产品图秒变电影级广告

Mobbi 的处理过程

Mobbi runs its three-keyframe food pipeline: it first generates two hand-held-tray opening-frame candidates grounded on the user's original photo — a white plate of glossy, red-hot chicken feet in a bright Nordic café mood with soft natural light and shallow depth of field. Once the user picks frame one, Mobbi uses it as the reference to generate two candidates each for the suspension frame — the plate tossed mid-air, chicken feet and soybeans radiating outward under warm golden light — and the landing-freeze frame, with the plate flat and re-plated on a spotless surface, serving a selection card at every round.

Mobbi 的处理过程

Mobbi feeds the three frames as start, middle, and end references into the video generation engine to produce the 9:16 vertical one-shot, keeping the native action sounds — the whoosh of the toss, the scattering ingredients, the landing impact. It then analyzes the footage frame by frame for rhythm and color mood, derives an upbeat BPM 118 music direction, generates the soundtrack with AI, and mixes it in at low volume with a two-second fade-out to composite the final film.