Motion 3, doing replacement animation
I got an interesting assignment, to create a talking ear of corn for a promo. I decided to make the corn out of clay, but not to do "claymation" on it, just add the eyes, eyebrows, and mouth as flat replacement animation. I could use some tips from anybody who's already done this. Lacking morphing software, I'll try to put up pictures later, but what I did was maybe "the hard way".
I brought all the mouth position images and eye position images into photoshop and created one massive layered photoshop file with alphas. My thinking here was that this way I can rotate and scale the master group as one unit and my individual layers never change their relative relationshiops.
I brought in an audio guide track into Motion 3.
I started with the bottom layer of the ear of corn turned on, and one basic expression of eyes looking forward, mouth closed and smiling.
I did the tracks for the eyes first. With the keyframe recorder on and the blinked/closed eyes track enabled, I keyframed eyeblinks of about 2 frames here and there along the length of the track. Then went to the eyes-open-ahead track, found where the blinks were, and keyframed the opacity of the open eyes down to zero for those same two frames. In a few spots along the track I did the same thing to switch to eyes looking off to the 3 o'clock position. Then I made an "eyebrow pass", throwing in a few raised eyebrows for 2 frames here and there. Same technique.
Then, I had to do multiple passes to spot the mouth positions. That's the hardest part.
I first went for the easier round open "O" mouth positions, playing the track and stopping everywhere I heard an "O" or "ooo" sound, and with the keyframing recorder on, cranking up the opacity of the "O" position for a couple of frames. While "in the neighborhood" of each "oh" position, I dropped down a layer and keyframed out the smile that was showing underneath for those same number of frames.
Wash hands, rinse, repeat, for each of the key basic vowel, consonant, and plosive shapes. For a day and a half.
This part was tricky because I could not find a way in Motion 3 to scrub the audio and hear the actual phonemes, so I had to be quick on start-stop playback of the guide track to locate the spot to place the mouth shapes, and my timing is not the best on the results so far. I'm close, but not always on it. I think my problem is too *many* mouth positions, seems like not enough time between the various shapes, and I may go Hanna-Barbera on this and reduce the number of mouth shapes to make it a little less kinetic looking.
I'm wondering if you guys can suggest a better or more efficient or easier workflow for this kind of thing, in Motion 3.
I can't wait to see some samples. This sounds fun.
Utah Video Productions
Check out my Motion Training DVD
Check out my Motion Tutorials
You made me a little curious, so I thought I'd see if I could streamline the process a bit.
Now I only messed with the eyes, but it was certainly easier than keyframing. First I built the left eye as two shapes: an eyeball and an iris. I grouped those 2 shapes and called it Open Eye. Then I duplicated the eyeball shape, changed the color to flesh and grouped it with a black bezier line across it like a closed eye. That group I placed below Open Eye so that it would only be visible if Open Eye were transparent.
To create the transparency I applied the Negate behavior to the Opacity of the Open Eye group and made it 4 frames (I felt any shorter didn't read as a blink). Then I just duplicated that behavior and moved it to various points on the timeline. I was finished with the blinks in about 30 seconds.
For some extra fun, I decided to play with the eyes moving. I used a combination of Randomize, Quantize, and Clamp behaviors all applied to the X position of the iris shape. After about 15 minutes of messing around with the parameters I was able to get the eyes to randomly shift. Combined with the blink, it looked really good to me. Best part was that I didn't have to do any keyframing which means I could lengthen the project to however long I need it, and not have to do anything more with the eyes.
Finally, since eyes work in pairs and blink in pairs, I simply cloned the LT EYE group and called that my RT EYE and I was done.
Here's the result:
Nice. What have you got for mouth transitions that is easy?
One Idea I had, but hadn't followed up on, was instead of making multiple layers of mouth positions in a stack, I could make a single layer with all of them on it in a grid, and could just flick from state to state by adjusting the x-y and cropping. But I don't know that that's going to be faster or easier.
I am wondering if there might be something that can be done using stroke controls. The main idea is to simplify the call-up and placement of each shape.
BTW, does Motion 4 have better audio scrubbing than 3 does? What I really need is an audio scrub as good as the one on FCP's timeline.
Failing that, I think what I will do the next time is when I create the audio guide track, I'll shoot a mouth closeup of my own mouth, and use that as a temporary reference layer. Kind of Clutch Cargo/ Sapce Angel/ Conan O'Brien, but it would only be there as a temp layer to guide the selection of the mouth shapes that would overlay it. Then, when all the cartoon mouth shapes are in place, just delete the human mouth layer.
Because it's creepy, man.
Sorry, no real tricks for the mouth. Because you really need to edit that based on the dialogue. Now, if you were going super lo-tech and only had an open/close mouth pieces, then I might suggest going with the audio parameter behavior.
If I were in your spot however, I'd be tempted to build the individual mouth pieces based on phonetics and draw them in Photoshop. then bring them in as stills into FCP and edit the mouth pieces there. It's far easier in my mind to scrub audio and cut the mouth shapes as clips next to each other than by doing some weird workaround with opacity, or cropped sequence replicators or some such. Once cut together, you can bring them into Motion to complete the animation.
Anyway, that's how I'd do it.