Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -9,10 +9,33 @@ from diffusers.models.attention_processor import AttnProcessor2_0
|
|
| 9 |
import gradio as gr
|
| 10 |
from PIL import Image
|
| 11 |
from transformers import AutoProcessor, AutoModelForCausalLM, pipeline
|
|
|
|
|
|
|
| 12 |
|
| 13 |
import subprocess
|
| 14 |
subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
# Download the model files
|
| 17 |
ckpt_dir = snapshot_download(repo_id="John6666/pony-realism-v21main-sdxl")
|
| 18 |
|
|
@@ -33,7 +56,6 @@ pipe.unet.set_attn_processor(AttnProcessor2_0())
|
|
| 33 |
# Define samplers
|
| 34 |
samplers = {
|
| 35 |
"Euler a": EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config),
|
| 36 |
-
"DPM++ 2M": DPMSolverMultistepScheduler.from_config(pipe.scheduler.config, algorithm_type="dpmsolver++", use_karras_sigmas=True),
|
| 37 |
"DPM++ SDE Karras": DPMSolverSDEScheduler.from_config(pipe.scheduler.config, use_karras_sigmas=True)
|
| 38 |
}
|
| 39 |
|
|
@@ -51,6 +73,12 @@ florence_processor = AutoProcessor.from_pretrained('microsoft/Florence-2-base',
|
|
| 51 |
enhancer_medium = pipeline("summarization", model="gokaygokay/Lamini-Prompt-Enchance", device=device)
|
| 52 |
enhancer_long = pipeline("summarization", model="gokaygokay/Lamini-Prompt-Enchance-Long", device=device)
|
| 53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
# Florence caption function
|
| 55 |
def florence_caption(image):
|
| 56 |
# Convert image to PIL if it's not already
|
|
@@ -85,11 +113,21 @@ def enhance_prompt(input_prompt, model_choice):
|
|
| 85 |
|
| 86 |
return enhanced_text
|
| 87 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 88 |
@spaces.GPU(duration=120)
|
| 89 |
def generate_image(additional_positive_prompt, additional_negative_prompt, height, width, num_inference_steps,
|
| 90 |
guidance_scale, num_images_per_prompt, use_random_seed, seed, sampler, clip_skip,
|
| 91 |
use_florence2, use_medium_enhancer, use_long_enhancer,
|
| 92 |
use_positive_prefix, use_positive_suffix, use_negative_prefix, use_negative_suffix,
|
|
|
|
| 93 |
input_image=None, progress=gr.Progress(track_tqdm=True)):
|
| 94 |
|
| 95 |
if use_random_seed:
|
|
@@ -138,7 +176,7 @@ def generate_image(additional_positive_prompt, additional_negative_prompt, heigh
|
|
| 138 |
full_negative_prompt += f", {DEFAULT_NEGATIVE_SUFFIX}"
|
| 139 |
|
| 140 |
try:
|
| 141 |
-
|
| 142 |
prompt=full_positive_prompt,
|
| 143 |
negative_prompt=full_negative_prompt,
|
| 144 |
height=height,
|
|
@@ -148,7 +186,15 @@ def generate_image(additional_positive_prompt, additional_negative_prompt, heigh
|
|
| 148 |
num_images_per_prompt=num_images_per_prompt,
|
| 149 |
generator=torch.Generator(pipe.device).manual_seed(seed)
|
| 150 |
).images
|
| 151 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
except Exception as e:
|
| 153 |
print(f"Error during image generation: {str(e)}")
|
| 154 |
return None, seed, full_positive_prompt, full_negative_prompt
|
|
@@ -188,6 +234,10 @@ with gr.Blocks(theme='bethecloud/storj_theme') as demo:
|
|
| 188 |
use_medium_enhancer = gr.Checkbox(label="Use Medium Prompt Enhancer", value=False)
|
| 189 |
use_long_enhancer = gr.Checkbox(label="Use Long Prompt Enhancer", value=False)
|
| 190 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 191 |
generate_btn = gr.Button("Generate Image")
|
| 192 |
|
| 193 |
with gr.Accordion("Prefix and Suffix Settings", open=True):
|
|
@@ -211,8 +261,6 @@ with gr.Blocks(theme='bethecloud/storj_theme') as demo:
|
|
| 211 |
value=True,
|
| 212 |
info=f"Suffix: {DEFAULT_NEGATIVE_SUFFIX}"
|
| 213 |
)
|
| 214 |
-
|
| 215 |
-
|
| 216 |
|
| 217 |
with gr.Column(scale=1):
|
| 218 |
output_gallery = gr.Gallery(label="Result", elem_id="gallery", show_label=False)
|
|
@@ -227,6 +275,7 @@ with gr.Blocks(theme='bethecloud/storj_theme') as demo:
|
|
| 227 |
guidance_scale, num_images_per_prompt, use_random_seed, seed, sampler,
|
| 228 |
clip_skip, use_florence2, use_medium_enhancer, use_long_enhancer,
|
| 229 |
use_positive_prefix, use_positive_suffix, use_negative_prefix, use_negative_suffix,
|
|
|
|
| 230 |
input_image
|
| 231 |
],
|
| 232 |
outputs=[output_gallery, seed_used, full_positive_prompt_used, full_negative_prompt_used]
|
|
|
|
| 9 |
import gradio as gr
|
| 10 |
from PIL import Image
|
| 11 |
from transformers import AutoProcessor, AutoModelForCausalLM, pipeline
|
| 12 |
+
import requests
|
| 13 |
+
from RealESRGAN import RealESRGAN
|
| 14 |
|
| 15 |
import subprocess
|
| 16 |
subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
|
| 17 |
|
| 18 |
+
def download_file(url, folder_path, filename):
|
| 19 |
+
if not os.path.exists(folder_path):
|
| 20 |
+
os.makedirs(folder_path)
|
| 21 |
+
file_path = os.path.join(folder_path, filename)
|
| 22 |
+
|
| 23 |
+
if os.path.isfile(file_path):
|
| 24 |
+
print(f"File already exists: {file_path}")
|
| 25 |
+
else:
|
| 26 |
+
response = requests.get(url, stream=True)
|
| 27 |
+
if response.status_code == 200:
|
| 28 |
+
with open(file_path, 'wb') as file:
|
| 29 |
+
for chunk in response.iter_content(chunk_size=1024):
|
| 30 |
+
file.write(chunk)
|
| 31 |
+
print(f"File successfully downloaded and saved: {file_path}")
|
| 32 |
+
else:
|
| 33 |
+
print(f"Error downloading the file. Status code: {response.status_code}")
|
| 34 |
+
|
| 35 |
+
# Download ESRGAN models
|
| 36 |
+
download_file("https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x2.pth", "models/upscalers/", "RealESRGAN_x2.pth")
|
| 37 |
+
download_file("https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4.pth", "models/upscalers/", "RealESRGAN_x4.pth")
|
| 38 |
+
|
| 39 |
# Download the model files
|
| 40 |
ckpt_dir = snapshot_download(repo_id="John6666/pony-realism-v21main-sdxl")
|
| 41 |
|
|
|
|
| 56 |
# Define samplers
|
| 57 |
samplers = {
|
| 58 |
"Euler a": EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config),
|
|
|
|
| 59 |
"DPM++ SDE Karras": DPMSolverSDEScheduler.from_config(pipe.scheduler.config, use_karras_sigmas=True)
|
| 60 |
}
|
| 61 |
|
|
|
|
| 73 |
enhancer_medium = pipeline("summarization", model="gokaygokay/Lamini-Prompt-Enchance", device=device)
|
| 74 |
enhancer_long = pipeline("summarization", model="gokaygokay/Lamini-Prompt-Enchance-Long", device=device)
|
| 75 |
|
| 76 |
+
# Initialize ESRGAN models
|
| 77 |
+
realesrgan_x2 = RealESRGAN(device, scale=2)
|
| 78 |
+
realesrgan_x2.load_weights('models/upscalers/RealESRGAN_x2.pth', download=False)
|
| 79 |
+
realesrgan_x4 = RealESRGAN(device, scale=4)
|
| 80 |
+
realesrgan_x4.load_weights('models/upscalers/RealESRGAN_x4.pth', download=False)
|
| 81 |
+
|
| 82 |
# Florence caption function
|
| 83 |
def florence_caption(image):
|
| 84 |
# Convert image to PIL if it's not already
|
|
|
|
| 113 |
|
| 114 |
return enhanced_text
|
| 115 |
|
| 116 |
+
# Upscale function
|
| 117 |
+
def upscale_image(image, scale):
|
| 118 |
+
if scale == 2:
|
| 119 |
+
return realesrgan_x2.predict(image)
|
| 120 |
+
elif scale == 4:
|
| 121 |
+
return realesrgan_x4.predict(image)
|
| 122 |
+
else:
|
| 123 |
+
return image
|
| 124 |
+
|
| 125 |
@spaces.GPU(duration=120)
|
| 126 |
def generate_image(additional_positive_prompt, additional_negative_prompt, height, width, num_inference_steps,
|
| 127 |
guidance_scale, num_images_per_prompt, use_random_seed, seed, sampler, clip_skip,
|
| 128 |
use_florence2, use_medium_enhancer, use_long_enhancer,
|
| 129 |
use_positive_prefix, use_positive_suffix, use_negative_prefix, use_negative_suffix,
|
| 130 |
+
use_upscaler, upscale_factor,
|
| 131 |
input_image=None, progress=gr.Progress(track_tqdm=True)):
|
| 132 |
|
| 133 |
if use_random_seed:
|
|
|
|
| 176 |
full_negative_prompt += f", {DEFAULT_NEGATIVE_SUFFIX}"
|
| 177 |
|
| 178 |
try:
|
| 179 |
+
images = pipe(
|
| 180 |
prompt=full_positive_prompt,
|
| 181 |
negative_prompt=full_negative_prompt,
|
| 182 |
height=height,
|
|
|
|
| 186 |
num_images_per_prompt=num_images_per_prompt,
|
| 187 |
generator=torch.Generator(pipe.device).manual_seed(seed)
|
| 188 |
).images
|
| 189 |
+
|
| 190 |
+
if use_upscaler:
|
| 191 |
+
upscaled_images = []
|
| 192 |
+
for img in images:
|
| 193 |
+
upscaled_img = upscale_image(img, upscale_factor)
|
| 194 |
+
upscaled_images.append(upscaled_img)
|
| 195 |
+
images = upscaled_images
|
| 196 |
+
|
| 197 |
+
return images, seed, full_positive_prompt, full_negative_prompt
|
| 198 |
except Exception as e:
|
| 199 |
print(f"Error during image generation: {str(e)}")
|
| 200 |
return None, seed, full_positive_prompt, full_negative_prompt
|
|
|
|
| 234 |
use_medium_enhancer = gr.Checkbox(label="Use Medium Prompt Enhancer", value=False)
|
| 235 |
use_long_enhancer = gr.Checkbox(label="Use Long Prompt Enhancer", value=False)
|
| 236 |
|
| 237 |
+
with gr.Accordion("Upscaler Settings", open=False):
|
| 238 |
+
use_upscaler = gr.Checkbox(label="Use Upscaler", value=False)
|
| 239 |
+
upscale_factor = gr.Radio(label="Upscale Factor", choices=[2, 4], value=2)
|
| 240 |
+
|
| 241 |
generate_btn = gr.Button("Generate Image")
|
| 242 |
|
| 243 |
with gr.Accordion("Prefix and Suffix Settings", open=True):
|
|
|
|
| 261 |
value=True,
|
| 262 |
info=f"Suffix: {DEFAULT_NEGATIVE_SUFFIX}"
|
| 263 |
)
|
|
|
|
|
|
|
| 264 |
|
| 265 |
with gr.Column(scale=1):
|
| 266 |
output_gallery = gr.Gallery(label="Result", elem_id="gallery", show_label=False)
|
|
|
|
| 275 |
guidance_scale, num_images_per_prompt, use_random_seed, seed, sampler,
|
| 276 |
clip_skip, use_florence2, use_medium_enhancer, use_long_enhancer,
|
| 277 |
use_positive_prefix, use_positive_suffix, use_negative_prefix, use_negative_suffix,
|
| 278 |
+
use_upscaler, upscale_factor,
|
| 279 |
input_image
|
| 280 |
],
|
| 281 |
outputs=[output_gallery, seed_used, full_positive_prompt_used, full_negative_prompt_used]
|