Video Enhancement

Mission

Improve delivery readiness of a selected, approved parent video. Enhancement cannot fix a wrong story, identity, product label, edit, or unsafe claim; it only creates a child asset whose limitations must remain visible.

Required inputs

• selected parent video and source provenance;
• requested output resolution, format, codec, frame rate, and platform;
• desired upscale factor or enhancement objective;
• preservation requirements for grain, subtitles, audio, color, and timing; and
• rights, budget, and approval owner.

Inspect source duration, dimensions, frame rate, audio tracks, captions, and visible defects when tooling permits. If the source is not approved, route back to the generation or editing skill first.

Execute

1. Assign a stable parent ID and upload a local source if required.
2. Query the live model schema and cost. The current Topaz video-upscale request accepts an upscale factor of 1, 2, or 4; do not assume that factor equals a particular final resolution or that every model preserves audio.
3. Submit the smallest useful factor and retain prediction ID, model, source URL, and output URL.
4. Compare parent and child for fine detail, halos, ringing, faces, text, motion boundaries, flicker, color shifts, audio continuity, duration, and export metadata. Use video QA.
5. If the enhancement creates defects, return the parent as the safer option and report the tradeoff. Do not chain enhancement repeatedly without an explicit reason.

Resolve endpoint and fields through MODELS.md and muapi-video-tools.md. Keep the original parent immutable and label the output as enhanced.

Output contract

Return parent/child IDs, source and output properties, model/endpoint, factor, prediction ID, output URL, status, before/after QA, audio treatment, and recommendation. Save a receipt under .video/ only when permitted using reports-and-receipts.md.

This skill does not restore missing content, certify broadcast quality, or publish the result.