# MiniMax-H3 Image/Text-to-Video Workflow (Easy & Advanced)
User-friendly, battle-tested ComfyUI workflows for MiniMax-H3 video generation with built-in audio creation (without external audio lipsync)—available as a fast Easy setup or a high-detail 2-Step Advanced pipeline.
### 🎤 Looking for the External Audio Lipsync Variant?
We also offer a dedicated version built specifically for **lipsyncing to your own songs or voice tracks**!
👉 **Watch the tutorial for the latest version here:**
https://civarchive.com/models/2899685/edit
### ✨ Key Features:
- **Start & End Frame Control:** Easily load a start image and an optional end frame, or toggle directly to pure Text-to-Video generation[cite: 2].
- **Integrated Audio Generation:** Sound effects, ambience, and background music are generated automatically alongside the visuals based on your prompt[cite: 2].
- **Easy & Advanced (2-Step) Modes:** Fast single-pass rendering (Easy) or multi-stage 2-step sampling with a low-res base pass followed by a latent upscale for crisp final details (Advanced)[cite: 2].
- **Speed & Quick Preview:** Ready for H3 Sparse Attention / Turbo LoRAs plus live preview feedback so you can cancel early if needed[cite: 2].
- **VRAM Optimized & Extendable:** Configured with Tiled Decode (tested on 16 GB VRAM) and end-frame extraction (Clipspace) for seamless scene extensions[cite: 2].Description
update, new features