The model was trained with the trigger phrase "steamboat willie style". I find best results from using this trigger phrase combined with "golden era animation".

Using with Diffusers

pip install git+https://github.com/huggingface/diffusers.git

import torch
from diffusers.utils import export_to_video
from diffusers import AutoencoderKLWan, WanPipeline
from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler

model_id = "Wan-AI/Wan2.1-T2V-14B-Diffusers"
vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
pipe = WanPipeline.from_pretrained(model_id, vae=vae, torch_dtype=torch.bfloat16)
pipe.scheduler = UniPCMultistepScheduler.from_config(
  pipe.scheduler.config,
  flow_shift=5.0
)
pipe.to("cuda")
pipe.load_lora_weights("benjamin-paine/steamboat-willie-14b")
pipe.enable_model_cpu_offload() # for low-vram environments

prompt = "steamboat willie style, golden era animation, an anthropomorphic cat character wearing a hat removes it and performs a courteous bow"
negative_prompt = "色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走"
output = pipe(
    prompt=prompt,
    negative_prompt=negative_prompt,
    height=720,
    width=1280,
    num_frames=81,
    guidance_scale=5.0,
    num_inference_steps=32
).frames[0]
export_to_video(output, "output.mp4", fps=16)

Download Model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.