Generate animated human videos from images & audio with Bytedance Omni Human

Generate Animated Human Videos from Images & Audio with Bytedance Omni Human

Built with n8n + Replicate

This workflow takes an image + audio, sends them to Bytedance’s omni-human model, waits for processing, and returns a generated video.

⚡ Section 1: Start & Authenticate

• ▶️ On clicking ‘execute’ → Manual trigger to start the workflow.
• 🔑 Set API Key → Stores your Replicate API key so future requests are authorized.

Benefit: Secures your API credentials in one place for easy re-use.

🛠️ Section 2: Send Video Generation Request

• 📡 Create Prediction → Makes a POST request to Replicate with:

  • image: input image URL
  • audio: input audio URL
  • Model version: 7ec44f5140c7338b3496cbf99ee8ea391a4bc18ff5d1677a146dfc936a91f65b

Benefit: Combines visual and audio inputs to start AI-powered video generation.

🔍 Section 3: Track the Prediction

• 📦 Extract Prediction ID → Saves predictionId, status, and predictionUrl for polling.

• ⏳ Wait → Pauses 2 seconds between status checks (to avoid spamming the API).

• 🔁 Check Prediction Status → Queries Replicate using the stored prediction URL.

• ✅ Check If Complete → Branches:

  • If status = succeeded → move forward.
  • Else → loop back to Wait until completion.

Benefit: Ensures workflow patiently waits for the video instead of timing out.

📽️ Section 4: Process & Return Results

• 📦 Process Result → Cleans the API response, returning:

  • status
  • video_url (generated video)
  • metrics
  • created_at & completed_at
  • model: bytedance/omni-human

Benefit: Gives you a neat, structured output with the generated video link ready to use.

📊 Workflow Overview

✅ Final Benefits

• 🎬 Turns static image + audio into full AI-generated video.
• 🔑 API key stored securely in workflow.
• 🔄 Handles async generation with auto-polling.
• 📤 Outputs clean JSON with direct video link.
• 🧩 Modular — you can connect results to Slack, Gmail, or Google Drive for auto-sharing.