首页快捷创作模型 API价格作品广场AI应用工作流模型库
wan2.2_t2v_low_14B_fp8_scaled
更新：2025-12-19 20:43｜发布时间：2025-12-19 20:43
555
0
4
wan2.2_t2v_low_14B_fp8_scaled - 1
头像
Gugugaga
类型
UNet
基础模型
Wan Video
发布时间
2025-12-19 20:43
文件签名
eed677458bf584f6e30fe3c8103fd739f45f07d7f959d95682ba162ca6913205
wan2.2_t2v_low_14B_fp8_scaled_wan2.2_t2v_low_14B_fp8_scaled.safetensors
13.31 GB
Reprinted from author: Tongyi Wanxiang

Wan 2.2

is a newly released next-generation multimodal generative model. Adopting an innovative MoE (Mixture of Experts) architecture, the model consists of high-noise and low-noise expert models partitioned by denoising timesteps, enabling the generation of higher-quality video content. Wan 2.2 features three core capabilities: Cinematic Aesthetics Control, deeply integrating professional film industry aesthetic standards to support multi-dimensional visual control over lighting, color, composition, etc.; Large-Scale Complex Motion, effortlessly reproducing various complex motions to enhance motion fluidity and controllability; Precise Semantic Adherence, excelling in complex scene understanding and multi-object generation to better reflect user creative intent. The model supports multiple generation modes including Text-to-Video and Image-to-Video, suitable for content creation, artistic creation, educational training, and other application scenarios.

Model Highlights

  • Cinematic Aesthetics Control: Professional camera language, supporting multi-dimensional visual control over lighting, color, composition, and more
  • Large-Scale Complex Motion: Smoothly reproduces various complex motions, enhancing motion controllability and naturalness
  • Precise Semantic Adherence: Complex scene understanding, multi-object generation, better reflecting creative intent
  • Efficient Compression Technology: High compression ratio VAE in the 5B version, memory optimization, supporting hybrid training

Wan 2.2 Open-Source Model Versions The Wan 2.2 series models are released under the Apache 2.0 open-source license, supporting commercial use. The Apache 2.0 license allows you to freely use, modify, and distribute these models, including for commercial purposes, provided you retain the original copyright notice and license text.

Model TypeModel NameParametersMain FeaturesModel Repository
Hybrid ModelWan2.2-TI2V-5B5BThe hybrid version supports both text-to-video and image-to-video, satisfying two core task requirements in a single model🤗 Wan2.2-TI2V-5B
Image-to-VideoWan2.2-I2V-A14B14BConverts static images into dynamic videos while maintaining content consistency and smooth motion transitions🤗 Wan2.2-I2V-A14B
Text-to-VideoWan2.2-T2V-A14B14BGenerates high-quality videos based on text descriptions, featuring cinematic aesthetic control and precise semantic alignment🤗 Wan2.2-T2V-A14B

Corresponding models:


Operating Steps:

  1. Ensure the first Load Diffusion Model node loads the wan2.2_t2v_high_noise_14B_fp8_scaled.safetensors model.
  2. Ensure the second Load Diffusion Model node loads the wan2.2_t2v_low_noise_14B_fp8_scaled.safetensors model.
  3. Ensure the Load CLIP node loads the umt5_xxl_fp8_e4m3fn_scaled.safetensors model.
  4. Ensure the Load VAE node loads the wan_2.1_vae.safetensors model.
  5. (Optional) In the EmptyHunyuanLatentVideo node, you can adjust the size settings and total video frame count (length).
  6. (Optional) If you need to modify the prompts (positive and negative), please edit them in the CLIP Text Encoder node from Step 5.
  7. Click the Run button or use the shortcut Ctrl(cmd) + Enter to execute video generation.




作品

暂无作品