Spaces:
Running
on
Zero
Running
on
Zero
File size: 11,009 Bytes
95ab1e6 3dc751e 95ab1e6 8f570a9 40b82c8 3dc751e 7e66050 95ab1e6 0d89801 d981a02 0d89801 031c42b 83cae6c 0d89801 e128936 72d2b99 741875a 95ab1e6 741875a 7c304e4 95ab1e6 7c304e4 3dc751e 7c304e4 0d89801 95ab1e6 8f570a9 72d2b99 8f570a9 bca1af7 8f570a9 95ab1e6 d981a02 95ab1e6 741875a e128936 d981a02 cfdaacd e128936 d981a02 0d89801 722b968 0d89801 d981a02 0d89801 722b968 0d89801 95ab1e6 d981a02 95ab1e6 d981a02 722b968 0d89801 95ab1e6 d981a02 95ab1e6 3dc751e 20fd04d a92a3f1 20fd04d a92a3f1 3dc751e 902e7a4 a92a3f1 3dc751e 0d89801 d981a02 0d89801 3e075bb e128936 805962c d981a02 805962c d981a02 805962c e128936 d981a02 cfdaacd e128936 72d2b99 e128936 72d2b99 e128936 72d2b99 e128936 72d2b99 e128936 0d89801 2b61647 722b968 7ea5176 722b968 3dc751e 0d89801 d981a02 cfdaacd e128936 0d89801 2b61647 0d89801 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 |
import cv2
import random
from typing import Tuple, Optional
import gradio as gr
import numpy as np
import spaces
import torch
from PIL import Image, ImageFilter
from diffusers import FluxInpaintPipeline
from gradio_client import Client, handle_file
MARKDOWN = """
# FLUX.1 Inpainting 🔥
Shoutout to [Black Forest Labs](https://huggingface.co/black-forest-labs) team for
creating this amazing model, and a big thanks to [Gothos](https://github.com/Gothos)
for taking it to the next level by enabling inpainting with the FLUX.
"""
MAX_SEED = np.iinfo(np.int32).max
IMAGE_SIZE = 1024
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
PIPE = FluxInpaintPipeline.from_pretrained(
"black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16).to(DEVICE)
CLIENT = Client("SkalskiP/florence-sam-masking")
# def remove_background(image: Image.Image, threshold: int = 50) -> Image.Image:
# image = image.convert("RGBA")
# data = image.getdata()
# new_data = []
# for item in data:
# avg = sum(item[:3]) / 3
# if avg < threshold:
# new_data.append((0, 0, 0, 0))
# else:
# new_data.append(item)
#
# image.putdata(new_data)
# return image
# EXAMPLES = [
# [
# {
# "background": Image.open(requests.get("https://media.roboflow.com/spaces/doge-2-image.png", stream=True).raw),
# "layers": [remove_background(Image.open(requests.get("https://media.roboflow.com/spaces/doge-2-mask-2.png", stream=True).raw))],
# "composite": Image.open(requests.get("https://media.roboflow.com/spaces/doge-2-composite-2.png", stream=True).raw),
# },
# "little lion",
# None,
# 42,
# False,
# 0.85,
# 30
# ],
# [
# {
# "background": Image.open(requests.get("https://media.roboflow.com/spaces/doge-2-image.png", stream=True).raw),
# "layers": [remove_background(Image.open(requests.get("https://media.roboflow.com/spaces/doge-2-mask-3.png", stream=True).raw))],
# "composite": Image.open(requests.get("https://media.roboflow.com/spaces/doge-2-composite-3.png", stream=True).raw),
# },
# "tattoos",
# None,
# 42,
# False,
# 0.85,
# 30
# ]
# ]
def calculate_image_dimensions_for_flux(
original_resolution_wh: Tuple[int, int],
maximum_dimension: int = IMAGE_SIZE
) -> Tuple[int, int]:
width, height = original_resolution_wh
if width > height:
scaling_factor = maximum_dimension / width
else:
scaling_factor = maximum_dimension / height
new_width = int(width * scaling_factor)
new_height = int(height * scaling_factor)
new_width = new_width - (new_width % 32)
new_height = new_height - (new_height % 32)
return new_width, new_height
def is_mask_empty(image: Image.Image) -> bool:
gray_img = image.convert("L")
pixels = list(gray_img.getdata())
return all(pixel == 0 for pixel in pixels)
def process_mask(
mask: Image.Image,
mask_inflation: Optional[int] = None,
mask_blur: Optional[int] = None
) -> Image.Image:
"""
Inflates and blurs the white regions of a mask.
Args:
mask (Image.Image): The input mask image.
mask_inflation (Optional[int]): The number of pixels to inflate the mask by.
mask_blur (Optional[int]): The radius of the Gaussian blur to apply.
Returns:
Image.Image: The processed mask with inflated and/or blurred regions.
"""
if mask_inflation and mask_inflation > 0:
mask_array = np.array(mask)
kernel = np.ones((mask_inflation, mask_inflation), np.uint8)
mask_array = cv2.dilate(mask_array, kernel, iterations=1)
mask = Image.fromarray(mask_array)
if mask_blur and mask_blur > 0:
mask = mask.filter(ImageFilter.GaussianBlur(radius=mask_blur))
return mask
@spaces.GPU(duration=100)
def process(
input_image_editor: dict,
inpainting_prompt_text: str,
masking_prompt_text: str,
mask_inflation_slider: int,
mask_blur_slider: int,
seed_slicer: int,
randomize_seed_checkbox: bool,
strength_slider: float,
num_inference_steps_slider: int,
progress=gr.Progress(track_tqdm=True)
):
if not inpainting_prompt_text:
gr.Info("Please enter a text prompt.")
return None, None
image_path = input_image_editor['background']
mask_path = input_image_editor['layers'][0]
image = Image.open(image_path)
mask = Image.open(mask_path)
if not image:
gr.Info("Please upload an image.")
return None, None
if is_mask_empty(mask) and not masking_prompt_text:
gr.Info("Please draw a mask or enter a masking prompt.")
return None, None
if not is_mask_empty(mask) and masking_prompt_text:
gr.Info("Both mask and masking prompt are provided. Please provide only one.")
return None, None
if is_mask_empty(mask):
mask = CLIENT.predict(
image_input=handle_file(image_path),
text_input=masking_prompt_text,
api_name="/process_image")
mask = Image.open(mask)
width, height = calculate_image_dimensions_for_flux(original_resolution_wh=image.size)
image = image.resize((width, height), Image.LANCZOS)
mask = mask.resize((width, height), Image.LANCZOS)
mask = process_mask(mask, mask_inflation=mask_inflation_slider, mask_blur=mask_blur_slider)
if randomize_seed_checkbox:
seed_slicer = random.randint(0, MAX_SEED)
generator = torch.Generator().manual_seed(seed_slicer)
result = PIPE(
prompt=inpainting_prompt_text,
image=image,
mask_image=mask,
width=width,
height=height,
strength=strength_slider,
generator=generator,
num_inference_steps=num_inference_steps_slider
).images[0]
print('INFERENCE DONE')
return result, mask
with gr.Blocks() as demo:
gr.Markdown(MARKDOWN)
with gr.Row():
with gr.Column():
input_image_editor_component = gr.ImageEditor(
label='Image',
type='filepath',
sources=["upload", "webcam"],
image_mode='RGB',
layers=False,
brush=gr.Brush(colors=["#FFFFFF"], color_mode="fixed"))
with gr.Row():
inpainting_prompt_text_component = gr.Text(
label="Prompt",
show_label=False,
max_lines=1,
placeholder="Enter text to generate inpainting",
container=False,
)
submit_button_component = gr.Button(
value='Submit', variant='primary', scale=0)
with gr.Accordion("Advanced Settings", open=False):
masking_prompt_text_component = gr.Text(
label="Prompt",
show_label=False,
max_lines=1,
placeholder="Enter text to generate masking",
container=False,
)
with gr.Row():
mask_inflation_slider_component = gr.Slider(
label="Mask inflation",
info="Adjusts the amount of mask edge expansion before "
"inpainting.",
minimum=0,
maximum=20,
step=1,
value=5,
)
mask_blur_slider_component = gr.Slider(
label="Mask blur",
info="Controls the intensity of the Gaussian blur applied to "
"the mask edges.",
minimum=0,
maximum=20,
step=1,
value=5,
)
seed_slicer_component = gr.Slider(
label="Seed",
minimum=0,
maximum=MAX_SEED,
step=1,
value=42,
)
randomize_seed_checkbox_component = gr.Checkbox(
label="Randomize seed", value=True)
with gr.Row():
strength_slider_component = gr.Slider(
label="Strength",
info="Indicates extent to transform the reference `image`. "
"Must be between 0 and 1. `image` is used as a starting "
"point and more noise is added the higher the `strength`.",
minimum=0,
maximum=1,
step=0.01,
value=0.85,
)
num_inference_steps_slider_component = gr.Slider(
label="Number of inference steps",
info="The number of denoising steps. More denoising steps "
"usually lead to a higher quality image at the",
minimum=1,
maximum=50,
step=1,
value=20,
)
with gr.Column():
output_image_component = gr.Image(
type='pil', image_mode='RGB', label='Generated image', format="png")
with gr.Accordion("Debug", open=False):
output_mask_component = gr.Image(
type='pil', image_mode='RGB', label='Input mask', format="png")
# with gr.Row():
# gr.Examples(
# fn=process,
# examples=EXAMPLES,
# inputs=[
# input_image_editor_component,
# inpainting_prompt_text_component,
# masking_prompt_text_component,
# seed_slicer_component,
# randomize_seed_checkbox_component,
# strength_slider_component,
# num_inference_steps_slider_component
# ],
# outputs=[
# output_image_component,
# output_mask_component
# ],
# run_on_click=True,
# cache_examples=True
# )
submit_button_component.click(
fn=process,
inputs=[
input_image_editor_component,
inpainting_prompt_text_component,
masking_prompt_text_component,
mask_inflation_slider_component,
mask_blur_slider_component,
seed_slicer_component,
randomize_seed_checkbox_component,
strength_slider_component,
num_inference_steps_slider_component
],
outputs=[
output_image_component,
output_mask_component
]
)
demo.launch(debug=False, show_error=True)
|