CompuServe Messages

I recently completed an animated 'cyber-ette' for an interactive kiosk application for AT&T. The kiosk application opens with an animation of a talking woman's head. The tools I used were 3D Studio on a 486DX2-66, and a Mac 840AV/Spigot Power AV running Photoshop and Premiere. We worked with a very limited budget and very short deadline, so a custom model and extensive motion data were out of the question. I needed to create a woman's head and synch her to a pre-recorded narration in about a day and a half, here's what I did… Step 1. Modeling the head. I started with a stock 3D mesh and added details like ear rings and more intricate clothing. I also took the original texture map and 'tweaked' it up, making the skin tones and shading brighter and more appealing (my wife's advice was very helpful here). Step 2. Creating key frames. Referring to a couple of my favorite books on animation, I took my completed mesh and created 5 main facial positions (8 or 10 would have been better, but I had a short deadline). I mostly used 3DS's bending tools to accomplish this. Step 3. Getting the reference motion info. I was unable to video tape the voice talent while she read the script (the recording was made in Colorado and I am in California). So I set up a video camera to tape my face as I read the script. I tried to read the script as closely to the voice talent's read as possible. If I were to do this again, I'd be sure to videotape the original talent, as re-reading the script in perfect sync is virtually impossible… I next captured the lip sync video on my Mac. The Spigot Power AV only captures at 30 fps, and my final animation was to play at 15 fps, so I converted the lip synch Quicktime movie to 15 fps in Premiere. Step 4. Morphing keyframes. I then setup my 3D Studio for the total number of frames in the animation. With each of my 'cyber-ettes' facial positions set as 'morph targets' I advanced my Quicktime movie frame by frame, and decided which facial position to use and set it as a morph position. Because I only had 5 facial positions, I had to make some compromises. If I were to do it again, I'd create about 12 or 15 facial expressions, adding a slight grin, eye blinks, etc. This would have greatly added to the realism of the final animation. Step 5. Adding head motion. Once the facial motions were completed, I then studied the Quicktime movie for head motions… slight movements upward during inhales, a slight downward here and there. I tried to mymic the video in my animation. I'd guess that I spent about an 8 to 10 hour day to actually complete the animation (plus another half day experimenting before I decided on this approach). Working with my Mac -and- my PC at the same time is often very effective, and in this case allowed me to view the video while simultaneously adjusting my animation and model to match. My client was very happy with the final outcome, and I've already created a French language version (fortunately I speak French as well!). Hope this helps, I can email you stills if you like (so long as you include credits for AT&T and M!ND GAMES). Paul Unterweiser, M!ND GAMES new media, Carlsbad, CA