Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,3 +1,306 @@
|
|
1 |
-
|
|
|
|
|
|
|
2 |
import os
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#!/usr/bin/env python
|
2 |
+
|
3 |
+
from __future__ import annotations
|
4 |
+
|
5 |
import os
|
6 |
+
import random
|
7 |
+
|
8 |
+
import gradio as gr
|
9 |
+
import numpy as np
|
10 |
+
import PIL.Image
|
11 |
+
import torch
|
12 |
+
from diffusers import AutoencoderKL, DiffusionPipeline
|
13 |
+
|
14 |
+
DESCRIPTION = "# SD-XL"
|
15 |
+
if not torch.cuda.is_available():
|
16 |
+
DESCRIPTION += "\n<p>Running on CPU 🥶 This demo does not work on CPU.</p>"
|
17 |
+
|
18 |
+
MAX_SEED = np.iinfo(np.int32).max
|
19 |
+
CACHE_EXAMPLES = torch.cuda.is_available() and os.getenv("CACHE_EXAMPLES") == "1"
|
20 |
+
MAX_IMAGE_SIZE = int(os.getenv("MAX_IMAGE_SIZE", "1024"))
|
21 |
+
USE_TORCH_COMPILE = os.getenv("USE_TORCH_COMPILE", "0") == "1"
|
22 |
+
ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD", "0") == "1"
|
23 |
+
ENABLE_REFINER = os.getenv("ENABLE_REFINER", "0") == "1"
|
24 |
+
|
25 |
+
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
26 |
+
if torch.cuda.is_available():
|
27 |
+
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
|
28 |
+
pipe = DiffusionPipeline.from_pretrained(
|
29 |
+
"segmind/SSD-1B",
|
30 |
+
vae=vae,
|
31 |
+
torch_dtype=torch.float16,
|
32 |
+
use_safetensors=True,
|
33 |
+
variant="fp16",
|
34 |
+
)
|
35 |
+
if ENABLE_REFINER:
|
36 |
+
refiner = DiffusionPipeline.from_pretrained(
|
37 |
+
"stabilityai/stable-diffusion-xl-refiner-1.0",
|
38 |
+
vae=vae,
|
39 |
+
torch_dtype=torch.float16,
|
40 |
+
use_safetensors=True,
|
41 |
+
variant="fp16",
|
42 |
+
)
|
43 |
+
|
44 |
+
if ENABLE_CPU_OFFLOAD:
|
45 |
+
pipe.enable_model_cpu_offload()
|
46 |
+
if ENABLE_REFINER:
|
47 |
+
refiner.enable_model_cpu_offload()
|
48 |
+
else:
|
49 |
+
pipe.to(device)
|
50 |
+
if ENABLE_REFINER:
|
51 |
+
refiner.to(device)
|
52 |
+
|
53 |
+
if USE_TORCH_COMPILE:
|
54 |
+
pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
|
55 |
+
if ENABLE_REFINER:
|
56 |
+
refiner.unet = torch.compile(refiner.unet, mode="reduce-overhead", fullgraph=True)
|
57 |
+
|
58 |
+
|
59 |
+
def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
|
60 |
+
if randomize_seed:
|
61 |
+
seed = random.randint(0, MAX_SEED)
|
62 |
+
return seed
|
63 |
+
|
64 |
+
|
65 |
+
def generate(
|
66 |
+
prompt: str,
|
67 |
+
negative_prompt: str = "",
|
68 |
+
prompt_2: str = "",
|
69 |
+
negative_prompt_2: str = "",
|
70 |
+
use_negative_prompt: bool = False,
|
71 |
+
use_prompt_2: bool = False,
|
72 |
+
use_negative_prompt_2: bool = False,
|
73 |
+
seed: int = 0,
|
74 |
+
width: int = 1024,
|
75 |
+
height: int = 1024,
|
76 |
+
guidance_scale_base: float = 5.0,
|
77 |
+
guidance_scale_refiner: float = 5.0,
|
78 |
+
num_inference_steps_base: int = 25,
|
79 |
+
num_inference_steps_refiner: int = 25,
|
80 |
+
apply_refiner: bool = False,
|
81 |
+
) -> PIL.Image.Image:
|
82 |
+
generator = torch.Generator().manual_seed(seed)
|
83 |
+
|
84 |
+
if not use_negative_prompt:
|
85 |
+
negative_prompt = None # type: ignore
|
86 |
+
if not use_prompt_2:
|
87 |
+
prompt_2 = None # type: ignore
|
88 |
+
if not use_negative_prompt_2:
|
89 |
+
negative_prompt_2 = None # type: ignore
|
90 |
+
|
91 |
+
if not apply_refiner:
|
92 |
+
return pipe(
|
93 |
+
prompt=prompt,
|
94 |
+
negative_prompt=negative_prompt,
|
95 |
+
prompt_2=prompt_2,
|
96 |
+
negative_prompt_2=negative_prompt_2,
|
97 |
+
width=width,
|
98 |
+
height=height,
|
99 |
+
guidance_scale=guidance_scale_base,
|
100 |
+
num_inference_steps=num_inference_steps_base,
|
101 |
+
generator=generator,
|
102 |
+
output_type="pil",
|
103 |
+
).images[0]
|
104 |
+
else:
|
105 |
+
latents = pipe(
|
106 |
+
prompt=prompt,
|
107 |
+
negative_prompt=negative_prompt,
|
108 |
+
prompt_2=prompt_2,
|
109 |
+
negative_prompt_2=negative_prompt_2,
|
110 |
+
width=width,
|
111 |
+
height=height,
|
112 |
+
guidance_scale=guidance_scale_base,
|
113 |
+
num_inference_steps=num_inference_steps_base,
|
114 |
+
generator=generator,
|
115 |
+
output_type="latent",
|
116 |
+
).images
|
117 |
+
image = refiner(
|
118 |
+
prompt=prompt,
|
119 |
+
negative_prompt=negative_prompt,
|
120 |
+
prompt_2=prompt_2,
|
121 |
+
negative_prompt_2=negative_prompt_2,
|
122 |
+
guidance_scale=guidance_scale_refiner,
|
123 |
+
num_inference_steps=num_inference_steps_refiner,
|
124 |
+
image=latents,
|
125 |
+
generator=generator,
|
126 |
+
).images[0]
|
127 |
+
return image
|
128 |
+
|
129 |
+
|
130 |
+
examples = [
|
131 |
+
"Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
|
132 |
+
"An astronaut riding a green horse",
|
133 |
+
]
|
134 |
+
|
135 |
+
with gr.Blocks(css="style.css") as demo:
|
136 |
+
gr.Markdown(DESCRIPTION)
|
137 |
+
gr.DuplicateButton(
|
138 |
+
value="Duplicate Space for private use",
|
139 |
+
elem_id="duplicate-button",
|
140 |
+
visible=os.getenv("SHOW_DUPLICATE_BUTTON") == "1",
|
141 |
+
)
|
142 |
+
with gr.Group():
|
143 |
+
with gr.Row():
|
144 |
+
prompt = gr.Text(
|
145 |
+
label="Prompt",
|
146 |
+
show_label=False,
|
147 |
+
max_lines=1,
|
148 |
+
placeholder="Enter your prompt",
|
149 |
+
container=False,
|
150 |
+
)
|
151 |
+
run_button = gr.Button("Run", scale=0)
|
152 |
+
result = gr.Image(label="Result", show_label=False)
|
153 |
+
with gr.Accordion("Advanced options", open=False):
|
154 |
+
with gr.Row():
|
155 |
+
use_negative_prompt = gr.Checkbox(label="Use negative prompt", value=False)
|
156 |
+
use_prompt_2 = gr.Checkbox(label="Use prompt 2", value=False)
|
157 |
+
use_negative_prompt_2 = gr.Checkbox(label="Use negative prompt 2", value=False)
|
158 |
+
negative_prompt = gr.Text(
|
159 |
+
label="Negative prompt",
|
160 |
+
max_lines=1,
|
161 |
+
placeholder="Enter a negative prompt",
|
162 |
+
visible=False,
|
163 |
+
)
|
164 |
+
prompt_2 = gr.Text(
|
165 |
+
label="Prompt 2",
|
166 |
+
max_lines=1,
|
167 |
+
placeholder="Enter your prompt",
|
168 |
+
visible=False,
|
169 |
+
)
|
170 |
+
negative_prompt_2 = gr.Text(
|
171 |
+
label="Negative prompt 2",
|
172 |
+
max_lines=1,
|
173 |
+
placeholder="Enter a negative prompt",
|
174 |
+
visible=False,
|
175 |
+
)
|
176 |
+
|
177 |
+
seed = gr.Slider(
|
178 |
+
label="Seed",
|
179 |
+
minimum=0,
|
180 |
+
maximum=MAX_SEED,
|
181 |
+
step=1,
|
182 |
+
value=0,
|
183 |
+
)
|
184 |
+
randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
|
185 |
+
with gr.Row():
|
186 |
+
width = gr.Slider(
|
187 |
+
label="Width",
|
188 |
+
minimum=256,
|
189 |
+
maximum=MAX_IMAGE_SIZE,
|
190 |
+
step=32,
|
191 |
+
value=1024,
|
192 |
+
)
|
193 |
+
height = gr.Slider(
|
194 |
+
label="Height",
|
195 |
+
minimum=256,
|
196 |
+
maximum=MAX_IMAGE_SIZE,
|
197 |
+
step=32,
|
198 |
+
value=1024,
|
199 |
+
)
|
200 |
+
apply_refiner = gr.Checkbox(label="Apply refiner", value=False, visible=ENABLE_REFINER)
|
201 |
+
with gr.Row():
|
202 |
+
guidance_scale_base = gr.Slider(
|
203 |
+
label="Guidance scale for base",
|
204 |
+
minimum=1,
|
205 |
+
maximum=20,
|
206 |
+
step=0.1,
|
207 |
+
value=5.0,
|
208 |
+
)
|
209 |
+
num_inference_steps_base = gr.Slider(
|
210 |
+
label="Number of inference steps for base",
|
211 |
+
minimum=10,
|
212 |
+
maximum=100,
|
213 |
+
step=1,
|
214 |
+
value=25,
|
215 |
+
)
|
216 |
+
with gr.Row(visible=False) as refiner_params:
|
217 |
+
guidance_scale_refiner = gr.Slider(
|
218 |
+
label="Guidance scale for refiner",
|
219 |
+
minimum=1,
|
220 |
+
maximum=20,
|
221 |
+
step=0.1,
|
222 |
+
value=5.0,
|
223 |
+
)
|
224 |
+
num_inference_steps_refiner = gr.Slider(
|
225 |
+
label="Number of inference steps for refiner",
|
226 |
+
minimum=10,
|
227 |
+
maximum=100,
|
228 |
+
step=1,
|
229 |
+
value=25,
|
230 |
+
)
|
231 |
+
|
232 |
+
gr.Examples(
|
233 |
+
examples=examples,
|
234 |
+
inputs=prompt,
|
235 |
+
outputs=result,
|
236 |
+
fn=generate,
|
237 |
+
cache_examples=CACHE_EXAMPLES,
|
238 |
+
)
|
239 |
+
|
240 |
+
use_negative_prompt.change(
|
241 |
+
fn=lambda x: gr.update(visible=x),
|
242 |
+
inputs=use_negative_prompt,
|
243 |
+
outputs=negative_prompt,
|
244 |
+
queue=False,
|
245 |
+
api_name=False,
|
246 |
+
)
|
247 |
+
use_prompt_2.change(
|
248 |
+
fn=lambda x: gr.update(visible=x),
|
249 |
+
inputs=use_prompt_2,
|
250 |
+
outputs=prompt_2,
|
251 |
+
queue=False,
|
252 |
+
api_name=False,
|
253 |
+
)
|
254 |
+
use_negative_prompt_2.change(
|
255 |
+
fn=lambda x: gr.update(visible=x),
|
256 |
+
inputs=use_negative_prompt_2,
|
257 |
+
outputs=negative_prompt_2,
|
258 |
+
queue=False,
|
259 |
+
api_name=False,
|
260 |
+
)
|
261 |
+
apply_refiner.change(
|
262 |
+
fn=lambda x: gr.update(visible=x),
|
263 |
+
inputs=apply_refiner,
|
264 |
+
outputs=refiner_params,
|
265 |
+
queue=False,
|
266 |
+
api_name=False,
|
267 |
+
)
|
268 |
+
|
269 |
+
gr.on(
|
270 |
+
triggers=[
|
271 |
+
prompt.submit,
|
272 |
+
negative_prompt.submit,
|
273 |
+
prompt_2.submit,
|
274 |
+
negative_prompt_2.submit,
|
275 |
+
run_button.click,
|
276 |
+
],
|
277 |
+
fn=randomize_seed_fn,
|
278 |
+
inputs=[seed, randomize_seed],
|
279 |
+
outputs=seed,
|
280 |
+
queue=False,
|
281 |
+
api_name=False,
|
282 |
+
).then(
|
283 |
+
fn=generate,
|
284 |
+
inputs=[
|
285 |
+
prompt,
|
286 |
+
negative_prompt,
|
287 |
+
prompt_2,
|
288 |
+
negative_prompt_2,
|
289 |
+
use_negative_prompt,
|
290 |
+
use_prompt_2,
|
291 |
+
use_negative_prompt_2,
|
292 |
+
seed,
|
293 |
+
width,
|
294 |
+
height,
|
295 |
+
guidance_scale_base,
|
296 |
+
guidance_scale_refiner,
|
297 |
+
num_inference_steps_base,
|
298 |
+
num_inference_steps_refiner,
|
299 |
+
apply_refiner,
|
300 |
+
],
|
301 |
+
outputs=result,
|
302 |
+
api_name="run",
|
303 |
+
)
|
304 |
+
|
305 |
+
if __name__ == "__main__":
|
306 |
+
demo.queue(max_size=20).launch()
|