Sony a7S and 120fps video

This is a partial bug report, partial “How To” workaround.

We had a production shoot with a Sony a7S camera at 120fps. At this speed, it only records 1280x720 resolution but that’s ok.

When the clips get imported into Cortex, they are interpreted as 23.976fps. This is great - they were purposely shot off speed to get a slow mo effect. But the problem I had was the the clips kept crashing Cohog Renderer. The problem is not with the video, it’s actually more with the audio. The audio played back at 1/5 speed (120fps/24fps = 5) which made the audio slow and stuttery. Since these clips were supposed to be MOS, I selected the clips in Cortex then right-clicked and selected “Remove Audio”. So now Cortex would only work with the video of these clips and ignore the audio. The clips converted just fine and Cohog did not crash.

I’ve sent these clips on to @dave for his team to take a look.

We had a similar issue in the past with offspeed GoPro footage.

In that case, I think we solved it by actually treating the footage as 48 fps. In this case, since there is no actual SMPTE timecode for 120 fps, we have to assign the timecode rate something valid (like your project framerate, in this case, 23.976).

But then the audio apparently isn’t handled correctly.

We’ll take a closer look.

In the mean time, your workaround is correct:
Remove audio to get footage like this to transcode…

@peter is going to take a look at this one in the near future.

I took a look and another workaround is to change the play speed to 120 fps. We are discussing how to handle this better because when we change the frame rate to the project frame rate we are also changing the length of the clip. It fails to render because its trying to get audio for frames that don’t actually exist.

The problem with changing the play speed is that you will end up skipping 4/5 frames in your render, which is probably not what you want to do…

@peter - If I read that right, are you suggesting changing the playback speed to 120fps. This would make the footage play back in real time, correct? And it would, as @dave suggested, throw away most of the footage and would defeat the purpose of shooting 120fps to begin with.

What if, as the rendering chugging along and suddenly is missing some audio, it just duplicates the last audio chunk? There would probably need to be checks to make sure A) the audio will pick up in a bit (ie the audio hasn’t fully stopped or reached EOF) and B) that the clip is some oddball like offspeed or the playback speed has been manually changed.

We decided to allow clips to be imported with what we call user defined framerates. This now means that the timecode will count to 120 in this case. We are leaving it up to the user to decide how they want to handle it after import. The recommended workflow would be to add these clips to a record timecode reel.

1 Like

This is fixed in v1.5.3 b5238.

This topic was automatically closed after 2 days. New replies are no longer allowed.