Hana-bi: Same image, same prompt, run through three video models at 1080p. Image: ChatGPT (keyframe written from title/thumbnail)
Cinematic video keyframe, square 1:1: fireworks bursting over a night harbour seen through a warm cafe window, silhouettes of two fictional friends at a table in soft bokeh, reflections on glass, frozen celebratory frame, no text, no logo. Square 1:1 composition.
- Wrote a single 1:1 still from the post prompt/caption/thumbnail; removed brands, logos, celebrity/IP elements and any need for a user photo
cinel made the replacements above to generate the example image.