stable-diffusion-webui/modules/textual_inversion/image_embedding.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

221 lines
8.2 KiB
Python
Raw Normal View History

2022-10-11 12:37:58 -06:00
import base64
import json
import warnings
2022-10-11 12:37:58 -06:00
import numpy as np
import zlib
from PIL import Image, ImageDraw
2022-10-11 12:37:58 -06:00
import torch
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
class EmbeddingEncoder(json.JSONEncoder):
def default(self, obj):
if isinstance(obj, torch.Tensor):
2022-10-12 06:13:25 -06:00
return {'TORCHTENSOR': obj.cpu().detach().numpy().tolist()}
2022-10-11 12:37:58 -06:00
return json.JSONEncoder.default(self, obj)
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
class EmbeddingDecoder(json.JSONDecoder):
def __init__(self, *args, **kwargs):
2023-05-10 02:19:16 -06:00
json.JSONDecoder.__init__(self, *args, object_hook=self.object_hook, **kwargs)
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
def object_hook(self, d):
if 'TORCHTENSOR' in d:
return torch.from_numpy(np.array(d['TORCHTENSOR']))
return d
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
def embedding_to_b64(data):
2022-10-12 06:13:25 -06:00
d = json.dumps(data, cls=EmbeddingEncoder)
2022-10-11 12:37:58 -06:00
return base64.b64encode(d.encode())
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
def embedding_from_b64(data):
d = base64.b64decode(data)
2022-10-12 06:13:25 -06:00
return json.loads(d, cls=EmbeddingDecoder)
2022-10-11 12:37:58 -06:00
def lcg(m=2**32, a=1664525, c=1013904223, seed=0):
while True:
seed = (a * seed + c) % m
2022-10-12 06:13:25 -06:00
yield seed % 255
2022-10-11 12:37:58 -06:00
def xor_block(block):
g = lcg()
randblock = np.array([next(g) for _ in range(np.product(block.shape))]).astype(np.uint8).reshape(block.shape)
2022-10-12 06:13:25 -06:00
return np.bitwise_xor(block.astype(np.uint8), randblock & 0x0F)
2022-10-11 12:37:58 -06:00
2022-10-12 06:13:25 -06:00
def style_block(block, sequence):
im = Image.new('RGB', (block.shape[1], block.shape[0]))
2022-10-11 12:37:58 -06:00
draw = ImageDraw.Draw(im)
2022-10-12 06:13:25 -06:00
i = 0
for x in range(-6, im.size[0], 8):
for yi, y in enumerate(range(-6, im.size[1], 8)):
offset = 0
if yi % 2 == 0:
offset = 4
shade = sequence[i % len(sequence)]
i += 1
draw.ellipse((x+offset, y, x+6+offset, y+6), fill=(shade, shade, shade))
2022-10-11 12:37:58 -06:00
fg = np.array(im).astype(np.uint8) & 0xF0
return block ^ fg
2022-10-12 06:13:25 -06:00
def insert_image_data_embed(image, data):
2022-10-11 12:37:58 -06:00
d = 3
2022-10-12 06:13:25 -06:00
data_compressed = zlib.compress(json.dumps(data, cls=EmbeddingEncoder).encode(), level=9)
data_np_ = np.frombuffer(data_compressed, np.uint8).copy()
2022-10-11 12:37:58 -06:00
data_np_high = data_np_ >> 4
2022-10-12 06:13:25 -06:00
data_np_low = data_np_ & 0x0F
2022-10-11 12:37:58 -06:00
h = image.size[1]
2022-10-12 06:13:25 -06:00
next_size = data_np_low.shape[0] + (h-(data_np_low.shape[0] % h))
next_size = next_size + ((h*d)-(next_size % (h*d)))
2022-10-11 12:37:58 -06:00
data_np_low = np.resize(data_np_low, next_size)
2022-10-12 06:13:25 -06:00
data_np_low = data_np_low.reshape((h, -1, d))
2022-10-11 12:37:58 -06:00
data_np_high = np.resize(data_np_high, next_size)
2022-10-12 06:13:25 -06:00
data_np_high = data_np_high.reshape((h, -1, d))
2022-10-11 12:37:58 -06:00
edge_style = list(data['string_to_param'].values())[0].cpu().detach().numpy().tolist()[0][:1024]
edge_style = (np.abs(edge_style)/np.max(np.abs(edge_style))*255).astype(np.uint8)
2022-10-12 06:13:25 -06:00
data_np_low = style_block(data_np_low, sequence=edge_style)
data_np_low = xor_block(data_np_low)
data_np_high = style_block(data_np_high, sequence=edge_style[::-1])
data_np_high = xor_block(data_np_high)
2022-10-11 12:37:58 -06:00
2022-10-12 06:13:25 -06:00
im_low = Image.fromarray(data_np_low, mode='RGB')
im_high = Image.fromarray(data_np_high, mode='RGB')
2022-10-11 12:37:58 -06:00
2022-10-12 06:13:25 -06:00
background = Image.new('RGB', (image.size[0]+im_low.size[0]+im_high.size[0]+2, image.size[1]), (0, 0, 0))
background.paste(im_low, (0, 0))
background.paste(image, (im_low.size[0]+1, 0))
background.paste(im_high, (im_low.size[0]+1+image.size[0]+1, 0))
2022-10-11 12:37:58 -06:00
return background
2022-10-12 06:13:25 -06:00
def crop_black(img, tol=0):
mask = (img > tol).all(2)
mask0, mask1 = mask.any(0), mask.any(1)
col_start, col_end = mask0.argmax(), mask.shape[1]-mask0[::-1].argmax()
row_start, row_end = mask1.argmax(), mask.shape[0]-mask1[::-1].argmax()
return img[row_start:row_end, col_start:col_end]
2022-10-11 12:37:58 -06:00
def extract_image_data_embed(image):
2022-10-12 06:13:25 -06:00
d = 3
outarr = crop_black(np.array(image.convert('RGB').getdata()).reshape(image.size[1], image.size[0], d).astype(np.uint8)) & 0x0F
black_cols = np.where(np.sum(outarr, axis=(0, 2)) == 0)
2022-10-11 12:37:58 -06:00
if black_cols[0].shape[0] < 2:
print('No Image data blocks found.')
return None
2022-10-12 06:13:25 -06:00
data_block_lower = outarr[:, :black_cols[0].min(), :].astype(np.uint8)
data_block_upper = outarr[:, black_cols[0].max()+1:, :].astype(np.uint8)
2022-10-11 12:37:58 -06:00
data_block_lower = xor_block(data_block_lower)
data_block_upper = xor_block(data_block_upper)
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
data_block = (data_block_upper << 4) | (data_block_lower)
data_block = data_block.flatten().tobytes()
data = zlib.decompress(data_block)
2022-10-12 06:13:25 -06:00
return json.loads(data, cls=EmbeddingDecoder)
2022-10-11 12:37:58 -06:00
2022-10-12 06:13:25 -06:00
def caption_image_overlay(srcimage, title, footerLeft, footerMid, footerRight, textfont=None):
from modules.images import get_font
if textfont:
warnings.warn(
'passing in a textfont to caption_image_overlay is deprecated and does nothing',
DeprecationWarning,
stacklevel=2,
)
2022-10-11 12:37:58 -06:00
from math import cos
image = srcimage.copy()
2022-10-21 07:47:21 -06:00
fontsize = 32
2022-10-11 12:37:58 -06:00
factor = 1.5
2022-10-12 06:13:25 -06:00
gradient = Image.new('RGBA', (1, image.size[1]), color=(0, 0, 0, 0))
2022-10-11 12:37:58 -06:00
for y in range(image.size[1]):
mag = 1-cos(y/image.size[1]*factor)
2022-10-12 06:13:25 -06:00
mag = max(mag, 1-cos((image.size[1]-y)/image.size[1]*factor*1.1))
gradient.putpixel((0, y), (0, 0, 0, int(mag*255)))
2022-10-11 12:37:58 -06:00
image = Image.alpha_composite(image.convert('RGBA'), gradient.resize(image.size))
draw = ImageDraw.Draw(image)
2022-10-21 07:47:21 -06:00
font = get_font(fontsize)
2022-10-11 12:37:58 -06:00
padding = 10
2022-10-12 06:13:25 -06:00
_, _, w, h = draw.textbbox((0, 0), title, font=font)
fontsize = min(int(fontsize * (((image.size[0]*0.75)-(padding*4))/w)), 72)
font = get_font(fontsize)
2022-10-12 06:13:25 -06:00
_, _, w, h = draw.textbbox((0, 0), title, font=font)
draw.text((padding, padding), title, anchor='lt', font=font, fill=(255, 255, 255, 230))
2022-10-11 12:37:58 -06:00
2022-10-12 06:13:25 -06:00
_, _, w, h = draw.textbbox((0, 0), footerLeft, font=font)
fontsize_left = min(int(fontsize * (((image.size[0]/3)-(padding))/w)), 72)
_, _, w, h = draw.textbbox((0, 0), footerMid, font=font)
fontsize_mid = min(int(fontsize * (((image.size[0]/3)-(padding))/w)), 72)
_, _, w, h = draw.textbbox((0, 0), footerRight, font=font)
fontsize_right = min(int(fontsize * (((image.size[0]/3)-(padding))/w)), 72)
2022-10-11 12:37:58 -06:00
font = get_font(min(fontsize_left, fontsize_mid, fontsize_right))
2022-10-11 12:37:58 -06:00
2022-10-12 06:13:25 -06:00
draw.text((padding, image.size[1]-padding), footerLeft, anchor='ls', font=font, fill=(255, 255, 255, 230))
draw.text((image.size[0]/2, image.size[1]-padding), footerMid, anchor='ms', font=font, fill=(255, 255, 255, 230))
draw.text((image.size[0]-padding, image.size[1]-padding), footerRight, anchor='rs', font=font, fill=(255, 255, 255, 230))
2022-10-11 12:37:58 -06:00
return image
2022-10-12 06:13:25 -06:00
2022-10-11 12:37:58 -06:00
if __name__ == '__main__':
2022-10-11 13:21:30 -06:00
testEmbed = Image.open('test_embedding.png')
data = extract_image_data_embed(testEmbed)
assert data is not None
data = embedding_from_b64(testEmbed.text['sd-ti-embedding'])
assert data is not None
2022-10-12 06:13:25 -06:00
image = Image.new('RGBA', (512, 512), (255, 255, 200, 255))
2022-10-11 12:37:58 -06:00
cap_image = caption_image_overlay(image, 'title', 'footerLeft', 'footerMid', 'footerRight')
2022-10-12 06:13:25 -06:00
test_embed = {'string_to_param': {'*': torch.from_numpy(np.random.random((2, 4096)))}}
2022-10-11 12:37:58 -06:00
embedded_image = insert_image_data_embed(cap_image, test_embed)
retrived_embed = extract_image_data_embed(embedded_image)
assert str(retrived_embed) == str(test_embed)
embedded_image2 = insert_image_data_embed(cap_image, retrived_embed)
assert embedded_image == embedded_image2
g = lcg()
shared_random = np.array([next(g) for _ in range(100)]).astype(np.uint8).tolist()
2022-10-12 06:13:25 -06:00
reference_random = [253, 242, 127, 44, 157, 27, 239, 133, 38, 79, 167, 4, 177,
95, 130, 79, 78, 14, 52, 215, 220, 194, 126, 28, 240, 179,
160, 153, 149, 50, 105, 14, 21, 218, 199, 18, 54, 198, 193,
38, 128, 19, 53, 195, 124, 75, 205, 12, 6, 145, 0, 28,
30, 148, 8, 45, 218, 171, 55, 249, 97, 166, 12, 35, 0,
41, 221, 122, 215, 170, 31, 113, 186, 97, 119, 31, 23, 185,
66, 140, 30, 41, 37, 63, 137, 109, 216, 55, 159, 145, 82,
2022-10-11 12:37:58 -06:00
204, 86, 73, 222, 44, 198, 118, 240, 97]
2022-10-12 06:13:25 -06:00
assert shared_random == reference_random
2022-10-11 12:37:58 -06:00
hunna_kay_random_sum = sum(np.array([next(g) for _ in range(100000)]).astype(np.uint8).tolist())
2022-10-11 12:55:54 -06:00
assert 12731374 == hunna_kay_random_sum