I imported a video rendered from Adobe Premiere (mp4 H264 1920x1080) and the audio from de video is imported 1/2 frame out of sync.
I reimported the video into premiere and into Davinci and the audio is in sync.
The project frame rate and the sample frame rate are the same, 24fps 48.000khz.
I exported OMF file from premiere, imported into Nuendo, and the audio is in sync
What am I doing wrong importing video with audio into Nuendo?
I’m using the last release of premiere and nuendo in W10.
Thanks for your kind help.
Edit: I imported the same premiere timeline rendered in prores with pcm audio with no issues.
I will have to write this down somewhere, and just copy/Paste.
Audio embedded in a video file is NEVER in sync.
Embedded audio can never-ever be used as sync reference.
Reason why we (Post) always need an AAF/OMF and BITC.
The only thing you can so is convert your video to ProRes and/or Dndx with a tool that is compensating for that offset that intentionaly is introduced to a videofile.
ER Media Toolkit is your (only) friend.
But … the AAF with 2pop and video with white Flashframe should be your onnly reference.
Ignore (and run away from) the embedded audio.
Always.
Estrange answer.
I’ve been doing audio post for film for a long time and never detected this issue. This is my first project with Nuendo 12.0.30. The issue could be with Premiere, may be, is the first export with the last version too.
My media editor friend is Adobe CC. I use Nuendo because is infinitely better than Audition.
Regards,
Lluís
‘Auto fix audio offset issue’ (enabled by default).
Due to extensive video decoding time, some highly compressed and complex codecs are designed to offset their included audio streams to ‘force’ video/audio playback synchronization while played in their native form via various media players in various devices, However, converting those codecs into none compressed form may introduce the opposite result!
Therefore, we designed a unique smart algorithm that detects those potential offset problems in both input and target (audio/video), and only if necessary adjust intelligently the audio stream offset to keep it ‘in sync’ with the target video, resulting in less than 1ms (more precisely, less than 45 samples) sync accuracy between video and audio streams with most popular mp4 containers, but never ever risking adjusting into negative values that result in audio material beginning cutoffs (like you may experience with most other conversion algorithms that try to deal with this issue). The fix applies to AAC and MP3 audio streams, all other audio codecs are unaffected in this process. If you choose to disable this option the audio will not be analyzed and processed with this unique algorithm and shall be passed on exactly as it originally created in the source (compressed) media form. This method is unique for ER Media ToolKit in both Video conversion and ‘Extract Audio to Wave’.
To answer your question: Steinberg does not convert the videofile in any way.
Nuendo/Cubase uses whatever shitty thing you throw into it.
As you have tested for yourself, a Prores ot Dndx will not show that problem.
And there are plenty more reasons to NOT use a heavily compressed videofile, so I will not get into that.
I have only two DAW, Audition and Nuendo.
Audition imports both, mp4 h264 and mov prores in perfect sync. Comparing both audio tracks, sems to be sample accurate.
In Nuendo, mp4 h264 is 1/2 frame delayed.
Of course I prefer to work with prores file, it plays smooth.
And of course I can re-render the h264 file to prores.
But I think Nuendo can do it better if Audition can.
I use this track only to check I’ve not lost any dialog while searching the best audio take.
And it’s helpful to check all dialog cuts are in perfect sync.
Allow me to chime in, @LluisV.
You’re assuming that Adobe software knows how to handle h264/h265 audio offsets correctly, and that is a wrong assumption. It can’t.
What it can do, is fix its own doing.
Hence, if an advanced video coding (AVC=h264) or an HEVC (aka H265) was created in Adobe software to begin with, then importing it back will allow Adobe software to correct the offset it (itself) created adequately, but unfortunately, they (and others), always assume their AVC offset is the only one that exists, so if you try to import an AVC (H264) or HEVC (H265) that was created with another tool that uses different decoder, you’ll see that it fails colossally.
The same thing with Protools, when you import an AVC that was created with Avid software, it will import correctly, but try another media that was created with a different AVC decoder and you’ll see that it does a poor job as well.
Here you can find a test example that you can use to understand the issue:
Download it and import all the included media files into your favorite DAW or NLVE.
Align them all on a single timeline.
Zoom in as much as possible, and look for audio offsets between the media files.
Note that the “Bullet-Train-6s-exported-as-uncompressed-PCM.wav” is the correct reference that everything else should align to.
Most likely that the audio in “Bullet-Train-6s-exported-as-h264.mp4” will not be handled correctly by your software.
In fact, in my experience, Adobe Premiere adjusts the audio into negative values that result in audio material beginning cutoffs, which is much worse than doing nothing!
That is why I designed the unique smart algorithm in ER Media ToolKit that attempts to fix any audio offset issue, something no other tool (that I know of) does.
Hi Sagi,
Thanks for your very kind explanation and for the test files.
You mean, then, all media players play h264 video with audio slightly out of sync always.
@LluisV , no, they don’t. It’s much more complicated than that.
The offset is applied so that the media will play back in-sync when used in conventional media players such as VLC, WMP (Windows Media Player), QT, MPC, Smartphones, and essentially any player that plays the media as-is. NLVEs and DAWs software are very different in this regard because they refer to each stream within the container by itself, not playing the container as a whole (mp4, mov, mxf, mkv, wmv, are all containers that can host lots of various stream such as audio, video, text, etc.)
Furthermore, DAWs are taking it one step further by physically splitting the audio streams from within the container, converting it into PCM audio (non-compressed form), so the user will have full control over the audio, but they mostly convert it as-is, not taking into account the embedded offset that was created for conventional media players.
In the ER Media ToolKit, I gave users the option to split the audio and video streams before they import them into their DAW or NLVE, and the audio goes through my unique smart algorithm that fixes its offset and at the same time converts it into a non-compressed form that you can drag into your work station without any additional conversions.
The other option is to convert the AVC (H264) media into ProRes or DNxHD without splitting the audio and video streams. In such a case, under the hood, the ER software does the same thing as I described above but does it into a temp audio file that is added (in its PCM form) to the output video file once the video conversion is done.
You can try the example folder I shared above and see what you get out of ER Media ToolKit by comparing the file named “Bullet-Train-6s-mp4-converted-to-ProRes-with-ER-Media.mov” with the original that been used to create it named “Bullet-Train-6s-exported-as-h264.mp4”
Asking for ProRes or DNxHD is the safest approach, but sending an mp4 is much more convenient for the sender, so as someone who’s not new to audio post production workflow, I don’t expect video editors to prepare ProRes for each cut, I just take whatever they supply and put everything through the ProResER Pro app before I import it into Nuendo. Easy for everyone.
