Listing Room Video Pack

Problem

Listing teams often receive room photos and need turn them into deliverable short clips without manual labeling, prompt assembly, and repeated generation submissions. This skill focuses on room-by-room delivery: each photo becomes its own clip, preserves source aspect and first-frame relationship, and avoids merging listings into one unbounded edit.

How It Works And Boundaries

The main path uses the packaged scripts/mcp_client.py to inspect image MIME, dimensions, aspect, byte size, and alpha channel. If room names are missing, it surfaces an understanding estimate before calling beatra.images.understand to label photos. It then admits image_to_video, confirms a supported 2–15 second duration for beatra.videos.animate, submits one motion per room, and does not attach driving_audio. Optional narration uses beatra.speech.synthesize; audio files are delivered alongside clips and not mixed into video. Talking-head openings or closings require licensed broker portraits and scripts, plus separate confirmation of voice rights, speech synthesis, and video entry.

Note: it does not perform splicing, sequencing, or editing. Use product-video-studio for product photos and image-to-motion for generic stills. Each paid stage is confirmed separately; on insufficient balance, top up and replay the same frozen payload.