---
tags:
- diffusion-single-file
- comfyui
license: apache-2.0
base_model:
- Wan-AI/Wan2.1-VACE-14B
- Wan-AI/Wan2.1-VACE-1.3B
---
Better fp8 scaled models (when measured against fp16) based on quantization code from https://github.com/Tencent-Hunyuan/HunyuanVideo/blob/main/hyvideo/modules/fp8_optimization.py

Can be used with: https://github.com/kijai/ComfyUI-WanVideoWrapper (latest version) and ComfyUI native WanVideo nodes.

14B-T2V comparison test without LoRAs, 25 steps, 832x480x81
---

<video controls autoplay width=50% src=https://cdn-uploads.huggingface.co/production/uploads/63297908f0b2fc94904a65b8/DwlAGbj20it1unZW54NDC.mp4></video>

2.2 A14B-T2V test
---

<video controls autoplay width=50% src=https://cdn-uploads.huggingface.co/production/uploads/63297908f0b2fc94904a65b8/6A_AZ7GN_uxeRH0vwsWkH.mp4></video>

<video controls autoplay width=50% src=https://cdn-uploads.huggingface.co/production/uploads/63297908f0b2fc94904a65b8/GpuqQ4YwoR3kjxkhuvP8P.mp4></video>

The e5m2 marked as v2 is the one uploaded here and these are all scaled even if I forgot to label properly.