Description

A regular video suddenly gains a little dog busy dancing. A coffee cup rim, windowsill, or steps in a travel shot can all become its tiny dance floor. Provide a video and your own transparent animated stickers, and this Skill helps decide where the dog stands, how large it should be, and when to switch characters—adding a playful contrast of “soulful dancing” to scenes of everyday life. This is Dancing Dog in video form: a 2D cartoon character jumps into a real-world scene, with its feet landing on visible objects. It suits daily-life recordings, travel Vlogs, café visits, pet videos, and everyday fun, and can also use your own characters. • Ground it in the scene: Review the video shot by shot, choose real support points such as tabletops, cup rims, windowsills, pool edges, or steps, and adjust the size based on distance. • Follow the rhythm: By default, keep the character appearing throughout the video. When suitable spots are available, arrange 1–3 characters at once, switch to a new group every four beats, and preserve the original audio. • Add natural accents: Try to avoid faces, hands, subtitles, and key actions. You can also add characters only during amusing pauses. You provide: an MP4/MOV video and transparent GIF, animated WebP, or PNG assets that you have permission to use. The animated dance comes from the GIF or animated WebP itself; PNG is a static sticker that can perform a subtle beat-synced motion. To rotate different characters, provide multiple assets. When the environment supports execution, delivery includes a finished MP4, placement and rhythm notes, and layout materials that can be adjusted further. You do not need to specify coordinates manually. 【Please read before using】 This Skill includes an AI placement workflow and a local compositing script. It requires an environment that can read video, review extracted frames, and run Python 3.11+, FFmpeg, and the relevant dependencies. Full video compositing in YouMind has not yet been verified. If the environment cannot run the script, the Skill will explain the limitation and provide the layout plus instructions for running it externally. AI determines support points by viewing the footage, while motion is interpolated between keyframes for the specified objects. General-purpose automatic tracking, semantic segmentation, and complex foreground-occlusion reconstruction are not included. The finished video should be checked for foot contact, subject avoidance, and shot transitions. This Skill does not include a sticker library or music. The examples include local compositing samples and original effect demonstrations; none are complete YouMind execution screenshots.