Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,27 +1,32 @@
|
|
1 |
import gradio as gr
|
2 |
import numpy as np
|
3 |
import random
|
4 |
-
from diffusers import
|
5 |
import torch
|
6 |
from time import time
|
7 |
|
8 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
9 |
model_repo_id = "stabilityai/sdxl-turbo"
|
10 |
|
11 |
-
#
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
MAX_SEED = np.iinfo(np.int32).max
|
27 |
MAX_IMAGE_SIZE = 1024
|
@@ -37,15 +42,15 @@ def infer(
|
|
37 |
num_inference_steps,
|
38 |
progress=gr.Progress(track_tqdm=True),
|
39 |
):
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
|
|
44 |
|
45 |
-
|
46 |
|
47 |
-
|
48 |
-
with torch.no_grad():
|
49 |
image = pipe(
|
50 |
prompt=prompt,
|
51 |
negative_prompt=negative_prompt,
|
@@ -56,9 +61,12 @@ def infer(
|
|
56 |
generator=generator,
|
57 |
).images[0]
|
58 |
|
59 |
-
|
60 |
-
|
61 |
-
|
|
|
|
|
|
|
62 |
|
63 |
examples = [
|
64 |
["Astronaut in a jungle, cold color palette, muted colors, detailed, 8k", 1024, 1024],
|
@@ -68,50 +76,19 @@ examples = [
|
|
68 |
|
69 |
css = """
|
70 |
:root {
|
71 |
-
--primary
|
72 |
-
--
|
73 |
-
--
|
74 |
-
--
|
75 |
-
--
|
76 |
-
--primary-500: #6e6af0;
|
77 |
-
--primary-600: #5a56e4;
|
78 |
-
--primary-700: #4a46c9;
|
79 |
-
--primary-800: #3e3ba3;
|
80 |
-
--primary-900: #383682;
|
81 |
-
--primary-950: #211f4d;
|
82 |
-
|
83 |
-
--surface-0: 255 255 255;
|
84 |
-
--surface-50: 248 250 252;
|
85 |
-
--surface-100: 241 245 249;
|
86 |
-
--surface-200: 226 232 240;
|
87 |
-
--surface-300: 203 213 225;
|
88 |
-
--surface-400: 148 163 184;
|
89 |
-
--surface-500: 100 116 139;
|
90 |
-
--surface-600: 71 85 105;
|
91 |
-
--surface-700: 45 55 72;
|
92 |
-
--surface-800: 30 41 59;
|
93 |
-
--surface-900: 15 23 42;
|
94 |
-
--surface-950: 3 6 23;
|
95 |
-
|
96 |
-
--text-primary: rgb(var(--surface-900));
|
97 |
-
--text-secondary: rgb(var(--surface-600));
|
98 |
}
|
99 |
|
100 |
.dark {
|
101 |
-
--primary
|
102 |
-
--
|
103 |
-
--
|
104 |
-
--
|
105 |
-
--
|
106 |
-
--primary-500: #6e6af0;
|
107 |
-
--primary-600: #8181f8;
|
108 |
-
--primary-700: #a5a5fc;
|
109 |
-
--primary-800: #c7c7fe;
|
110 |
-
--primary-900: #e0e0ff;
|
111 |
-
--primary-950: #f0f0ff;
|
112 |
-
|
113 |
-
--text-primary: rgb(var(--surface-100));
|
114 |
-
--text-secondary: rgb(var(--surface-300));
|
115 |
}
|
116 |
|
117 |
#col-container {
|
@@ -128,28 +105,21 @@ css = """
|
|
128 |
.header h1 {
|
129 |
font-size: 2.5rem;
|
130 |
font-weight: 700;
|
131 |
-
color: var(--primary
|
132 |
margin-bottom: 10px;
|
133 |
}
|
134 |
|
135 |
.header p {
|
136 |
-
color: var(--text
|
|
|
137 |
}
|
138 |
|
139 |
.prompt-container, .result-container, .advanced-settings {
|
140 |
-
background
|
141 |
border-radius: 12px;
|
142 |
padding: 20px;
|
143 |
-
box-shadow:
|
144 |
margin-bottom: 20px;
|
145 |
-
border: 1px solid rgb(var(--surface-200));
|
146 |
-
}
|
147 |
-
|
148 |
-
.dark .prompt-container,
|
149 |
-
.dark .result-container,
|
150 |
-
.dark .advanced-settings {
|
151 |
-
background-color: rgb(var(--surface-800));
|
152 |
-
border-color: rgb(var(--surface-700));
|
153 |
}
|
154 |
|
155 |
.advanced-settings .form {
|
@@ -158,10 +128,6 @@ css = """
|
|
158 |
gap: 16px;
|
159 |
}
|
160 |
|
161 |
-
.advanced-settings .form > * {
|
162 |
-
margin-bottom: 0 !important;
|
163 |
-
}
|
164 |
-
|
165 |
.control-row {
|
166 |
display: flex;
|
167 |
gap: 10px;
|
@@ -169,13 +135,13 @@ css = """
|
|
169 |
}
|
170 |
|
171 |
.btn-primary {
|
172 |
-
background: var(--primary
|
173 |
border: none !important;
|
174 |
color: white !important;
|
175 |
}
|
176 |
|
177 |
.btn-primary:hover {
|
178 |
-
|
179 |
}
|
180 |
|
181 |
.examples {
|
@@ -186,44 +152,23 @@ css = """
|
|
186 |
}
|
187 |
|
188 |
.example-prompt {
|
189 |
-
background:
|
190 |
padding: 12px;
|
191 |
border-radius: 8px;
|
192 |
cursor: pointer;
|
193 |
transition: all 0.2s;
|
194 |
-
border: 1px solid rgb(var(--surface-200));
|
195 |
-
}
|
196 |
-
|
197 |
-
.dark .example-prompt {
|
198 |
-
background: rgb(var(--surface-700));
|
199 |
-
border-color: rgb(var(--surface-600));
|
200 |
}
|
201 |
|
202 |
.example-prompt:hover {
|
203 |
-
background: var(--primary-100);
|
204 |
transform: translateY(-2px);
|
205 |
-
|
206 |
-
}
|
207 |
-
|
208 |
-
.dark .example-prompt:hover {
|
209 |
-
background: var(--primary-800);
|
210 |
-
border-color: var(--primary-600);
|
211 |
-
}
|
212 |
-
|
213 |
-
.example-img {
|
214 |
-
width: 100%;
|
215 |
-
height: 120px;
|
216 |
-
object-fit: cover;
|
217 |
-
border-radius: 6px;
|
218 |
-
margin-bottom: 8px;
|
219 |
}
|
220 |
|
221 |
-
/* Theme toggle button */
|
222 |
.theme-toggle {
|
223 |
position: absolute;
|
224 |
top: 20px;
|
225 |
right: 20px;
|
226 |
-
background: var(--
|
227 |
border: none;
|
228 |
border-radius: 50%;
|
229 |
width: 40px;
|
@@ -232,47 +177,12 @@ css = """
|
|
232 |
align-items: center;
|
233 |
justify-content: center;
|
234 |
cursor: pointer;
|
235 |
-
transition: all 0.2s;
|
236 |
-
}
|
237 |
-
|
238 |
-
.theme-toggle:hover {
|
239 |
-
background: var(--primary-200);
|
240 |
-
}
|
241 |
-
|
242 |
-
.dark .theme-toggle {
|
243 |
-
background: var(--primary-800);
|
244 |
-
}
|
245 |
-
|
246 |
-
.dark .theme-toggle:hover {
|
247 |
-
background: var(--primary-700);
|
248 |
}
|
249 |
|
250 |
@media (max-width: 768px) {
|
251 |
.advanced-settings .form {
|
252 |
grid-template-columns: 1fr;
|
253 |
}
|
254 |
-
|
255 |
-
.theme-toggle {
|
256 |
-
top: 10px;
|
257 |
-
right: 10px;
|
258 |
-
}
|
259 |
-
}
|
260 |
-
|
261 |
-
/* Loading animation */
|
262 |
-
@keyframes spin {
|
263 |
-
0% { transform: rotate(0deg); }
|
264 |
-
100% { transform: rotate(360deg); }
|
265 |
-
}
|
266 |
-
|
267 |
-
.loading-spinner {
|
268 |
-
display: inline-block;
|
269 |
-
width: 20px;
|
270 |
-
height: 20px;
|
271 |
-
border: 3px solid rgba(255, 255, 255, 0.3);
|
272 |
-
border-radius: 50%;
|
273 |
-
border-top-color: white;
|
274 |
-
animation: spin 1s ease-in-out infinite;
|
275 |
-
margin-right: 8px;
|
276 |
}
|
277 |
"""
|
278 |
|
@@ -294,13 +204,6 @@ document.addEventListener('DOMContentLoaded', () => {
|
|
294 |
themeToggle.innerHTML = savedTheme === 'dark' ? '☀️' : '🌙';
|
295 |
themeToggle.onclick = toggleTheme;
|
296 |
document.body.appendChild(themeToggle);
|
297 |
-
|
298 |
-
// Update icon when theme changes
|
299 |
-
document.body.addEventListener('click', (e) => {
|
300 |
-
if (e.target === themeToggle) {
|
301 |
-
themeToggle.innerHTML = document.body.classList.contains('dark') ? '☀️' : '🌙';
|
302 |
-
}
|
303 |
-
});
|
304 |
});
|
305 |
"""
|
306 |
|
@@ -338,8 +241,7 @@ with gr.Blocks(css=css, js=js, theme=gr.themes.Soft()) as demo:
|
|
338 |
result = gr.Image(
|
339 |
label="Generated Image",
|
340 |
show_label=False,
|
341 |
-
height=500
|
342 |
-
elem_id="output-image"
|
343 |
)
|
344 |
with gr.Row():
|
345 |
seed_info = gr.Textbox(
|
@@ -372,7 +274,6 @@ with gr.Blocks(css=css, js=js, theme=gr.themes.Soft()) as demo:
|
|
372 |
randomize_seed = gr.Checkbox(
|
373 |
label="Randomize seed",
|
374 |
value=True,
|
375 |
-
interactive=True
|
376 |
)
|
377 |
|
378 |
with gr.Column():
|
@@ -415,8 +316,7 @@ with gr.Blocks(css=css, js=js, theme=gr.themes.Soft()) as demo:
|
|
415 |
examples=[[example[0], example[1], example[2]]],
|
416 |
inputs=[prompt, width, height],
|
417 |
label="",
|
418 |
-
examples_per_page=20
|
419 |
-
elem_id=f"example-{example[0][:10]}"
|
420 |
)
|
421 |
|
422 |
run_button.click(
|
|
|
1 |
import gradio as gr
|
2 |
import numpy as np
|
3 |
import random
|
4 |
+
from diffusers import DiffusionPipeline
|
5 |
import torch
|
6 |
from time import time
|
7 |
|
8 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
9 |
model_repo_id = "stabilityai/sdxl-turbo"
|
10 |
|
11 |
+
# Simplified model loading
|
12 |
+
try:
|
13 |
+
if torch.cuda.is_available():
|
14 |
+
torch_dtype = torch.float16
|
15 |
+
pipe = DiffusionPipeline.from_pretrained(
|
16 |
+
model_repo_id,
|
17 |
+
torch_dtype=torch_dtype,
|
18 |
+
variant="fp16",
|
19 |
+
use_safetensors=True
|
20 |
+
).to(device)
|
21 |
+
else:
|
22 |
+
torch_dtype = torch.float32
|
23 |
+
pipe = DiffusionPipeline.from_pretrained(
|
24 |
+
model_repo_id,
|
25 |
+
torch_dtype=torch_dtype
|
26 |
+
).to(device)
|
27 |
+
except Exception as e:
|
28 |
+
print(f"Error loading model: {e}")
|
29 |
+
raise
|
30 |
|
31 |
MAX_SEED = np.iinfo(np.int32).max
|
32 |
MAX_IMAGE_SIZE = 1024
|
|
|
42 |
num_inference_steps,
|
43 |
progress=gr.Progress(track_tqdm=True),
|
44 |
):
|
45 |
+
try:
|
46 |
+
start_time = time()
|
47 |
+
|
48 |
+
if randomize_seed:
|
49 |
+
seed = random.randint(0, MAX_SEED)
|
50 |
|
51 |
+
generator = torch.Generator(device=device).manual_seed(seed)
|
52 |
|
53 |
+
# Generate image
|
|
|
54 |
image = pipe(
|
55 |
prompt=prompt,
|
56 |
negative_prompt=negative_prompt,
|
|
|
61 |
generator=generator,
|
62 |
).images[0]
|
63 |
|
64 |
+
gen_time = time() - start_time
|
65 |
+
|
66 |
+
return image, seed, f"Generated in {gen_time:.2f}s"
|
67 |
+
except Exception as e:
|
68 |
+
print(f"Error during inference: {e}")
|
69 |
+
return None, seed, f"Error: {str(e)}"
|
70 |
|
71 |
examples = [
|
72 |
["Astronaut in a jungle, cold color palette, muted colors, detailed, 8k", 1024, 1024],
|
|
|
76 |
|
77 |
css = """
|
78 |
:root {
|
79 |
+
--primary: #6e6af0;
|
80 |
+
--secondary: #f5f5f7;
|
81 |
+
--accent: #f5f5f7;
|
82 |
+
--text: #1e1e1e;
|
83 |
+
--shadow: 0 4px 12px rgba(0, 0, 0, 0.1);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
84 |
}
|
85 |
|
86 |
.dark {
|
87 |
+
--primary: #a5a5fc;
|
88 |
+
--secondary: #2d3748;
|
89 |
+
--accent: #4a5568;
|
90 |
+
--text: #f7fafc;
|
91 |
+
--shadow: 0 4px 12px rgba(0, 0, 0, 0.3);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
}
|
93 |
|
94 |
#col-container {
|
|
|
105 |
.header h1 {
|
106 |
font-size: 2.5rem;
|
107 |
font-weight: 700;
|
108 |
+
color: var(--primary);
|
109 |
margin-bottom: 10px;
|
110 |
}
|
111 |
|
112 |
.header p {
|
113 |
+
color: var(--text);
|
114 |
+
opacity: 0.8;
|
115 |
}
|
116 |
|
117 |
.prompt-container, .result-container, .advanced-settings {
|
118 |
+
background: var(--secondary);
|
119 |
border-radius: 12px;
|
120 |
padding: 20px;
|
121 |
+
box-shadow: var(--shadow);
|
122 |
margin-bottom: 20px;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
123 |
}
|
124 |
|
125 |
.advanced-settings .form {
|
|
|
128 |
gap: 16px;
|
129 |
}
|
130 |
|
|
|
|
|
|
|
|
|
131 |
.control-row {
|
132 |
display: flex;
|
133 |
gap: 10px;
|
|
|
135 |
}
|
136 |
|
137 |
.btn-primary {
|
138 |
+
background: var(--primary) !important;
|
139 |
border: none !important;
|
140 |
color: white !important;
|
141 |
}
|
142 |
|
143 |
.btn-primary:hover {
|
144 |
+
opacity: 0.9 !important;
|
145 |
}
|
146 |
|
147 |
.examples {
|
|
|
152 |
}
|
153 |
|
154 |
.example-prompt {
|
155 |
+
background: var(--secondary);
|
156 |
padding: 12px;
|
157 |
border-radius: 8px;
|
158 |
cursor: pointer;
|
159 |
transition: all 0.2s;
|
|
|
|
|
|
|
|
|
|
|
|
|
160 |
}
|
161 |
|
162 |
.example-prompt:hover {
|
|
|
163 |
transform: translateY(-2px);
|
164 |
+
box-shadow: var(--shadow);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
165 |
}
|
166 |
|
|
|
167 |
.theme-toggle {
|
168 |
position: absolute;
|
169 |
top: 20px;
|
170 |
right: 20px;
|
171 |
+
background: var(--secondary);
|
172 |
border: none;
|
173 |
border-radius: 50%;
|
174 |
width: 40px;
|
|
|
177 |
align-items: center;
|
178 |
justify-content: center;
|
179 |
cursor: pointer;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
180 |
}
|
181 |
|
182 |
@media (max-width: 768px) {
|
183 |
.advanced-settings .form {
|
184 |
grid-template-columns: 1fr;
|
185 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
186 |
}
|
187 |
"""
|
188 |
|
|
|
204 |
themeToggle.innerHTML = savedTheme === 'dark' ? '☀️' : '🌙';
|
205 |
themeToggle.onclick = toggleTheme;
|
206 |
document.body.appendChild(themeToggle);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
207 |
});
|
208 |
"""
|
209 |
|
|
|
241 |
result = gr.Image(
|
242 |
label="Generated Image",
|
243 |
show_label=False,
|
244 |
+
height=500
|
|
|
245 |
)
|
246 |
with gr.Row():
|
247 |
seed_info = gr.Textbox(
|
|
|
274 |
randomize_seed = gr.Checkbox(
|
275 |
label="Randomize seed",
|
276 |
value=True,
|
|
|
277 |
)
|
278 |
|
279 |
with gr.Column():
|
|
|
316 |
examples=[[example[0], example[1], example[2]]],
|
317 |
inputs=[prompt, width, height],
|
318 |
label="",
|
319 |
+
examples_per_page=20
|
|
|
320 |
)
|
321 |
|
322 |
run_button.click(
|