Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -85,24 +85,30 @@ def resize_image(image, resolution):
|
|
| 85 |
new_h = int(h * ratio)
|
| 86 |
return image.resize((new_w, new_h), Image.LANCZOS)
|
| 87 |
|
| 88 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
width, height = 1024, 576
|
| 90 |
image = Image.new("RGB", (width, height), "white")
|
| 91 |
draw = ImageDraw.Draw(image)
|
| 92 |
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
if os.path.exists(font_path):
|
| 98 |
-
try:
|
| 99 |
-
font = ImageFont.truetype(font_path, size=size)
|
| 100 |
-
print(f"Using font: {font_file}")
|
| 101 |
-
break
|
| 102 |
-
except IOError:
|
| 103 |
-
print(f"Error loading font: {font_file}")
|
| 104 |
-
if font is None:
|
| 105 |
-
print("No suitable font found. Using default font.")
|
| 106 |
font = ImageFont.load_default()
|
| 107 |
|
| 108 |
lines = text.split('\n')
|
|
@@ -153,7 +159,7 @@ def text_to_image(text, size=72, position="middle-center"):
|
|
| 153 |
return image
|
| 154 |
|
| 155 |
@spaces.GPU
|
| 156 |
-
def infer_canny(prompt, text_for_image, text_position, font_size,
|
| 157 |
negative_prompt = "nsfw, facial shadows, low resolution, jpeg artifacts, blurry, bad quality, dark face, neon lights",
|
| 158 |
seed = 397886929,
|
| 159 |
randomize_seed = False,
|
|
@@ -171,8 +177,8 @@ def infer_canny(prompt, text_for_image, text_position, font_size,
|
|
| 171 |
seed = random.randint(0, MAX_SEED)
|
| 172 |
generator = torch.Generator().manual_seed(seed)
|
| 173 |
|
| 174 |
-
# Generate text image
|
| 175 |
-
init_image = text_to_image(text_for_image, size=font_size, position=text_position)
|
| 176 |
init_image = resize_image(init_image, MAX_IMAGE_SIZE)
|
| 177 |
|
| 178 |
pipe = pipe_canny.to("cuda")
|
|
@@ -191,7 +197,7 @@ def infer_canny(prompt, text_for_image, text_position, font_size,
|
|
| 191 |
num_images_per_prompt=1,
|
| 192 |
generator=generator,
|
| 193 |
).images[0]
|
| 194 |
-
return image, seed
|
| 195 |
|
| 196 |
def update_button_states(selected_position):
|
| 197 |
return [
|
|
@@ -199,6 +205,20 @@ def update_button_states(selected_position):
|
|
| 199 |
for pos in position_list
|
| 200 |
]
|
| 201 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 202 |
position_list = [
|
| 203 |
"top-left", "top-left-center", "top-center", "top-right-center", "top-right",
|
| 204 |
"upper-left", "upper-left-center", "upper-center", "upper-right-center", "upper-right",
|
|
@@ -207,7 +227,6 @@ position_list = [
|
|
| 207 |
"bottom-left", "bottom-left-center", "bottom-center", "bottom-right-center", "bottom-right"
|
| 208 |
]
|
| 209 |
|
| 210 |
-
|
| 211 |
css = """
|
| 212 |
footer {
|
| 213 |
visibility: hidden;
|
|
@@ -247,14 +266,14 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as Kolors:
|
|
| 247 |
label="Prompt",
|
| 248 |
placeholder="Enter your prompt",
|
| 249 |
lines=2,
|
| 250 |
-
value="coffee in a cup bokeh --ar 85:128 --v 6.0 --style raw5, 4K, 리얼리티 사진"
|
| 251 |
)
|
| 252 |
with gr.Row():
|
| 253 |
text_for_image = gr.Textbox(
|
| 254 |
label="Text for Image Generation",
|
| 255 |
placeholder="Enter text to be converted into an image",
|
| 256 |
lines=3,
|
| 257 |
-
value="대한 萬世 GO"
|
| 258 |
)
|
| 259 |
with gr.Row():
|
| 260 |
with gr.Column():
|
|
@@ -273,6 +292,15 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as Kolors:
|
|
| 273 |
step=1,
|
| 274 |
value=72
|
| 275 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 276 |
with gr.Accordion("Advanced Settings", open=False):
|
| 277 |
negative_prompt = gr.Textbox(
|
| 278 |
label="Negative prompt",
|
|
@@ -329,18 +357,22 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as Kolors:
|
|
| 329 |
with gr.Row():
|
| 330 |
canny_button = gr.Button("Start", elem_id="button")
|
| 331 |
|
| 332 |
-
|
| 333 |
with gr.Column(elem_id="col-right"):
|
| 334 |
-
result = gr.Image(label="Result", show_label=False)
|
| 335 |
seed_used = gr.Number(label="Seed Used")
|
| 336 |
|
| 337 |
-
|
| 338 |
-
|
| 339 |
-
|
| 340 |
-
|
|
|
|
| 341 |
)
|
| 342 |
-
|
| 343 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 344 |
|
| 345 |
# Set initial button states
|
| 346 |
Kolors.load(update_button_states, inputs=[text_position], outputs=position_buttons)
|
|
|
|
| 85 |
new_h = int(h * ratio)
|
| 86 |
return image.resize((new_w, new_h), Image.LANCZOS)
|
| 87 |
|
| 88 |
+
def get_font_files():
|
| 89 |
+
return [f for f in os.listdir() if f.endswith('.ttf')]
|
| 90 |
+
|
| 91 |
+
def preview_font(font_name, preview_text):
|
| 92 |
+
font_size = 24
|
| 93 |
+
image = Image.new('RGB', (400, 100), color='white')
|
| 94 |
+
draw = ImageDraw.Draw(image)
|
| 95 |
+
try:
|
| 96 |
+
font = ImageFont.truetype(font_name, font_size)
|
| 97 |
+
draw.text((10, 10), f"{font_name}:", font=font, fill='black')
|
| 98 |
+
draw.text((10, 40), preview_text, font=font, fill='black')
|
| 99 |
+
except IOError:
|
| 100 |
+
draw.text((10, 10), f"Error loading font: {font_name}", fill='black')
|
| 101 |
+
return image
|
| 102 |
+
|
| 103 |
+
def text_to_image(text, size=72, position="middle-center", font_name="Arial_Unicode.ttf"):
|
| 104 |
width, height = 1024, 576
|
| 105 |
image = Image.new("RGB", (width, height), "white")
|
| 106 |
draw = ImageDraw.Draw(image)
|
| 107 |
|
| 108 |
+
try:
|
| 109 |
+
font = ImageFont.truetype(font_name, size=size)
|
| 110 |
+
except IOError:
|
| 111 |
+
print(f"Error loading font: {font_name}. Using default font.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 112 |
font = ImageFont.load_default()
|
| 113 |
|
| 114 |
lines = text.split('\n')
|
|
|
|
| 159 |
return image
|
| 160 |
|
| 161 |
@spaces.GPU
|
| 162 |
+
def infer_canny(prompt, text_for_image, text_position, font_size, font_name,
|
| 163 |
negative_prompt = "nsfw, facial shadows, low resolution, jpeg artifacts, blurry, bad quality, dark face, neon lights",
|
| 164 |
seed = 397886929,
|
| 165 |
randomize_seed = False,
|
|
|
|
| 177 |
seed = random.randint(0, MAX_SEED)
|
| 178 |
generator = torch.Generator().manual_seed(seed)
|
| 179 |
|
| 180 |
+
# Generate text image with selected font
|
| 181 |
+
init_image = text_to_image(text_for_image, size=font_size, position=text_position, font_name=font_name)
|
| 182 |
init_image = resize_image(init_image, MAX_IMAGE_SIZE)
|
| 183 |
|
| 184 |
pipe = pipe_canny.to("cuda")
|
|
|
|
| 197 |
num_images_per_prompt=1,
|
| 198 |
generator=generator,
|
| 199 |
).images[0]
|
| 200 |
+
return image, seed
|
| 201 |
|
| 202 |
def update_button_states(selected_position):
|
| 203 |
return [
|
|
|
|
| 205 |
for pos in position_list
|
| 206 |
]
|
| 207 |
|
| 208 |
+
|
| 209 |
+
def generate_font_preview(font_name):
|
| 210 |
+
font_size = 24
|
| 211 |
+
image = Image.new('RGB', (400, 100), color='white')
|
| 212 |
+
draw = ImageDraw.Draw(image)
|
| 213 |
+
try:
|
| 214 |
+
font = ImageFont.truetype(font_name, font_size)
|
| 215 |
+
draw.text((10, 10), f"{font_name}:", font=font, fill='black')
|
| 216 |
+
draw.text((10, 40), "AaBbCc 123 가나다", font=font, fill='black')
|
| 217 |
+
except IOError:
|
| 218 |
+
draw.text((10, 10), f"Error loading font: {font_name}", fill='black')
|
| 219 |
+
return image
|
| 220 |
+
|
| 221 |
+
|
| 222 |
position_list = [
|
| 223 |
"top-left", "top-left-center", "top-center", "top-right-center", "top-right",
|
| 224 |
"upper-left", "upper-left-center", "upper-center", "upper-right-center", "upper-right",
|
|
|
|
| 227 |
"bottom-left", "bottom-left-center", "bottom-center", "bottom-right-center", "bottom-right"
|
| 228 |
]
|
| 229 |
|
|
|
|
| 230 |
css = """
|
| 231 |
footer {
|
| 232 |
visibility: hidden;
|
|
|
|
| 266 |
label="Prompt",
|
| 267 |
placeholder="Enter your prompt",
|
| 268 |
lines=2,
|
| 269 |
+
value="coffee in a cup bokeh --ar 85:128 --v 6.0 --style raw5, 4K, 리얼리티 사진"
|
| 270 |
)
|
| 271 |
with gr.Row():
|
| 272 |
text_for_image = gr.Textbox(
|
| 273 |
label="Text for Image Generation",
|
| 274 |
placeholder="Enter text to be converted into an image",
|
| 275 |
lines=3,
|
| 276 |
+
value="대한 萬世 GO"
|
| 277 |
)
|
| 278 |
with gr.Row():
|
| 279 |
with gr.Column():
|
|
|
|
| 292 |
step=1,
|
| 293 |
value=72
|
| 294 |
)
|
| 295 |
+
with gr.Row():
|
| 296 |
+
font_dropdown = gr.Dropdown(choices=get_font_files(), label="Select Font", value="Arial_Unicode.ttf")
|
| 297 |
+
|
| 298 |
+
with gr.Row():
|
| 299 |
+
preview_text = gr.Textbox(label="Preview Text", value="Hello, World!")
|
| 300 |
+
preview_button = gr.Button("Preview Font")
|
| 301 |
+
|
| 302 |
+
font_preview = gr.Image(label="Font Preview")
|
| 303 |
+
|
| 304 |
with gr.Accordion("Advanced Settings", open=False):
|
| 305 |
negative_prompt = gr.Textbox(
|
| 306 |
label="Negative prompt",
|
|
|
|
| 357 |
with gr.Row():
|
| 358 |
canny_button = gr.Button("Start", elem_id="button")
|
| 359 |
|
|
|
|
| 360 |
with gr.Column(elem_id="col-right"):
|
| 361 |
+
result = gr.Image(label="Result", show_label=False)
|
| 362 |
seed_used = gr.Number(label="Seed Used")
|
| 363 |
|
| 364 |
+
# Update font preview when button is clicked
|
| 365 |
+
preview_button.click(
|
| 366 |
+
fn=preview_font,
|
| 367 |
+
inputs=[font_dropdown, preview_text],
|
| 368 |
+
outputs=font_preview
|
| 369 |
)
|
|
|
|
| 370 |
|
| 371 |
+
canny_button.click(
|
| 372 |
+
fn=infer_canny,
|
| 373 |
+
inputs=[prompt, text_for_image, text_position, font_size, font_dropdown, negative_prompt, seed, randomize_seed, guidance_scale, num_inference_steps, controlnet_conditioning_scale, control_guidance_end, strength],
|
| 374 |
+
outputs=[result, seed_used]
|
| 375 |
+
)
|
| 376 |
|
| 377 |
# Set initial button states
|
| 378 |
Kolors.load(update_button_states, inputs=[text_position], outputs=position_buttons)
|