@KyleSygma Those are my initial thoughts. The nice part of sharing the idea is that other can question and add input, like you did, thanks. It makes easier when the time comes to start coding

RT @jonstephens85: I am super excited to be rebooting the Computer Vision Decoded podcast with my cohost, @JaredHeinly! Here is a preview s…

@jonstephens85 @ClassyDogFilms I knew you would share valuable info. Thanks a lot @jonstephens85 🙏

What if I use an autoencoder to create a latent space and use that latent space to drive the movement, just like it happens with reallusion accupose? I think it can work, but I got a problem, no free time to test the idea 😣 maybe until next month I find a moment to test.

@olli_pelz @animturnbull Wow, great animation. It's so well done, looks like it's simple to do. I'm almost sure it's not as simple as it seems.

@DaluzLuciano LOL. That is really funny, as I don't use that addon anymore. Didn't like the results 😅

@DaluzLuciano I see that using SD as the final pass will make the character extremely photo real. I tested something like that at first, but didn't like the path I was going.

@olli_pelz The AI part that I used was only to have more info on something, instead of spending a lot of time searching. And I can make unique characters... I think ...

@DaluzLuciano I remember those. They look great. But I think we did different things. you used something from metahuman and processed it in SD, right? Or you created the metahuman character based on the stable diffusion image?

@nickseavert Looking great 💪👏

@jrgandara Ouvi hj, acabou tao rapido, agora so no meio ou no final do ano, pra gente ouvir o Joao falandosobre os panettones 😊

Get while you can

@curtisjamesholt That was a hard one to answer without swearing

@Shateiel3D Yeah, that would be great. I downloaded and installed the free version to test, but I'm with almost no free time, couldnt test much.