of value 'anymore'. Upon testing though it gave a potentially slightly clearer result than if I didn't hit it before rendering.
-I've mixdowned the audio separately (i.e. rendered just the audio). What I got was what looks like the audio alignment heard on vse playback. At least it had a different alignment than the audio+video has (see picture below). The quality of audio is also noticably better if mixdowned (best quality observed if mixdowned and "update animation cache" is hit right before rendering)
-I'm going to render audio and video separately (mixdown for audio, render for video), and recombine in ffmpeg without re-encoding to see if they do infact lineup the same way as in the vse playback if this is done so. I'm pretty sure they do as I've done a previous 30 minute project this way and didn't have this issue. Edit: THIS FAILED, msgs about lack of timecodes. there is no timecode for this project footage but im getting the same error in another project that does has a blender timecode
-I'm going to try rendering everything AGAIN in a different container, I used .avi. I will try .mkv AND I will NOT render with an audio codec (I used aac before), I will render a .wav mixdown and combine with ffmpeg and see if it works. If that doesn't work I'll try rendering as 720p instead of 1080p, since the footage used is 720p. If that doesn't work I will try rendering with a different codec other than h.264. And if THAT doesn't work I will make everything a png, render that all out, and then see make into a video using ffmpeg and the .wav mixdowned file. And if THAT doesn't work I'll play the audio and rip it with audacity record and plug it in via ffmpeg.
Top waveform is the audio+video version. Bottom waveform is the mixdowned waveform. They don't lineup. They should lineup! The top one is shorter, I guess makes sense because a rendered movie is timed according to frames rather than time, and time is how the audio is rendered. So its okay that they are different lengths....but its NOT okay that they aren't lining up everywhere inbetween. They started lined up, then as the video hit the 2 minute mark and 8 min as seen below its offset.
*When I say mixdown or mixdowned all I mean is I rendered out 'just the audio' using the properties>render>audio special menu that gives you specific options you can't get when you render video.