Compute head movement from speech function in 23.0 after (2024 not fixed either)

[C27618681pmp5]

This problem has been bothering me since version 23.0, the video is version 23.6.

When I imported the mp3 and used the Compute head movement from speeche function, the character moved like this.

and adjusting the face's Pose-to-Pose Movement parameter doesn't smooth out the movement.
How to set it to smooth the movement?

[David Arbor]

I moved this post to the Bugs forum. I also just noticed this issue yesterday.

[C27618681pmp5]

Finally, I hopefully this will be fixed in the next version. Thank you.

[axelj5676904]

Got that problem too. Also in version 24.0

[C27618681pmp5]

Could you do something to help? @oksamurai 

[oksamurai]

Thanks for the bug - I noticed this recently too. We're looking into it.

[axelj5676904]

Any update?

[charlesb60477526]

Any updates on this? It's making Character Animator basically unuable for my needs.

[Sillvi36577260818j]

still not fixed still not  fixed!

[axelj5676904]

An update was installed today and the bug is still there. This massive error has not been fixed for well over a year or two.

Does anyone know how I can still get the old version V22? This feature worked without any problems in V22. I have given up hope that the bug will be fixed, @oksamurai 

[oksamurai]

Unfortunately yes, I haven't heard of any progress on this bug. I'm really sorry; I know it's frustrating to have something working and become a critical part of your workflow and then break like this. My limited understanding is that the bug is more complex than it may seem on the surface, but I don't know the specifics under the hood.

The workaround I would recommend would be what we show in the recent Express Animation tutorial. Basically, your triggered Replays become your different poses and you swap between them to achieve the pose-to-pose look. If you only wanted the head you could make a set of Face-only replays and then do the other things like arms and lip sync separately. The blend handles between poses would equate to the duration / smoothness of the transition between poses. This system is essentially what we do when creating Express characters with "automatic" animation. I know it's not the same, as pose-to-pose just worked in the background without needing this more manual setup, but it should ultimately achieve a similar effect.

Again, I'm sorry this bug has lingered for over a year without a clear fix. If anything changes I will udpate this thread, but my recommendation for the forseeable future would be to explore different workflows.