File size: 5,551 Bytes
fcf0449
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4663a72
 
 
 
fcf0449
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4663a72
fcf0449
 
 
 
 
 
 
f2ea589
fcf0449
 
 
 
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
import numpy as np
import torch
from tqdm import tqdm
from pathlib import Path
import os

import clip

imagenet_templates = [
    'a bad photo of a {}.',
    'a photo of many {}.',
    'a sculpture of a {}.',
    'a photo of the hard to see {}.',
    'a low resolution photo of the {}.',
    'a rendering of a {}.',
    'graffiti of a {}.',
    'a bad photo of the {}.',
    'a cropped photo of the {}.',
    'a tattoo of a {}.',
    'the embroidered {}.',
    'a photo of a hard to see {}.',
    'a bright photo of a {}.',
    'a photo of a clean {}.',
    'a photo of a dirty {}.',
    'a dark photo of the {}.',
    'a drawing of a {}.',
    'a photo of my {}.',
    'the plastic {}.',
    'a photo of the cool {}.',
    'a close-up photo of a {}.',
    'a black and white photo of the {}.',
    'a painting of the {}.',
    'a painting of a {}.',
    'a pixelated photo of the {}.',
    'a sculpture of the {}.',
    'a bright photo of the {}.',
    'a cropped photo of a {}.',
    'a plastic {}.',
    'a photo of the dirty {}.',
    'a jpeg corrupted photo of a {}.',
    'a blurry photo of the {}.',
    'a photo of the {}.',
    'a good photo of the {}.',
    'a rendering of the {}.',
    'a {} in a video game.',
    'a photo of one {}.',
    'a doodle of a {}.',
    'a close-up photo of the {}.',
    'a photo of a {}.',
    'the origami {}.',
    'the {} in a video game.',
    'a sketch of a {}.',
    'a doodle of the {}.',
    'a origami {}.',
    'a low resolution photo of a {}.',
    'the toy {}.',
    'a rendition of the {}.',
    'a photo of the clean {}.',
    'a photo of a large {}.',
    'a rendition of a {}.',
    'a photo of a nice {}.',
    'a photo of a weird {}.',
    'a blurry photo of a {}.',
    'a cartoon {}.',
    'art of a {}.',
    'a sketch of the {}.',
    'a embroidered {}.',
    'a pixelated photo of a {}.',
    'itap of the {}.',
    'a jpeg corrupted photo of the {}.',
    'a good photo of a {}.',
    'a plushie {}.',
    'a photo of the nice {}.',
    'a photo of the small {}.',
    'a photo of the weird {}.',
    'the cartoon {}.',
    'art of the {}.',
    'a drawing of the {}.',
    'a photo of the large {}.',
    'a black and white photo of a {}.',
    'the plushie {}.',
    'a dark photo of a {}.',
    'itap of a {}.',
    'graffiti of the {}.',
    'a toy {}.',
    'itap of my {}.',
    'a photo of a cool {}.',
    'a photo of a small {}.',
    'a tattoo of the {}.',
]

FFHQ_CODE_INDICES = [(0, 512), (512, 1024), (1024, 1536), (1536, 2048), (2560, 3072), (3072, 3584), (4096, 4608), (4608, 5120), (5632, 6144), (6144, 6656), (7168, 7680), (7680, 7936), (8192, 8448), (8448, 8576), (8704, 8832), (8832, 8896), (8960, 9024), (9024, 9056)] + \
                    [(2048, 2560), (3584, 4096), (5120, 5632), (6656, 7168), (7936, 8192), (8576, 8704), (8896, 8960), (9056, 9088)]

def zeroshot_classifier(model, classnames, templates, device):

    with torch.no_grad():
        zeroshot_weights = []
        for classname in tqdm(classnames):
            texts = [template.format(classname) for template in templates]  # format with class
            texts = clip.tokenize(texts).to(device)  # tokenize
            class_embeddings = model.encode_text(texts)  # embed with text encoder
            class_embeddings /= class_embeddings.norm(dim=-1, keepdim=True)
            class_embedding = class_embeddings.mean(dim=0)
            class_embedding /= class_embedding.norm()
            zeroshot_weights.append(class_embedding)
        zeroshot_weights = torch.stack(zeroshot_weights, dim=1).to(device)
    return zeroshot_weights


def get_direction(neutral_class, target_class, beta, di, clip_model=None):

    device = "cuda" if torch.cuda.is_available() else "cpu"

    if clip_model is None:
        clip_model, _ = clip.load("ViT-B/32", device=device)

    class_names = [neutral_class, target_class]
    class_weights = zeroshot_classifier(clip_model, class_names, imagenet_templates, device)

    dt = class_weights[:, 1] - class_weights[:, 0]
    dt = dt / dt.norm()

    dt = dt.float()
    di = di.float()

    relevance = di @ dt
    mask = relevance.abs() > beta
    direction = relevance * mask
    direction_max = direction.abs().max()
    if direction_max > 0:
        direction = direction / direction_max
    else:
        raise ValueError(f'Beta value {beta} is too high for mapping from {neutral_class} to {target_class},'
                         f' try setting it to a lower value')
    return direction

def style_tensor_to_style_dict(style_tensor, refernce_generator):
    style_layers = refernce_generator.modulation_layers

    style_dict = {}
    for layer_idx, layer in enumerate(style_layers):
        style_dict[layer] = style_tensor[:, FFHQ_CODE_INDICES[layer_idx][0]:FFHQ_CODE_INDICES[layer_idx][1]]

    return style_dict

def style_dict_to_style_tensor(style_dict, reference_generator):
    style_layers = reference_generator.modulation_layers

    style_tensor = torch.zeros(size=(1, 9088))
    for layer in style_dict:
        layer_idx = style_layers.index(layer)
        style_tensor[:, FFHQ_CODE_INDICES[layer_idx][0]:FFHQ_CODE_INDICES[layer_idx][1]] = style_dict[layer]

    return style_tensor

def project_code_with_styleclip(source_latent, source_class, target_class, alpha, beta, reference_generator, di, clip_model=None):
        edit_direction = get_direction(source_class, target_class, beta, di, clip_model)

        source_s = style_dict_to_style_tensor(source_latent, reference_generator)

        return source_s + alpha * edit_direction