Spaces:
Running
on
L40S
Running
on
L40S
File size: 6,593 Bytes
3964763 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 |
# Copyright 2024-2025 The Alibaba Wan Team Authors. All rights reserved.
import argparse
import os.path as osp
import os
import sys
import warnings
import gradio as gr
warnings.filterwarnings('ignore')
# Model
sys.path.insert(0, os.path.sep.join(osp.realpath(__file__).split(os.path.sep)[:-2]))
import wan
from wan.configs import WAN_CONFIGS
from wan.utils.prompt_extend import DashScopePromptExpander, QwenPromptExpander
from wan.utils.utils import cache_video
# Global Var
prompt_expander = None
wan_t2v = None
# Button Func
def prompt_enc(prompt, tar_lang):
global prompt_expander
prompt_output = prompt_expander(prompt, tar_lang=tar_lang.lower())
if prompt_output.status == False:
return prompt
else:
return prompt_output.prompt
def t2v_generation(txt2vid_prompt, resolution, sd_steps, guide_scale,
shift_scale, seed, n_prompt):
global wan_t2v
# print(f"{txt2vid_prompt},{resolution},{sd_steps},{guide_scale},{shift_scale},{seed},{n_prompt}")
W = int(resolution.split("*")[0])
H = int(resolution.split("*")[1])
video = wan_t2v.generate(
txt2vid_prompt,
size=(W, H),
shift=shift_scale,
sampling_steps=sd_steps,
guide_scale=guide_scale,
n_prompt=n_prompt,
seed=seed,
offload_model=True)
cache_video(
tensor=video[None],
save_file="example.mp4",
fps=16,
nrow=1,
normalize=True,
value_range=(-1, 1))
return "example.mp4"
# Interface
def gradio_interface():
with gr.Blocks() as demo:
gr.Markdown("""
<div style="text-align: center; font-size: 32px; font-weight: bold; margin-bottom: 20px;">
Wan2.1 (T2V-14B)
</div>
<div style="text-align: center; font-size: 16px; font-weight: normal; margin-bottom: 20px;">
Wan: Open and Advanced Large-Scale Video Generative Models.
</div>
""")
with gr.Row():
with gr.Column():
txt2vid_prompt = gr.Textbox(
label="Prompt",
placeholder="Describe the video you want to generate",
)
tar_lang = gr.Radio(
choices=["CH", "EN"],
label="Target language of prompt enhance",
value="CH")
run_p_button = gr.Button(value="Prompt Enhance")
with gr.Accordion("Advanced Options", open=True):
resolution = gr.Dropdown(
label='Resolution(Width*Height)',
choices=[
'720*1280', '1280*720', '960*960', '1088*832',
'832*1088', '480*832', '832*480', '624*624',
'704*544', '544*704'
],
value='720*1280')
with gr.Row():
sd_steps = gr.Slider(
label="Diffusion steps",
minimum=1,
maximum=1000,
value=50,
step=1)
guide_scale = gr.Slider(
label="Guide scale",
minimum=0,
maximum=20,
value=5.0,
step=1)
with gr.Row():
shift_scale = gr.Slider(
label="Shift scale",
minimum=0,
maximum=10,
value=5.0,
step=1)
seed = gr.Slider(
label="Seed",
minimum=-1,
maximum=2147483647,
step=1,
value=-1)
n_prompt = gr.Textbox(
label="Negative Prompt",
placeholder="Describe the negative prompt you want to add"
)
run_t2v_button = gr.Button("Generate Video")
with gr.Column():
result_gallery = gr.Video(
label='Generated Video', interactive=False, height=600)
run_p_button.click(
fn=prompt_enc,
inputs=[txt2vid_prompt, tar_lang],
outputs=[txt2vid_prompt])
run_t2v_button.click(
fn=t2v_generation,
inputs=[
txt2vid_prompt, resolution, sd_steps, guide_scale, shift_scale,
seed, n_prompt
],
outputs=[result_gallery],
)
return demo
# Main
def _parse_args():
parser = argparse.ArgumentParser(
description="Generate a video from a text prompt or image using Gradio")
parser.add_argument(
"--ckpt_dir",
type=str,
default="cache",
help="The path to the checkpoint directory.")
parser.add_argument(
"--prompt_extend_method",
type=str,
default="local_qwen",
choices=["dashscope", "local_qwen"],
help="The prompt extend method to use.")
parser.add_argument(
"--prompt_extend_model",
type=str,
default=None,
help="The prompt extend model to use.")
args = parser.parse_args()
return args
if __name__ == '__main__':
args = _parse_args()
print("Step1: Init prompt_expander...", end='', flush=True)
if args.prompt_extend_method == "dashscope":
prompt_expander = DashScopePromptExpander(
model_name=args.prompt_extend_model, is_vl=False)
elif args.prompt_extend_method == "local_qwen":
prompt_expander = QwenPromptExpander(
model_name=args.prompt_extend_model, is_vl=False, device=0)
else:
raise NotImplementedError(
f"Unsupport prompt_extend_method: {args.prompt_extend_method}")
print("done", flush=True)
print("Step2: Init 14B t2v model...", end='', flush=True)
cfg = WAN_CONFIGS['t2v-14B']
wan_t2v = wan.WanT2V(
config=cfg,
checkpoint_dir=args.ckpt_dir,
device_id=0,
rank=0,
t5_fsdp=False,
dit_fsdp=False,
use_usp=False,
)
print("done", flush=True)
demo = gradio_interface()
demo.launch(server_name="0.0.0.0", share=False, server_port=7860)
|