I’m currently looking to batch convert 60 large multichannel wavs at 23.976fps to 24 and 25fps for a feature deliverables to indie distributor for streaming and cinema.
There is the two schools:
a) “Resample” by changing the header info so that the file runs faster. ie. in batch plugins set the SB Resampler plugin to 48048hz for 23.976 to 24fps and 50050hz for 23.976 to 25fps. There were no presets for this and I had to make my own custom sample frequency settings within resampler. Am I doing it right? Some say changing the header info like this (apart from the slight munchinizing of the voice) can lead to some player incompatibility.
edit: I have found another way of doing this: Using no plugins but go to Options tab (of the Output/Format/Options/XML area) and checking the box "Reinterpret Header Sample Rate and setting it to 48048 or 50050hz for 24 or 25fps respectively. Is this the preferred way? I’m dubious because the wav then presents itself as a 50khz wav for example. This is surely going to bounce in QC?
b) Using the monopass internal plugin Pitch Correction or Time Stretch (which one is recommended?) to stretch to the new length causing smearing artifacts. Some still prefer this to method a) because there is no pitch change.
Which method do you use and is there a batch preset or Resampler/Time Stretch preset I am missing? You’d think there would be Pull-up and Pull-down presets in either.
Edit: Resampler’s Channel Processing settings only lists mono or stereo pairs. Why is there no multichannel option? I did’t even think it was possible to only process say L/R of a 5.1 surround file to change the frequency. So I am assuming that this is another oversight of the gui when adjusting plgins in batch processor?
You should not use the resampler from the file format dialog.
You use the Resampler plugin, which has no surround limitation.
And you’d better use the batch processor for this.
Yet when I go to the SB Resampler plugin settings from the Custom Plugin Chain box in batch processor the available Channel Processing options are still only mono and stereo pairs that make up a surround file, but no all encompassing multichannel options. Is this for a reason? I have a bunch of multichannel files loaded in the File window.
And just to confirm, using the plugin this way has been said to actually resample, using maths and interpolation, the file instead of do the pull-up workaround. Is this correct?
Nb: It would be logical to include the resampler in the monopass list instead of browsing for it (with no search options) in the Steinberg Plugins List.
Using the Resampler plugin set to 50050hz for example and output format set to 48000 (as it must present as a 48000hz file to ingestion systems) results in a file with the same length as the original. A pull-up needs to have a slightly shorter total length to be in sync, with TC in metadata re-referenced to the destination frame rate. So that wont work. It would if there was an addition option in the resampler for batch that switched to preserve time or pitch. But there isn’t that option. So this isnt a workable pull-up or pull-down workflow.
What about if it was 2 processes:
1. I had no plugins inserted, set output option to Reinterpret Header Samplerate to 50050 whilst output format is set to 48khz/24bit. Apparently this sets playback rate separately at 50050 but output container at 48000 in the metadata only.
And then bake that in to a true 48000hz playrate (so there is no incompatibility problems) in a separate batch process using the plugin set to 48000 (a flattening of all the metadata so to speak)
Some might say this 2nd bake in phase isnt necessary since the metatag trick works in all ingestion scenarios. What are your thoughts?
When source rate in video and Nuendo project is 23.976fps and I want to generate a set of 25fps audio to go with the same source video but video is re-clocked to run faster/shorter at 25fps (by using the metadata only method). This requires the duration of audio to also be shorter. ie not preserving time (or pitch). Yet the resampling plugin does not have a switch for this. It inherently preserves time.
The whole workflow should be analogous to a video tape (and it’s sound) running slightly faster thus shorter duration.
Thanks, so there is two methods. Let’s use 23.976 to 25fps as the example:
Use Resampler plugin set to 50050hz/“Best” quality. Followed by Dither plugin set to 24bit. Then in Format tab at bottom File Format>Edit Single Format> Sample Rate 48000, Bit Depth 24bit (this setting changes the sample rate stamp?)
In Options tab bottom left, enable Reinterpret Header Sample Rate to 50050 (this apparently sets how the header is read at the input, not how it is written at the output). Then Use Resampler plugin set to 48000hz/Best followed by Dither plugin set to 24bit with noise shaping off. Then also to be optionally sure, in Format tab set to 48000hz/24bit.
I have tested method 2 and it works. Length is shorter, and the pitch is higher (which is the necessary tradeoff of using varirate method instead of a granular Time Stretch)
Method 1 doesnt work. The length remains unchanged, the pitch remains unchanged.
PG please confirm that method 2 is the correct and cleanest way to achieve a pull-up from 23.976 to 25fps?