As well, to make these characters move and Just a Dog Mom with tattoos flower shirt, you have to do motion capture of a real person, body and face, and map it onto the digital model of the person or another creature, like Gollum in LOTR. This is all 1-1, as the CG character can only say or act the lines that the person has recorded. Sometimes you get near the end of production, and you realize that you need another 5 lines of dialog, along with the accompanying facial motion capture. It is a pain to bring the actor back in if they even are available, set up all the motion capture rig the same, record then new dialog and mocap, edit it and it is expensive.
Just a Dog Mom with tattoos flower shirt, hoodie, sweater and v-neck t-shirt
Offical Just a Dog Mom with tattoos flower shirt
So, we have been working on methods that use deep learning to train on the initial motion capture data of the Just a Dog Mom with tattoos flower shirt, then extrapolate that model to a few more lines of dialog, for which the deep learning model can generate the appropriate facial motion data. This is more profound than it first seems. Now imagine this taken to the extreme: put a subject in a complete performance capture environment, which records every detail of what they say, including their words, intonation, speech patterns, voice stress, and it also does full motion capture on body and face, recording lip-sync, facial expressions, and even micro-expressions, eye movement, pupil dilation, sweating.