pix2latent/examples/invert_stylegan2_cars_hybrid_ng.py at master · minyoungg/pix2latent · GitHub

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
import os, os.path as osp
import numpy as np
import argparse

import torch
import torch.nn as nn

from pix2latent.model.stylegan2 import StyleGAN2

from pix2latent import VariableManager, save_variables
from pix2latent.optimizer import HybridNevergradOptimizer
from pix2latent.utils import image, video

import pix2latent.loss_functions as LF
import pix2latent.utils.function_hooks as hook
import pix2latent.distribution as dist


parser = argparse.ArgumentParser()
parser.add_argument('--ng_method', type=str, default='CMA')
parser.add_argument('--lr', type=float, default=0.05)
parser.add_argument('--latent_noise', type=float, default=0.05)
parser.add_argument('--truncate', type=float, default=2.0)
parser.add_argument('--make_video', action='store_true')
parser.add_argument('--num_samples', type=int, default=4)
parser.add_argument('--max_minibatch', type=int, default=9)
args = parser.parse_args()


### ---- initialize --- ###

model = StyleGAN2(model='cars', search='z')

filename = './images/car-example.png'

target = image.read(filename, as_transformed_tensor=True, im_size=512,
                    transform_style='stylegan')

# we apply a mask since the generated resolution is 384 x 512
loss_mask = torch.zeros((3, 512, 512))
loss_mask[:, 64:-64, :].data += 1.0

weight = loss_mask


fn = filename.split('/')[-1].split('.')[0]
save_dir = f'./results/stylegan2_cars/hybridng_{args.ng_method}_{fn}'


model = StyleGAN2(search='z')
model = nn.DataParallel(model)
loss_fn = LF.ProjectionLoss()


var_manager = VariableManager()

var_manager.register(
                variable_name='z',
                shape=(512,),
                default=None,
                grad_free=True,
                distribution=dist.TruncatedNormalModulo(
                                            sigma=1.0,
                                            trunc=args.truncate
                                            ),
                var_type='input',
                learning_rate=args.lr,
                hook_fn=hook.Compose(
                            hook.NormalPerturb(sigma=args.latent_noise),
                            hook.Clamp(trunc=args.truncate),
                            )
                )

var_manager.register(
                variable_name='target',
                shape=(3, 512, 512),
                requires_grad=False,
                default=target,
                var_type='output'
                )

var_manager.register(
                variable_name='weight',
                shape=(3, 512, 512),
                requires_grad=False,
                default=weight,
                var_type='output'
                )

var_manager.register(
                variable_name='loss_mask',
                shape=(3, 512, 512),
                requires_grad=False,
                default=loss_mask,
                var_type='output'
                )


### ---- optimize --- ###

opt = HybridNevergradOptimizer(
            args.ng_method, model, var_manager, loss_fn,
            max_batch_size=args.max_minibatch,
            log=args.make_video
            )

opt.log_resize_factor = 0.5

vars, out, loss = opt.optimize(
            num_samples=args.num_samples, meta_steps=30,
            grad_steps=50, last_grad_steps=300,
            )


### ---- save results ---- #

vars.loss = loss
os.makedirs(save_dir, exist_ok=True)

save_variables(osp.join(save_dir, 'vars.npy'), vars)

if args.make_video:
    video.make_video(osp.join(save_dir, 'out.mp4'), out)

image.save(osp.join(save_dir, 'target.jpg'), target)
image.save(osp.join(save_dir, 'mask.jpg'), image.binarize(weight))
image.save(osp.join(save_dir, 'out.jpg'), out[-1])
np.save(osp.join(save_dir, 'tracked.npy'), opt.tracked)