mirror of
https://github.com/modelscope/DiffSynth-Studio.git
synced 2026-03-18 22:08:13 +00:00
32 lines
1.8 KiB
Python
32 lines
1.8 KiB
Python
import torch
|
||
from PIL import Image
|
||
from diffsynth.utils.data import save_video, VideoData
|
||
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
|
||
from modelscope import dataset_snapshot_download
|
||
|
||
|
||
pipe = WanVideoPipeline.from_pretrained(
|
||
torch_dtype=torch.bfloat16,
|
||
device="cuda",
|
||
model_configs=[
|
||
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
|
||
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
|
||
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
|
||
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
|
||
],
|
||
)
|
||
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-14B-Control_lora/epoch-4.safetensors", alpha=1)
|
||
|
||
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
|
||
video = [video[i] for i in range(81)]
|
||
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
|
||
|
||
# Control video
|
||
video = pipe(
|
||
prompt="from sunset to night, a small town, light, house, river",
|
||
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",
|
||
control_video=video, reference_image=reference_image,
|
||
seed=1, tiled=True
|
||
)
|
||
save_video(video, "video_Wan2.1-Fun-V1.1-14B-Control.mp4", fps=15, quality=5)
|