Waveform's phase issues
-
- KVRian
- 500 posts since 3 Dec, 2021
Been having a bit of a play around.
So there's a couple of things that have bugged me about Waveform.
1. It doesn't null. When you render anything a tiny (tiny!) phase shift is introduced. You can see it in the video. Now if you're doing sound design and layering with percussive elements and you're rendering down one layer but not the others then this is having an effect, especially on the transients. It's only small but it accumulates with multiple renders.
2. Sometimes I swear that when I have layered hits on different tracks, like snare layers etc, the playback is fractionally inconsistent. Again sounds a bit phasey? Or am I going mad? (I've really asked myself that many times)
What I've done here:
Got a few kicks in a track as per normal. Rendered the track to a new track. Removed the auto pitch etc on the rendered .wav to reduce variables. The I've inverted the polarity on the volume and pan plugin on the 2nd track. A perfect copy of track 1 would now null to silence (it doesn't, I know this already) So now I set the polarity back to normal, reverse the rendered .wav clip and render that 2nd track again, so this time being in reverse, the phase shift should cancel out. Reverse this render back the right way, flip the polarity again and now it should null with my original track. It does. Kind of.
So here I have two tracks that should null, but depending on where I start the playback, they don't always, as you can hear. Waveform doesn't seem to be syncing the two tracks perfectly with every pass.
Does this matter? I'm not sure. It's noticeable, because I've noticed it and thought I was going mad. If I'm in loop mode and really trying to fine tune things am I hearing the in phase or slightly out of phase version? Option B is always on the table (Option B is I might have completely screwed something up and have no idea what I'm doing)
-
- KVRist
- 238 posts since 24 Sep, 2005
You may well be mad, like me, yet either way, I could duplicat this effect on my system. First ran the null test with the samples on two tracks before rendering to make sure those did null, which they did perfectly. Rendered the samples and the render does not null with the source samples. It is indeed on the leading transients that it is not nulling.