File size: 12,612 Bytes
3dabe4a |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 |
from modules.sd_samplers_kdiffusion import KDiffusionSampler
from modules import shared, scripts, script_callbacks
import gradio as gr
from scripts.cg_xyz import xyz_support
VERSION = "v1.1.0"
DYNAMIC_RANGE = [3.25, 2.5, 2.5, 2.5]
Default_LUTs = {"C": 0.01, "M": 0.5, "Y": -0.13, "K": 0}
def normalize_tensor(x, r):
X = x.detach().clone()
ratio = r / max(abs(float(X.min())), abs(float(X.max())))
X *= max(ratio, 0.99)
return X
original_callback = KDiffusionSampler.callback_state
def center_callback(self, d):
if not self.diffcg_enable or getattr(self.p, "_ad_inner", False):
return original_callback(self, d)
X = d["denoised"].detach().clone()
batchSize = X.size(0)
channels = len(self.LUTs)
for b in range(batchSize):
for c in range(channels):
if self.diffcg_recenter_strength > 0.0:
d["denoised"][b][c] += (
self.LUTs[c] - X[b][c].mean()
) * self.diffcg_recenter_strength
if self.diffcg_normalize and (d["i"] + 1) > self.diffcg_last_step // 2:
d["denoised"][b][c] = normalize_tensor(X[b][c], DYNAMIC_RANGE[c])
return original_callback(self, d)
KDiffusionSampler.callback_state = center_callback
# ["None", "txt2img", "img2img", "Both"]
ac = getattr(shared.opts, "always_center", "None")
an = getattr(shared.opts, "always_normalize", "None")
def_sd = getattr(shared.opts, "default_arch", "1.5")
adv_opt = getattr(shared.opts, "show_center_opt", False)
c_t2i = ac in ("txt2img", "Both")
c_i2i = ac in ("img2img", "Both")
n_t2i = an in ("txt2img", "Both")
n_i2i = an in ("img2img", "Both")
class DiffusionCG(scripts.Script):
def __init__(self):
self.xyzCache = {}
xyz_support(self.xyzCache)
def title(self):
return "DiffusionCG"
def show(self, is_img2img):
return scripts.AlwaysVisible
def ui(self, is_img2img):
with gr.Accordion(f"Diffusion CG {VERSION}", open=False):
with gr.Row():
enableG = gr.Checkbox(
label="Enable (Global)",
value=(
((not is_img2img) and (c_t2i or n_t2i))
or (is_img2img and (c_i2i or n_i2i))
),
)
sd_ver = gr.Radio(
["1.5", "XL"], value=def_sd, label="Stable Diffusion Version"
)
with gr.Row():
with gr.Group():
gr.Markdown('<h3 align="center">Recenter</h3>')
if not is_img2img:
v = 1.0 if c_t2i else 0.0
else:
v = 1.0 if c_i2i else 0.0
rc_str = gr.Slider(
label="Effect Strength",
minimum=0.0,
maximum=1.0,
step=0.2,
value=v,
)
with gr.Group():
gr.Markdown('<h3 align="center">Normalization</h3>')
enableN = gr.Checkbox(
label="Activate",
value=(((not is_img2img) and n_t2i) or (is_img2img and n_i2i)),
)
with gr.Accordion("Recenter Settings", visible=adv_opt, open=False):
with gr.Group(visible=(def_sd == "1.5")) as setting15:
C = gr.Slider(
label="C",
minimum=-1.00,
maximum=1.00,
step=0.01,
value=Default_LUTs["C"],
)
M = gr.Slider(
label="M",
minimum=-1.00,
maximum=1.00,
step=0.01,
value=Default_LUTs["M"],
)
Y = gr.Slider(
label="Y",
minimum=-1.00,
maximum=1.00,
step=0.01,
value=Default_LUTs["Y"],
)
K = gr.Slider(
label="K",
minimum=-1.00,
maximum=1.00,
step=0.01,
value=Default_LUTs["K"],
)
with gr.Group(visible=(def_sd == "XL")) as settingXL:
L = gr.Slider(
label="L", minimum=-1.00, maximum=1.00, step=0.01, value=0.0
)
a = gr.Slider(
label="a", minimum=-1.00, maximum=1.00, step=0.01, value=0.0
)
b = gr.Slider(
label="b", minimum=-1.00, maximum=1.00, step=0.01, value=0.0
)
def on_radio_change(choice):
if choice == "1.5":
return [
gr.Group.update(visible=True),
gr.Group.update(visible=False),
]
else:
return [
gr.Group.update(visible=False),
gr.Group.update(visible=True),
]
sd_ver.change(on_radio_change, sd_ver, [setting15, settingXL])
self.paste_field_names = [
(rc_str, "ReCenter Str"),
(enableN, "Normalization"),
(sd_ver, "SD_ver"),
]
self.infotext_fields = [
(rc_str, "ReCenter Str"),
(enableN, "Normalization"),
(sd_ver, "SD_ver"),
]
if adv_opt:
self.paste_field_names += [
(
C,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[0])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
M,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[1])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
Y,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[2])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
K,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[3])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
L,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[0])
if len(d.get("LUTs", "").split(",")) == 3
else gr.update()
),
),
(
a,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[1])
if len(d.get("LUTs", "").split(",")) == 3
else gr.update()
),
),
(
b,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[2])
if len(d.get("LUTs", "").split(",")) == 3
else gr.update()
),
),
]
self.infotext_fields += [
(
C,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[0])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
M,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[1])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
Y,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[2])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
K,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[3])
if len(d.get("LUTs", "").split(",")) == 4
else gr.update()
),
),
(
L,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[0])
if len(d.get("LUTs", "").split(",")) == 3
else gr.update()
),
),
(
a,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[1])
if len(d.get("LUTs", "").split(",")) == 3
else gr.update()
),
),
(
b,
lambda d: (
float(d["LUTs"].strip("[]").split(",")[2])
if len(d.get("LUTs", "").split(",")) == 3
else gr.update()
),
),
]
for comp in [enableG, sd_ver, rc_str, enableN, C, M, Y, K, L, a, b]:
comp.do_not_save_to_config = True
return [enableG, sd_ver, rc_str, enableN, C, M, Y, K, L, a, b]
def before_hr(self, p, *args):
KDiffusionSampler.diffcg_normalize = False
def process(
self,
p,
enableG: bool,
sd_ver: str,
rc_str: float,
enableN: bool,
C: float,
M: float,
Y: float,
K: float,
L: float,
a: float,
b: float,
):
if "enableG" in self.xyzCache.keys():
enableG = self.xyzCache["enableG"].lower().strip() == "true"
del self.xyzCache["enableG"]
KDiffusionSampler.diffcg_enable = enableG
if not enableG:
if len(self.xyzCache.keys()) > 0:
print("\n[Diff. CG] X [X/Y/Z Plot] Extension is not Enabled!\n")
self.xyzCache.clear()
return p
if "rc_str" in self.xyzCache.keys():
rc_str = float(self.xyzCache["rc_str"])
if "enableN" in self.xyzCache.keys():
enableN = self.xyzCache["enableN"].lower().strip() == "true"
if adv_opt:
C = self.xyzCache.get("C", C)
M = self.xyzCache.get("M", M)
Y = self.xyzCache.get("Y", Y)
K = self.xyzCache.get("K", K)
L = self.xyzCache.get("L", L)
a = self.xyzCache.get("a", a)
b = self.xyzCache.get("b", b)
if sd_ver == "1.5":
KDiffusionSampler.LUTs = [-K, -M, C, Y]
else:
KDiffusionSampler.LUTs = [L, -a, b]
KDiffusionSampler.diffcg_recenter_strength = rc_str
KDiffusionSampler.diffcg_normalize = enableN
if (
not hasattr(p, "enable_hr")
and not shared.opts.img2img_fix_steps
and getattr(p, "denoising_strength", 1.0) < 1.0
):
KDiffusionSampler.diffcg_last_step = int(p.steps * p.denoising_strength) + 1
else:
KDiffusionSampler.diffcg_last_step = p.steps
p.extra_generation_params.update(
{
"ReCenter Str": rc_str,
"Normalization": enableN,
"SD_ver": sd_ver,
}
)
if adv_opt:
if def_sd == "1.5":
p.extra_generation_params["LUTs"] = f"[{C}, {M}, {Y}, {K}]"
else:
p.extra_generation_params["LUTs"] = f"[{L}, {a}, {b}]"
self.xyzCache.clear()
def restore_callback():
KDiffusionSampler.callback_state = original_callback
script_callbacks.on_script_unloaded(restore_callback)
|