Douyin UGC Ad Video Creator

The Specific Problem It Solves

Creating a UGC-style video ad for Douyin (TikTok) e-commerce often involves coordinating product visuals, influencer performance, and voiceover scripts across disjointed stages. The manual process demands high consistency, pacing, and platform adaptation, risking quality issues if stages are misaligned. This skill standardizes these steps, leveraging an AI toolchain to reduce friction between raw product assets and a publishable video.

Core Capabilities & Key Steps

The skill follows a clear standard pipeline to transform inputs into output. It requires an inspectable product image, confirmed product information (including factual claims and evidence), and an influencer direction. The core process is:
- Input Inspection & Modeling: Analyze the product image's MIME type, dimensions, and other properties; log all visible features and merchant claims. This step defines a key "product moment" (e.g., unboxing or demo) as an anchor for creation.
- Model Selection & First-Frame Generation: Read available image_to_image, text_to_speech, and image_to_video model cards via beatra.models.list. Generate an influencer-style first frame using beatra.images.transform, where the product is the visual focus, blended with natural handheld camera feel and scene interaction.
- Voiceover Script & Speech Synthesis: Write a concise 8-12 second voiceover script based on the product and approved claims. Select a suitable voice via beatra.voices.list and generate audio using text_to_speech.
- Staged Confirmation & Paid Production: The process is split into a preparation stage (first frame, voiceover) and a video stage. Each stage requires presenting a detailed plan (e.g., canvas, script, model, cost) to the user for approval, generating a stable client_request_id. Production is executed solely via the bundled MCP client—never using host connectors or REST fallbacks.
- Video Synthesis & Delivery: Use beatra.videos.animate to combine the first frame and voiceover audio into the final vertical video. Delivery includes checking product visibility, voiceover clarity, and platform compatibility, followed by a focused next-step suggestion.

Applicability Boundaries & Important Notes

This skill is suitable for scenarios requiring an AI influencer to appear in a short vertical video, explaining and demonstrating a product, such as Douyin UGC ads, unboxing recommendations, etc. Use other tools for:
- Pure product shorts without influencer appearance → product-video-studio
- Existing video footage needing voiceover insertion → talking-avatar-video
- Voiceover audio only → short-form-video-audio
- Multi-camera or non-influencer-led videos → beatra-ai-video-studio

Key considerations include:
- All factual claims must trace back to user-provided evidence and must not imply unverified personal experience.
- Each paid call (e.g., modifying the first frame, rewriting the script) constitutes new paid work and requires re-confirmation.
- Product appearance, labels, materials, and other key visual information must be retained, but post-delivery only checks actual results, not guaranteed consistency.
- Updates are checked silently at runtime, but failures do not block current task execution; control via --auto off.