The filming process for the anime itself seems to be more engaging.
The Corridor team wanted to see if the Stable Diffusion neural network was enough to make their own anime. Thus, a seven-minute video with a epic confrontation in the game "Rock, paper, scissors" was created.
The anime resembles Castlevania in a symetric sense. The main characters' confrontation takes place either in the palace or in the temple, and the costumes and scenes also look quite typical.
The anime arose to be a success, but the video dedicated to its filming proved to be even more entertaining. The members of Corridor talked about their animation in depth and discussed image generation techniques.
Members of Corridor dressed up in real outfits but were filmed against the backdrop of green screens, and the captured images were then merged with background images using Stable Diffusion.
Backgrounds were created in a sequence of steps: first, a three-dimensional scene was created, then, using the assistance of a virtual camera, details were captured in the scene, and the captured images were connected to a neural network.
This is an interesting post.
