WAN 2.2 SVI - Infinite Video Generation - ComfyUI One Click Installer

WAN 2.2 SVI just dropped — and this update changes everything for infinite-length video generation inside ComfyUI. With the new Stable Video Infinity (SVI) LoRAs, you can now create seamless long-form videos that extend far beyond single clips. This release builds directly on WAN 2.2's 14B model, bringing enhanced stability, temporal coherence, and improved generative transitions between scenes.

The current workflow is set up to generate 20-second videos, but it can easily be modified to add more samplers for extended durations. Each sampler acts as a chain of connected clips, blending one seamlessly into the next. The final output is automatically concatenated into a smooth, continuous video.

Preloaded Models Within the Installer (Low VRAM)

• umt5-xxl-encoder-Q5_K_M.gguf (clip folder) — Hugging Face

• wan_2.1_vae.safetensors (vae folder) — Hugging Face

• Wan2.2-I2V-A14B-LowNoise-Q3_K_S.gguf (unet folder) — Hugging Face

• Wan2.2-I2V-A14B-HighNoise-Q3_K_S.gguf (unet folder) — Hugging Face

• 2xLexicaRRDBNet_Sharp.pth upscale model — Hugging Face

• high_noise_model.safetensors Lightx2v LoRA — Hugging Face

• low_noise_model.safetensors Lightx2v LoRA — Hugging Face

• SVI_v2_PRO_Wan2.2-I2V-A14B_HIGH_lora_rank_128_fp16.safetensors — Hugging Face

• SVI_v2_PRO_Wan2.2-I2V-A14B_LOW_lora_rank_128_fp16.safetensors — Hugging Face

Standard WAN 2.2 14B diffusion models are also supported. Download them from Hugging Face. ComfyUI Manager: GitHub.

Speed

Generate 20-second 480p videos (6 steps) in under a minute per 5-second clip on an RTX 4090 (24 GB VRAM). The workflow dynamically scales with your hardware — 720p and higher frame rates (24fps+) are achievable on more powerful GPUs.

System Requirements

• Nvidia RTX 30XX / 40XX / 50XX GPU (FP16 supported)

• CUDA-compatible GPU (minimum 6 GB VRAM, 12 GB+ recommended)

• Windows OS

• Minimum 40 GB free storage

What's Included

• Portable ComfyUI Windows Installer, fully configured for WAN 2.2 multi-clip long sequences

• Automated downloads for all required models and nodes

• User-friendly workflow optimized for beginners and advanced users

• Fast Groups Bypasser support for enabling or disabling sections on the fly

Usage Notes

• You can use either the WAN 2.2 14B GGUF models or the standard WAN 2.2 14B diffusion models within the workflow.

• Toggle workflow layers with the Fast Groups Bypasser node (left sidebar).

• Write detailed, consistent prompts for each 5-second segment to ensure smooth visual transitions and thematic continuity.