DiffSynth-Studio 2.0 major update

This commit is contained in:
root
2025-12-04 16:33:07 +08:00
parent afd101f345
commit 72af7122b3
758 changed files with 26462 additions and 2221398 deletions

View File

@@ -1,19 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video
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="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="Wan2.2_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="Wan2.2_VAE.pth"),
],
tokenizer_config=ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="google/umt5-xxl/"),
)
pipe.enable_vram_management()
# Text-to-video
video = pipe(
@@ -23,7 +23,7 @@ video = pipe(
height=704, width=1248,
num_frames=121,
)
save_video(video, "video1.mp4", fps=15, quality=5)
save_video(video, "video_1_Wan2.2-TI2V-5B.mp4", fps=15, quality=5)
# Image-to-video
dataset_snapshot_download(
@@ -40,4 +40,4 @@ video = pipe(
input_image=input_image,
num_frames=121,
)
save_video(video, "video2.mp4", fps=15, quality=5)
save_video(video, "video_2_Wan2.2-TI2V-5B.mp4", fps=15, quality=5)