AI enthusiast Esdrico has turned La La Land's dance scene into an anime-style animation.
Visual Arts & Animation Student and AI enthusiast Esdrico demonstrated an amazing use case for a recently-open-sourced text-to-image diffusion model Stable Diffusion. Using the AI, the author managed to recreate a dance scene from the 2016 musical La La Land and turn it into an anime-style animation. To create the animation, the artist utilized Euler Ancestral sampler and the "a film still of lalaland, artwork by studio ghibli, makoto shinkai, pixv" prompt. The result took around 30 minutes to generate and runs at a framerate of 8 FPS.
"I leave seed behavior as inter, the strength is from the init image strength. Also, I pause the process frequently to adjust the settings depending on how much is the coherence between the frames and also check the spacing," comments the creator.
And here are some more of Esdrico's AI-powered projects:
You can learn more about the author's other projects here. Also, don't forget to join our Reddit page and our Telegram channel, follow us on Instagram and Twitter, where we share breakdowns, the latest news, awesome artworks, and more.