Something I added back in v0.02, but never did a proper videos (since /mlp/ doesn’t let you post videos with sound) :
Realtime Lip-syncing, using Oculus’ library.
It’s imperfect, but considering those models don’t have proper visemes blendshapes it’s probably the best approximation I will get.
Twilight’s voice machine-learning-created by the Pony Voice Preservation Project.
For now it’s not tied to anything in-game, but in the future it should be used for dialogues (if a voice generator is ready by then) … or at least with emotes.
Also, new design for Tw-AI-light : loading BlackJack’s legs the same way I do with Pillow Case’s hoodie, and hiding the body’s “real” legs by mes the corresponding bones’ links.
It’s a 4chan thread so the links don’t stay up forever, but here is the current one :
https://boards.4channel.org/mlp/thread/34514258
Other samples of what they made recently :
https://clyp.it/z3qpzzie
https://vocaroo.com/i/s13Fa6h24wbh
https://clyp.it/ezboiyrf
links? that voice was… mumbly, but kind of uncanny.
A group of Anons on 4chan using machine-learning to create an AI capable of imitating the show’s voices.
what’s the pony voice preservation project?
Edited