Motion and sound, rebuilt from four sketches
aicreataro's making-of starts with four AI sketches and a track. In Blender the footage is one plate, split into grains that read shape, depth, and where motion was fastest.
The making-of is more interesting than a mystery would have been. aicreataro starts with very little: four kinds of AI video as underdrawings, plus a piece of music. Inside Blender, the source image is a single plate. It gets divided into fine grains. From the video, three measurements: the person's shape, the depth, and where the motion was fastest.
That is a technique you can repeat, which is why a making-of earns the feature. "Visualizing motion and sound" could have stayed a vibe. Instead the post names the plate, the particles, and the three signals. You can see which decision made the body appear.
The materials are few on purpose. The system in the middle is the work.
Made with
aicreataro
@aicreataro
「動きと音をビジュアライズ」のメイキング 簡単ではありますが、参考になれば嬉しいです。 ■素材は少ない ・下絵は４種類のAI動画+曲 ■Blenderの中の仕組み ・元の映像は1枚の板。それを細かい粒に分けている ・動画から測るのは3つ。人物の形、奥行き、どこが速く動いたか https://t.co/DRQz8a5t9S
2.3K likes
- likes
- 2,256
- replies
- 15
Captured when this was curated, not live. —
