12 production-tested ComfyUI workflows for AI content creation — characters, product videos, music, voice cloning, and more.
Built for a real client project. Battle-tested in production. Now open-source.
| # | Workflow | Description |
|---|---|---|
| 01 | AI UGC Video + Voice Cloning | Generate product videos with AI characters and cloned voice (InfiniTalk + VibeVoice) |
| 02 | AI UGC Video + Prebuilt Audio | Same pipeline but with pre-recorded audio instead of cloning |
| 03 | Trend Copy Video (Image-Guided) | Copy any trending video style using a reference image (Wan 2.2 Animate) |
| 04 | Trend Copy Video (Text-Only) | Generate trend-style videos from text prompts without a reference image |
| # | Workflow | Description |
|---|---|---|
| 05 | Multi-Character Group (Qwen 2511) | Generate 5 consistent characters with individual LoRAs in a single pipeline |
| 06 | Multi-Character Regional Prompting | Place multiple characters in one image using regional attention control |
| 07 | Image Realism Enhancer | Add photorealistic skin, lighting, and detail to AI-generated images |
| 08 | Product Placement (Qwen 2511) | Edit AI characters to hold or interact with real products |
| # | Workflow | Description |
|---|---|---|
| 09 | Talking Avatar (InfiniTalk) | Generate lip-synced talking character videos from image + audio |
| 10 | AI Music Generation (AceStep) | Create original songs with lyrics, genre, and instrument control |
| # | Workflow | Description |
|---|---|---|
| 11 | Flux2 Dataset + Auto Captions | Generate training images with automatic Florence2 captioning |
| 12 | Flux2 Batch Dataset (Prompt Queue) | Batch-generate dataset images using prompt queue automation |
Small single-purpose workflows in the bonus-utilities folder:
- Audio Separator — Split vocals from instrumentals
- Caption Generator — Auto-caption image datasets (WD14 Tagger)
- Voice Design (Qwen TTS) — Design custom voices with text-to-speech
- Video Combine — Merge multiple video clips with audio
- Singing Audio Merge — Combine and amplify audio tracks
- Open ComfyUI
- Drag and drop any
workflow.jsoninto the canvas - Install missing custom nodes (ComfyUI Manager recommended)
- Download required models (listed in each workflow's README)
- Run
- ComfyUI (latest)
- ComfyUI Manager (for auto-installing custom nodes)
- GPU with 12GB+ VRAM (24GB+ recommended for video workflows)
MIT License — use these however you want.
If any of these workflows save you time, drop a star. It helps others find them.