Talking

[Jkoseattle]

My characters need to talk, which basically means moving around my Mouth Shape switch layer for as long as I need. I'm not mimicking actual syllables here, just making it look like their mouths are moving. I created an Action to do this, but when inserting it into an animation, the action goes for as long as it goes and that's that. I need more granularity on the length of mouth movement, as it coordinates with other character movement. (For example, they talk for 3 seconds and then lift their arms. The talking must be 3 seconds, but my talking action is two seconds long.) Here are the options I have thought of and why they are problematic. Suggestions?

1. Make the Talking action short enough, i.e. one second, so that I can just repeat it and at least have granularity of one second. PROBLEM: The mouth movement repeating that quickly looks too phony and obviously looped. Needs to be longer than that.

2. Make the Talking action as long as I want, and when using, only insert a copy (not a reference) and remove the keyframes I don't need that make it too long. PROBLEM: Seems inelegant somehow, and vaguely feels like I'm going to run into some problem somewhere.

3. Make several Talking actions, i.e. half second, 1 second, 2 seconds, 3 seconds, and choose the appropriate action in the animation. PROBLEM: Can't see any downsides here, if nothing better is suggested this is how I'll do it, but I don't love it.

3. Forget about actions and just manually animate the mouth switch when I need it. PROBLEM: I will be retired by the time I finish any animations.

[MrMiracle77]

Moho does have a feature that allows a Switch layer to cycle through shapes based on the relative amplitude of an audio sample. This is found on the 'Switch' tab of the Switch layer properties. You can select an audio layer in the timeline and after hitting 'ok' or 'apply' Moho will assign Switch changes to the timeline based on amplitude, with the highest position assigned to the loudest sounds and the bottom position assigned to the quietest. It isn't perfect, but it eliminates the appearance of 'looping'. You can also give multiple mouth shapes the same layer name and Moho will randomly pick one from the set for that amplitude.

You can also use Papagoyo, a free software that generates a 'dat' file that can be handed off to Moho for lipsync.

[Greenlaw]

BTW, Moho 13.5.5 has improvements for getting proper mouth shapes when using the automated option. Here's the description from their website...

Automatic Lipsync is back with some improvements:

Select a switch layer with mouth visemes, open the Layer Settings dialog, go to Switch layer settings, and select an audio layer to sync to -> automatic lip sync

When you set up a mouth switch layer, you can create extra viseme layers for more variation. For example, besides the "AI" and "MBP" layers, you can also have "AI 2", "AI 3", "MBP 2", "MBP 3", as many as you want. When lip syncing, Moho will pick a sub-variation randomly. These variants should still be proper visemes, but they can be slightly different for more variation

I never tested this because I prefer to keyframe my mouth shapes, but if you use it I'm curious to see how well it works.

[Jkoseattle]

So cool. I'm looking forward to messing around with lip syncing, but in this case I should have specified that I'm not using audio at all, but rather word balloons. Although I may add VO later, we'll see how quickly I get this thing done if I have time for that.

So, in this case the question is more about how to adjust the duration of actions. After typing my initial question up though, I decided that having several similar actions of differing lengths was the easiest way to move forward.

[slowtiger]

Uhm, you could use whatever speech audio to create the keys, but keep it silent later on.

Or put your mouth shapes into a switch layer, create an action with all contents, and control this with a smart bone set to noisy, this will create random mouths.

[Greenlaw]

Ah, I see. You just want generic 'mouth flapping'.

That reminds me: slowtiger posted an interesting idea the other day, a technique for randomizing the switch layer which I think can be adapted for this situation.

Just create a Smart Bone Dial (SBD) to open and close the mouth using the Switch Layer, then when you want the mouth to start yapping, create a key and set the interpolation to Noisy. When you want the mouth to stop, dial the SBD to the closed mouth postion and set the interpolation to Step. This will keep the mouth closed until you set the next Noisy key.

Or, if you're using vectors only, make an SBD that basically just opens and closes the mouth drawing and use the same keyframing method described above.

If you want to also automate the keyframing part, make a regular action that pastes the 'on' and 'off' keys for you in the timeline, and then slide the keys for the timing and duration you want. After that, you can just copy/paste those keys on the timeline to repeat as needed.

I haven't tried any of this myself so there may be steps I'm overlooking, but in theory anyway I think this should do what you're asking.

[MrMiracle77]

So cool. I'm looking forward to messing around with lip syncing, but in this case I should have specified that I'm not using audio at all, but rather word balloons. Although I may add VO later, we'll see how quickly I get this thing done if I have time for that.

So, in this case the question is more about how to adjust the duration of actions. After typing my initial question up though, I decided that having several similar actions of differing lengths was the easiest way to move forward.

Moho can record an audio layer. Just talk gibberish for as long as you need and then use that sample to reference the mouth shape.

[Jkoseattle]

ONG so glad I asked. I forgot all about the a Noisy interpolation. Good idea. Also gibberish might be good, though honestly if I’m going that’s why not record myself doing the actual text so that the syllables will look right and just not use the audio for public consumption.