Download Wan 2.2 workflow and models to generate cinematic videos in ComfyUI

Wan2.2 was released on July 28, 2025. On that day, only the nightly version of ComfyUI could display the Wan2.2 workflow in the template. You can use the "Reinstall/Update" feature provided by LM Downloader to update, or switch the ComfyUI program to the nightly version using ComfyUI Manager.

Generation Demo

You can save this original image to your computer for testing.

Wan2.2 TI2V 5B – Image and Text to Video

Wan2.2’s TI2V-5B is a compact yet highly efficient AI model, serving as the lightweight version in Alibaba’s Wan2.2 series. It can quickly generate 720P HD videos (24fps) from either text or images. Leveraging advanced compression technology, it runs smoothly on consumer-grade GPUs, delivering fast speeds and high-quality output. Supporting both text-to-video and image-to-video generation, it’s perfect for rapidly testing ideas or sparking inspiration.

The vehicle motion and camera tracking effects in this video are quite good.

While this also generates a video, the car moves backward, which is not the intended result. If the output doesn’t meet expectations, you can modify the seed and regenerate.

Wan2.2 TI2V 5B – Text-to-Video

For text-to-video generation, simply disable the input image node to create videos purely from text:

Prmopts:

A futuristic cyberpunk car speeding through a neon-lit city at night during heavy rain, captured with an FPV (First Person View) camera angle. The vehicle dynamically follows the fast-paced motion as it races along an elevated highway, moving at extreme speed with high wheel rotation, splashing water dramatically onto the wet pavement. Raindrops on the car’s surface stream backward due to the intense velocity, creating a dynamic visual effect. Neon lights reflect vividly on the glossy, rain-soaked road, enhancing the immersive and high-tech atmosphere. Glowing tire trails and futuristic energy streaks follow the motion of the car, emphasizing the sense of speed and fluidity. In the background, drones hover in the air, firing occasional red laser beams, adding tension and sci-fi action to the scene. The lighting is cinematic, with smooth motion blur, high-tech aesthetics, and a fast-paced, thrilling vibe.

Wan2.2 I2V 14B Text-to-Image

Original image:

Generation result:

English Prompts:

A beautiful Chinese cartoon-style girl in flowing Hanfu bends down gracefully to pick a pink-white lotus flower by a summer pond, then dances joyfully while holding the blossom. The camera orbits dynamically around her—starting with a top-down shot of the picking motion, transitioning smoothly to a medium eye-level view, and culminating in a 360-degree surround shot to capture her spinning dance.

The background features an ink-wash-style lotus pond with misty mountains and drifting petals, rendered in a soft palette of celadon greens and blush pinks. Subtle glowing particles enhance the dreamlike atmosphere. The character’s expressions are vivid, with her sleeves and ribbons fluttering fluidly to emphasize traditional Chinese elegance.

Chinese Prompts:

中国古风卡通少女，身着飘逸汉服，在夏日荷塘边俯身轻摘一枝粉白荷花，随后手持荷花在落英缤纷中欢快旋转起舞。镜头以[环绕运镜]动态跟随，从俯拍摘花动作开始，逐渐拉远转为平视视角，最后360度环绕展现舞蹈全景。背景需有水墨风格的荷塘、朦胧远山和飘落的花瓣，整体色调柔和雅致（青绿与浅粉为主），加入细微的光晕粒子效果增强梦幻感。角色表情生动，衣袂随动作飘动，强调古风韵律美。

Download Workflows

Note

Windows: NVIDIA GPU with 8GB+ VRAM required (CUDA drivers must be updated)
macOS: M-series chips only. We are still in the testing phase and will release a Mac-compatible workflow later.
Storage: 30GB+ free disk space minimum

Wan2.2 TI2V 5B

We tested the workflow of Wan2.2 TI2V 5B - one workflow with two modes of operation:

Image-to-Video: Enable the input image node to generate videos using both an image and text prompts.
Text-to-Video: Disable the input image node to generate videos using text prompts alone.

⏬Download Wan 2.2 TI2V 5B Workflow (for Windows)

If your GPU has limited VRAM and cannot run the original model, you can opt for the GGUF version instead.

⏬Download Wan 2.2 TI2V 5B GGUF Workflow (for Windows)

Model Storage Location

ComfyUI/
├───📂 models/
│   ├───📂 diffusion_models/
│   │   └───wan2.2_ti2v_5B_fp16.safetensors
│   ├───📂 text_encoders/
│   │   └─── umt5_xxl_fp8_e4m3fn_scaled.safetensors 
│   └───📂 vae/
│       └── wan2.2_vae.safetensors

From： Comfy-Org/Wan_2.2_ComfyUI_Repackaged

Wan2.2 I2V 14B Workflow

⏬Download Wan 2.2 I2V 14B Workflow (for Windows)

If your VRAM is limited and unable to run the original model, you can opt for the GGUF version.

⏬Download Wan 2.2 TI2V 14B GGUF Workflow (适用于Windows)

Model Storage Location

ComfyUI/
├───📂 models/
│   ├───📂 diffusion_models/
│   │   ├─── wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors or wan2.2_i2v_low_noise_14B_fp16.safetensors // Select a model according to your VRAM capacity
│   │   └─── wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors or wan2.2_i2v_high_noise_14B_fp16.safetensors // Select a model according to your VRAM capacity
│   ├───📂 text_encoders/
│   │   └─── umt5_xxl_fp8_e4m3fn_scaled.safetensors 
│   └───📂 vae/
│       └── wan_2.1_vae.safetensors

Note：vae file wan_2.1_vae.safetensors

Download Model Files

In the ComfyUI runtime window, you can find the "Model Download" feature. We have provided some commonly used models—simply click the download button to start downloading, saving you the hassle of searching for models elsewhere.

For more models, check out the following websites:

Mainland China users:

Global users:

Contact us

If you still encounter issues, please contact our technical support team. tech@daiyl.com

Download Wan 2.2 workflow and models to generate cinematic videos in ComfyUI ​

Generation Demo ​

Wan2.2 TI2V 5B – Image and Text to Video ​

Wan2.2 TI2V 5B – Text-to-Video ​

Wan2.2 I2V 14B Text-to-Image ​

Download Workflows ​

Wan2.2 TI2V 5B ​

Wan2.2 I2V 14B Workflow ​

Download Model Files ​

For more models, check out the following websites: ​

Contact us ​