add AI generated garment
Browse files
app.py
CHANGED
|
@@ -49,8 +49,35 @@ def url_to_base64(url):
|
|
| 49 |
print(f"Error converting URL to base64: {str(e)}")
|
| 50 |
return None
|
| 51 |
|
| 52 |
-
def run_viton(model_image_path
|
| 53 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
try:
|
| 55 |
api_url = os.environ.get("SERVER_URL")
|
| 56 |
print(f"Using API URL: {api_url}") # Add this to debug
|
|
@@ -110,6 +137,94 @@ def run_viton(model_image_path, garment_image_path, model_url, garment_url,
|
|
| 110 |
img = base64_to_image(img_b64, output_path) # Remove 'self.'
|
| 111 |
generated_images.append(img)
|
| 112 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
print(f"Successfully generated {len(generated_images)} images")
|
| 114 |
return generated_images
|
| 115 |
else:
|
|
@@ -124,48 +239,53 @@ block = gr.Blocks().queue()
|
|
| 124 |
with block:
|
| 125 |
with gr.Row():
|
| 126 |
gr.Markdown("# Virtual Try-On")
|
| 127 |
-
with gr.Row():
|
| 128 |
-
gr.Markdown("**Instructions:** You can either upload images using the file upload interface or provide direct URLs to images. URL inputs will take priority over uploaded files.")
|
| 129 |
with gr.Row():
|
| 130 |
with gr.Column():
|
|
|
|
| 131 |
model_url = gr.Textbox(
|
| 132 |
label="Enter Model Image URL",
|
| 133 |
)
|
| 134 |
vton_img = gr.Image(label="Model", sources=['upload', 'webcam'], type="filepath", height=384)
|
| 135 |
example = gr.Examples(
|
| 136 |
inputs=vton_img,
|
| 137 |
-
examples_per_page=
|
| 138 |
examples=[
|
| 139 |
-
os.path.join(example_path, 'model/model_8.png'),
|
| 140 |
os.path.join(example_path, 'model/model_2.png'),
|
| 141 |
os.path.join(example_path, 'model/model_7.png'),
|
| 142 |
os.path.join(example_path, 'model/model_4.png'),
|
| 143 |
os.path.join(example_path, 'model/model_5.png'),
|
| 144 |
])
|
| 145 |
with gr.Column():
|
|
|
|
| 146 |
garment_url = gr.Textbox(
|
| 147 |
-
label="Enter Garment Image URL",
|
|
|
|
|
|
|
|
|
|
| 148 |
)
|
| 149 |
garm_img = gr.Image(label="Garment", sources=['upload', 'webcam'], type="filepath", height=384)
|
| 150 |
example = gr.Examples(
|
| 151 |
inputs=garm_img,
|
| 152 |
-
examples_per_page=
|
| 153 |
examples=[
|
| 154 |
-
os.path.join(example_path, 'garment/00055_00.jpg'),
|
| 155 |
os.path.join(example_path, 'garment/07764_00.jpg'),
|
| 156 |
os.path.join(example_path, 'garment/03032_00.jpg'),
|
| 157 |
os.path.join(example_path, 'garment/048554_1.jpg'),
|
| 158 |
os.path.join(example_path, 'garment/049805_1.jpg'),
|
| 159 |
])
|
| 160 |
with gr.Column():
|
| 161 |
-
|
|
|
|
| 162 |
with gr.Column():
|
| 163 |
-
run_button = gr.Button(value="
|
|
|
|
| 164 |
n_steps = gr.Slider(label="Steps", minimum=20, maximum=40, value=20, step=1)
|
| 165 |
image_scale = gr.Slider(label="Guidance scale", minimum=1.0, maximum=5.0, value=2.0, step=0.1)
|
| 166 |
seed = gr.Slider(label="Seed", minimum=-1, maximum=2147483647, step=1, value=-1)
|
| 167 |
|
| 168 |
-
|
| 169 |
-
run_button.click(fn=run_viton, inputs=
|
|
|
|
|
|
|
| 170 |
|
| 171 |
block.launch(mcp_server=True)
|
|
|
|
| 49 |
print(f"Error converting URL to base64: {str(e)}")
|
| 50 |
return None
|
| 51 |
|
| 52 |
+
def run_viton(model_image_path: str = None,
|
| 53 |
+
garment_image_path: str = None,
|
| 54 |
+
model_url: str = None,
|
| 55 |
+
garment_url: str = None,
|
| 56 |
+
n_steps=20,
|
| 57 |
+
image_scale=2.0,
|
| 58 |
+
seed=-1
|
| 59 |
+
):
|
| 60 |
+
"""
|
| 61 |
+
Run the Virtual Try-On model with provided images path or URLs.
|
| 62 |
+
|
| 63 |
+
Args:
|
| 64 |
+
model_image_path (str): Path to the model image file.
|
| 65 |
+
garment_image_path (str): Path to the garment image file.
|
| 66 |
+
model_url (str): URL of the model image.
|
| 67 |
+
garment_url (str): URL of the garment image.
|
| 68 |
+
n_steps (int): Number of steps for the model.
|
| 69 |
+
image_scale (float): Scale for the generated images.
|
| 70 |
+
seed (int): Random seed for reproducibility.
|
| 71 |
+
|
| 72 |
+
Returns:
|
| 73 |
+
list: List of generated images in base64 format.
|
| 74 |
+
"""
|
| 75 |
+
if not model_image_path and not model_url:
|
| 76 |
+
print("Error: No model image provided")
|
| 77 |
+
return []
|
| 78 |
+
if not garment_image_path and not garment_url:
|
| 79 |
+
print("Error: No garment image provided")
|
| 80 |
+
return []
|
| 81 |
try:
|
| 82 |
api_url = os.environ.get("SERVER_URL")
|
| 83 |
print(f"Using API URL: {api_url}") # Add this to debug
|
|
|
|
| 137 |
img = base64_to_image(img_b64, output_path) # Remove 'self.'
|
| 138 |
generated_images.append(img)
|
| 139 |
|
| 140 |
+
print(f"Successfully generated {len(generated_images)} images")
|
| 141 |
+
return generated_images
|
| 142 |
+
else:
|
| 143 |
+
print(f"Request failed with status code: {response.status_code}")
|
| 144 |
+
return []
|
| 145 |
+
|
| 146 |
+
except Exception as e:
|
| 147 |
+
print(f"Exception occurred: {str(e)}") # Add this
|
| 148 |
+
return []
|
| 149 |
+
|
| 150 |
+
def run_new_garment(model_image_path: str = None,
|
| 151 |
+
garment_prompt: str = None,
|
| 152 |
+
model_url: str = None,
|
| 153 |
+
n_steps=20,
|
| 154 |
+
image_scale=2.0,
|
| 155 |
+
seed=-1
|
| 156 |
+
):
|
| 157 |
+
"""
|
| 158 |
+
Run the Virtual Try-On model with provided model image and garment image generated using FLUX.1-dev
|
| 159 |
+
based on description of the garment obtained from the user
|
| 160 |
+
|
| 161 |
+
Args:
|
| 162 |
+
model_image_path (str): Path to the model image file.
|
| 163 |
+
garment_prompt (str): Description of the garment.
|
| 164 |
+
model_url (str): URL of the model image.
|
| 165 |
+
n_steps (int): Number of steps for the model.
|
| 166 |
+
image_scale (float): Scale for the generated images.
|
| 167 |
+
seed (int): Random seed for reproducibility.
|
| 168 |
+
|
| 169 |
+
Returns:
|
| 170 |
+
list: List of generated images in base64 format.
|
| 171 |
+
"""
|
| 172 |
+
if not model_image_path and not model_url:
|
| 173 |
+
print("Error: No model image provided")
|
| 174 |
+
return []
|
| 175 |
+
if not garment_prompt or not garment_prompt.strip():
|
| 176 |
+
print("Error: No garment description provided")
|
| 177 |
+
return []
|
| 178 |
+
try:
|
| 179 |
+
api_url = os.environ.get("SERVER_URL")
|
| 180 |
+
print(f"Using API URL: {api_url}") # Add this to debug
|
| 181 |
+
|
| 182 |
+
# Determine which inputs to use (file upload or URL)
|
| 183 |
+
model_b64 = None
|
| 184 |
+
|
| 185 |
+
# Handle model image
|
| 186 |
+
if model_url and model_url.strip():
|
| 187 |
+
print(f"Using model URL: {model_url}")
|
| 188 |
+
model_b64 = url_to_base64(model_url.strip())
|
| 189 |
+
elif model_image_path:
|
| 190 |
+
print(f"Using model file: {model_image_path}")
|
| 191 |
+
model_b64 = image_to_base64(model_image_path)
|
| 192 |
+
|
| 193 |
+
# Check if we have both images
|
| 194 |
+
if not model_b64 or not garment_prompt:
|
| 195 |
+
print("Error: Missing model or garment description")
|
| 196 |
+
return []
|
| 197 |
+
|
| 198 |
+
# Prepare request
|
| 199 |
+
request_data = {
|
| 200 |
+
"model_image_base64": model_b64,
|
| 201 |
+
"garment_prompt": garment_prompt.strip(),
|
| 202 |
+
"n_samples": 1,
|
| 203 |
+
"n_steps": n_steps,
|
| 204 |
+
"image_scale": image_scale,
|
| 205 |
+
"seed": seed
|
| 206 |
+
}
|
| 207 |
+
|
| 208 |
+
# Send request
|
| 209 |
+
response = requests.post(f"{api_url}/new-garment",
|
| 210 |
+
json=request_data,
|
| 211 |
+
timeout=300)
|
| 212 |
+
|
| 213 |
+
print(f"Request sent to {api_url}/new-garment")
|
| 214 |
+
print(f"Response status code: {response.status_code}")
|
| 215 |
+
|
| 216 |
+
if response.status_code == 200:
|
| 217 |
+
result = response.json()
|
| 218 |
+
if result.get("error"):
|
| 219 |
+
print(f"Error: {result['error']}")
|
| 220 |
+
return []
|
| 221 |
+
|
| 222 |
+
generated_images = []
|
| 223 |
+
for i, img_b64 in enumerate(result.get("images_base64", [])):
|
| 224 |
+
output_path = f"flux_output_{i}.png"
|
| 225 |
+
img = base64_to_image(img_b64, output_path) # Remove 'self.'
|
| 226 |
+
generated_images.append(img)
|
| 227 |
+
|
| 228 |
print(f"Successfully generated {len(generated_images)} images")
|
| 229 |
return generated_images
|
| 230 |
else:
|
|
|
|
| 239 |
with block:
|
| 240 |
with gr.Row():
|
| 241 |
gr.Markdown("# Virtual Try-On")
|
|
|
|
|
|
|
| 242 |
with gr.Row():
|
| 243 |
with gr.Column():
|
| 244 |
+
gr.Markdown("### Provide image or URL of upper body photo")
|
| 245 |
model_url = gr.Textbox(
|
| 246 |
label="Enter Model Image URL",
|
| 247 |
)
|
| 248 |
vton_img = gr.Image(label="Model", sources=['upload', 'webcam'], type="filepath", height=384)
|
| 249 |
example = gr.Examples(
|
| 250 |
inputs=vton_img,
|
| 251 |
+
examples_per_page=4,
|
| 252 |
examples=[
|
|
|
|
| 253 |
os.path.join(example_path, 'model/model_2.png'),
|
| 254 |
os.path.join(example_path, 'model/model_7.png'),
|
| 255 |
os.path.join(example_path, 'model/model_4.png'),
|
| 256 |
os.path.join(example_path, 'model/model_5.png'),
|
| 257 |
])
|
| 258 |
with gr.Column():
|
| 259 |
+
gr.Markdown("### Provide image, URL or description of a garment")
|
| 260 |
garment_url = gr.Textbox(
|
| 261 |
+
label="Enter Garment Image URL",
|
| 262 |
+
)
|
| 263 |
+
garment_promt = gr.Textbox(
|
| 264 |
+
label="Describe Garment",
|
| 265 |
)
|
| 266 |
garm_img = gr.Image(label="Garment", sources=['upload', 'webcam'], type="filepath", height=384)
|
| 267 |
example = gr.Examples(
|
| 268 |
inputs=garm_img,
|
| 269 |
+
examples_per_page=4,
|
| 270 |
examples=[
|
|
|
|
| 271 |
os.path.join(example_path, 'garment/07764_00.jpg'),
|
| 272 |
os.path.join(example_path, 'garment/03032_00.jpg'),
|
| 273 |
os.path.join(example_path, 'garment/048554_1.jpg'),
|
| 274 |
os.path.join(example_path, 'garment/049805_1.jpg'),
|
| 275 |
])
|
| 276 |
with gr.Column():
|
| 277 |
+
gr.Markdown("### 2D Result")
|
| 278 |
+
result_gallery = gr.Gallery(label='Output 2D', show_label=False, elem_id="gallery", preview=True, scale=1)
|
| 279 |
with gr.Column():
|
| 280 |
+
run_button = gr.Button(value="Try On with your garment")
|
| 281 |
+
run_button2 = gr.Button(value="Try On with AI generated garment")
|
| 282 |
n_steps = gr.Slider(label="Steps", minimum=20, maximum=40, value=20, step=1)
|
| 283 |
image_scale = gr.Slider(label="Guidance scale", minimum=1.0, maximum=5.0, value=2.0, step=0.1)
|
| 284 |
seed = gr.Slider(label="Seed", minimum=-1, maximum=2147483647, step=1, value=-1)
|
| 285 |
|
| 286 |
+
ips1 = [vton_img, garm_img, model_url, garment_url, n_steps, image_scale, seed]
|
| 287 |
+
run_button.click(fn=run_viton, inputs=ips1, outputs=result_gallery)
|
| 288 |
+
ips2 = [vton_img, garment_promt, model_url, n_steps, image_scale, seed]
|
| 289 |
+
run_button2.click(fn=run_new_garment, inputs=ips2, outputs=result_gallery)
|
| 290 |
|
| 291 |
block.launch(mcp_server=True)
|