Spaces:
Paused
Paused
File size: 13,561 Bytes
343e5a8 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 |
import random
import base64
from io import BytesIO
import gradio as gr
import numpy as np
import torch
import torchvision.transforms.functional as F
from transformers import AutoTokenizer, CLIPTextModel
tokenizer = AutoTokenizer.from_pretrained("stabilityai/sd-turbo", subfolder="tokenizer")
style_list = [
{
"name": "Cinematic",
"prompt": "cinematic still {prompt} . emotional, harmonious, vignette, highly detailed, high budget, bokeh, cinemascope, moody, epic, gorgeous, film grain, grainy",
},
{
"name": "3D Model",
"prompt": "professional 3d model {prompt} . octane render, highly detailed, volumetric, dramatic lighting",
},
{
"name": "Anime",
"prompt": "anime artwork {prompt} . anime style, key visual, vibrant, studio anime, highly detailed",
},
{
"name": "Digital Art",
"prompt": "concept art {prompt} . digital artwork, illustrative, painterly, matte painting, highly detailed",
},
{
"name": "Photographic",
"prompt": "cinematic photo {prompt} . 35mm photograph, film, bokeh, professional, 4k, highly detailed",
},
{
"name": "Pixel art",
"prompt": "pixel-art {prompt} . low-res, blocky, pixel art style, 8-bit graphics",
},
{
"name": "Fantasy art",
"prompt": "ethereal fantasy concept art of {prompt} . magnificent, celestial, ethereal, painterly, epic, majestic, magical, fantasy art, cover art, dreamy",
},
{
"name": "Neonpunk",
"prompt": "neonpunk style {prompt} . cyberpunk, vaporwave, neon, vibes, vibrant, stunningly beautiful, crisp, detailed, sleek, ultramodern, magenta highlights, dark purple shadows, high contrast, cinematic, ultra detailed, intricate, professional",
},
{
"name": "Manga",
"prompt": "manga style {prompt} . vibrant, high-energy, detailed, iconic, Japanese comic style",
},
]
styles = {k["name"]: k["prompt"] for k in style_list}
STYLE_NAMES = list(styles.keys())
DEFAULT_STYLE_NAME = "Fantasy art"
MAX_SEED = np.iinfo(np.int32).max
def pil_image_to_data_uri(img, format="PNG"):
buffered = BytesIO()
img.save(buffered, format=format)
img_str = base64.b64encode(buffered.getvalue()).decode()
return f"data:image/{format.lower()};base64,{img_str}"
def run(image, prompt, prompt_template, style_name, seed):
print(f"prompt: {prompt}")
print("sketch updated")
if image is None:
ones = Image.new("L", (512, 512), 255)
temp_uri = pil_image_to_data_uri(ones)
return ones, gr.update(link=temp_uri), gr.update(link=temp_uri)
prompt = prompt_template.replace("{prompt}", prompt)
image = image.convert("RGB")
image_t = F.to_tensor(image) > 0.5
print(f"seed={seed}")
caption_tokens = tokenizer(prompt, max_length=tokenizer.model_max_length, padding="max_length", truncation=True, return_tensors="pt").input_ids.cpu()
with torch.no_grad():
c_t = image_t.unsqueeze(0)
torch.manual_seed(seed)
B, C, H, W = c_t.shape
noise = torch.randn((1, 4, H // 8, W // 8))
output_image = torch.from_numpy(tokenizer.compiled_model([c_t.to(torch.float32), caption_tokens, noise])[0])
output_pil = F.to_pil_image(output_image[0].cpu() * 0.5 + 0.5)
input_sketch_uri = pil_image_to_data_uri(Image.fromarray(255 - np.array(image)))
output_image_uri = pil_image_to_data_uri(output_pil)
return (
output_pil,
gr.update(link=input_sketch_uri),
gr.update(link=output_image_uri),
)
def update_canvas(use_line, use_eraser):
if use_eraser:
_color = "#ffffff"
brush_size = 20
if use_line:
_color = "#000000"
brush_size = 4
return gr.update(brush_radius=brush_size, brush_color=_color, interactive=True)
def upload_sketch(file):
_img = Image.open(file.name)
_img = _img.convert("L")
return gr.update(value=_img, source="upload", interactive=True)
scripts = """
async () => {
globalThis.theSketchDownloadFunction = () => {
console.log("test")
var link = document.createElement("a");
dataUri = document.getElementById('download_sketch').href
link.setAttribute("href", dataUri)
link.setAttribute("download", "sketch.png")
document.body.appendChild(link); // Required for Firefox
link.click();
document.body.removeChild(link); // Clean up
// also call the output download function
theOutputDownloadFunction();
return false
}
globalThis.theOutputDownloadFunction = () => {
console.log("test output download function")
var link = document.createElement("a");
dataUri = document.getElementById('download_output').href
link.setAttribute("href", dataUri);
link.setAttribute("download", "output.png");
document.body.appendChild(link); // Required for Firefox
link.click();
document.body.removeChild(link); // Clean up
return false
}
globalThis.UNDO_SKETCH_FUNCTION = () => {
console.log("undo sketch function")
var button_undo = document.querySelector('#input_image > div.image-container.svelte-p3y7hu > div.svelte-s6ybro > button:nth-child(1)');
// Create a new 'click' event
var event = new MouseEvent('click', {
'view': window,
'bubbles': true,
'cancelable': true
});
button_undo.dispatchEvent(event);
}
globalThis.DELETE_SKETCH_FUNCTION = () => {
console.log("delete sketch function")
var button_del = document.querySelector('#input_image > div.image-container.svelte-p3y7hu > div.svelte-s6ybro > button:nth-child(2)');
// Create a new 'click' event
var event = new MouseEvent('click', {
'view': window,
'bubbles': true,
'cancelable': true
});
button_del.dispatchEvent(event);
}
globalThis.togglePencil = () => {
el_pencil = document.getElementById('my-toggle-pencil');
el_pencil.classList.toggle('clicked');
// simulate a click on the gradio button
btn_gradio = document.querySelector("#cb-line > label > input");
var event = new MouseEvent('click', {
'view': window,
'bubbles': true,
'cancelable': true
});
btn_gradio.dispatchEvent(event);
if (el_pencil.classList.contains('clicked')) {
document.getElementById('my-toggle-eraser').classList.remove('clicked');
document.getElementById('my-div-pencil').style.backgroundColor = "gray";
document.getElementById('my-div-eraser').style.backgroundColor = "white";
}
else {
document.getElementById('my-toggle-eraser').classList.add('clicked');
document.getElementById('my-div-pencil').style.backgroundColor = "white";
document.getElementById('my-div-eraser').style.backgroundColor = "gray";
}
}
globalThis.toggleEraser = () => {
element = document.getElementById('my-toggle-eraser');
element.classList.toggle('clicked');
// simulate a click on the gradio button
btn_gradio = document.querySelector("#cb-eraser > label > input");
var event = new MouseEvent('click', {
'view': window,
'bubbles': true,
'cancelable': true
});
btn_gradio.dispatchEvent(event);
if (element.classList.contains('clicked')) {
document.getElementById('my-toggle-pencil').classList.remove('clicked');
document.getElementById('my-div-pencil').style.backgroundColor = "white";
document.getElementById('my-div-eraser').style.backgroundColor = "gray";
}
else {
document.getElementById('my-toggle-pencil').classList.add('clicked');
document.getElementById('my-div-pencil').style.backgroundColor = "gray";
document.getElementById('my-div-eraser').style.backgroundColor = "white";
}
}
}
"""
with gr.Blocks(css="style.css") as demo:
# these are hidden buttons that are used to trigger the canvas changes
line = gr.Checkbox(label="line", value=False, elem_id="cb-line")
eraser = gr.Checkbox(label="eraser", value=False, elem_id="cb-eraser")
with gr.Row(elem_id="main_row"):
with gr.Column(elem_id="column_input"):
gr.Markdown("## INPUT", elem_id="input_header")
image = gr.Image(
source="canvas",
tool="color-sketch",
type="pil",
image_mode="L",
invert_colors=True,
shape=(512, 512),
brush_radius=4,
height=440,
width=440,
brush_color="#000000",
interactive=True,
show_download_button=True,
elem_id="input_image",
show_label=False,
)
download_sketch = gr.Button("Download sketch", scale=1, elem_id="download_sketch")
gr.HTML(
"""
<div class="button-row">
<div id="my-div-pencil" class="pad2"> <button id="my-toggle-pencil" onclick="return togglePencil(this)"></button> </div>
<div id="my-div-eraser" class="pad2"> <button id="my-toggle-eraser" onclick="return toggleEraser(this)"></button> </div>
<div class="pad2"> <button id="my-button-undo" onclick="return UNDO_SKETCH_FUNCTION(this)"></button> </div>
<div class="pad2"> <button id="my-button-clear" onclick="return DELETE_SKETCH_FUNCTION(this)"></button> </div>
<div class="pad2"> <button href="TODO" download="image" id="my-button-down" onclick='return theSketchDownloadFunction()'></button> </div>
</div>
"""
)
# gr.Markdown("## Prompt", elem_id="tools_header")
prompt = gr.Textbox(label="Prompt", value="", show_label=True)
with gr.Row():
style = gr.Dropdown(
label="Style",
choices=STYLE_NAMES,
value=DEFAULT_STYLE_NAME,
scale=1,
)
prompt_temp = gr.Textbox(
label="Prompt Style Template",
value=styles[DEFAULT_STYLE_NAME],
scale=2,
max_lines=1,
)
with gr.Row():
seed = gr.Textbox(label="Seed", value=42, scale=1, min_width=50)
randomize_seed = gr.Button("Random", scale=1, min_width=50)
with gr.Column(elem_id="column_process", min_width=50, scale=0.4):
gr.Markdown("## pix2pix-turbo", elem_id="description")
run_button = gr.Button("Run", min_width=50)
with gr.Column(elem_id="column_output"):
gr.Markdown("## OUTPUT", elem_id="output_header")
result = gr.Image(
label="Result",
height=440,
width=440,
elem_id="output_image",
show_label=False,
show_download_button=True,
)
download_output = gr.Button("Download output", elem_id="download_output")
gr.Markdown("### Instructions")
gr.Markdown("**1**. Enter a text prompt (e.g. cat)")
gr.Markdown("**2**. Start sketching")
gr.Markdown("**3**. Change the image style using a style template")
gr.Markdown("**4**. Try different seeds to generate different results")
eraser.change(
fn=lambda x: gr.update(value=not x),
inputs=[eraser],
outputs=[line],
queue=False,
api_name=False,
).then(update_canvas, [line, eraser], [image])
line.change(
fn=lambda x: gr.update(value=not x),
inputs=[line],
outputs=[eraser],
queue=False,
api_name=False,
).then(update_canvas, [line, eraser], [image])
demo.load(None, None, None, _js=scripts)
randomize_seed.click(
lambda x: random.randint(0, MAX_SEED),
inputs=[],
outputs=seed,
queue=False,
api_name=False,
)
inputs = [image, prompt, prompt_temp, style, seed]
outputs = [result, download_sketch, download_output]
prompt.submit(fn=run, inputs=inputs, outputs=outputs, api_name=False)
style.change(
lambda x: styles[x],
inputs=[style],
outputs=[prompt_temp],
queue=False,
api_name=False,
).then(
fn=run,
inputs=inputs,
outputs=outputs,
api_name=False,
)
run_button.click(fn=run, inputs=inputs, outputs=outputs, api_name=False)
image.change(run, inputs=inputs, outputs=outputs, queue=False, api_name=False)
try:
demo.queue().launch(debug=False)
except Exception:
demo.queue().launch(debug=False, share=True)
# if you are launching remotely, specify server_name and server_port
# demo.launch(server_name='your server name', server_port='server port in int')
# Read more in the docs: https://gradio.app/docs/ |