Moho Forum

Hey there just wondering if anybody has any strategies or links to info about refining automatic lip syncing.

I am using 6 mouths in general, the Preston Blair mouth shapes, but it still looks really terrible. I am adding effects to the audio, such as changing the sound of the voice before I import it into Moho, though I can still clearly understand what is being said. Perhaps this is effecting the accuracy? Any info from somebody with experience who has struggled with automatic lip syncing would be really really awesome.

Cheers,

Matt

Just to clarify - are you talking about importing an audio file into your project and then going to the switch layer's settings and using that audio file in the "select audio sync source" drop down? If so, I don't think you'll ever get a super accurate lip sync from that method, as I believe it really only looks at the volume of the audio file. Sadly, nothing really beats the process of lip syncing manually.

You may want to look into this script to speed up your manual lip sync: https://www.lostmarble.com/forum/viewto ... 12&t=34078

And if you're interested in automating that process a bit more, take a look at this thread: viewtopic.php?f=15&p=193367#p193367

I'd suggest before automating lipsync, first animate a shot manually to understand the style you want to automate. As a bonus you'll have a reference point to compare the automated animation to your own to figure out what's going wrong.

I always prefer to manually do the lip synch... unless you've ' a lot' of dialogue and no enough time to do it properly.
What I find better to automate is the complimentary gross motions of the head and spine using the audio file, nut do the actual lip sync by hand, it's quick and clean.
You can always exaggerate and pose out body the poses later, but it adds a layer of life and breath to the character for free.
See this video about it.

https://www.youtube.com/embed/6dcZXYlLK4A

[youtube]6dcZXYlLK4A[/youtube]

If you mean linking it to the audio source and letting Moho guess... I like this method!

I suggest skipping Preston Blair and just making a series of mouth shapes from closed to wide open. Then, at the top end, add a couple of special shapes if your dialogue needs it.

My mouth switch layer, from the bottom layer up, would be something like - closed, slight open, medium open, big open, really big open... then maybe an "oo" shape or a snarl or teeth, or something else that adds some character and helps with the sounds identified in the script.

I'll let Moho do it's guessing, then it needs to be cleaned up manually. Usually Moho puts in more switches that is really necessary, so I go through and simplify - if Moho has it go from mouth 1, to 3 to 5, for example, I might take out mouth 3 and let the interpolation handle getting from 1 to 5.

When I need one of the special shapes, I just choose it from the switch layer and let interpolation handle the transition.

I also usually have to slide some of the mouth switches around... the start of a phrase sometimes has the mouth switch placed a few frames off in one direction or another.

Next step is to watch it as a whole and tweak anything that seems off.

Final step is to then animate the eyes, brows, face, etc to add character and accent. Good eye moves can help distract from the mouth. I sometimes add blinks for accents, or to indicate a change of thought.

Awesome thanks heaps everybody for the info, really appreciate it.