-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathreadme_styleganv2ada.txt
618 lines (332 loc) · 35.4 KB
/
readme_styleganv2ada.txt
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
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
# 如果命令不能成功执行，说明咩酱实现中，，，
nvidia-smi
pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple
梯度对齐：
1.(原版仓库也要设置)设置 StyleGANv2ADAModel 的
self.style_mixing_prob = -1.0
self.align_grad = True
解除上面语句的注释即可。
以及，对下面所有的以if self.align_grad:开头的代码块解除注释
if self.align_grad:
xxx
计算loss_Gpl那里，
pl_noise = torch.randn_like(gen_img) / np.sqrt(gen_img.shape[2] * gen_img.shape[3])
改为
pl_noise = torch.ones_like(gen_img) / np.sqrt(gen_img.shape[2] * gen_img.shape[3])
run_D()方法中，注释掉：
img = self.augment_pipe(img)
解除注释：
# debug_percentile = 0.7
# img = self.augment_pipe(img, debug_percentile)
2.trainer.py下面代码解除注释
# 对齐梯度用
# if self.rank == 0:
# if (self.iter + 1) == 20:
# self.save_ckpt(ckpt_name="%d" % (self.epoch + 1))
3.(原版仓库也要设置)设置 SynthesisLayer 的
self.use_noise = False
4.(原版仓库也要设置)设置 StyleGANv2ADA_SynthesisNetwork 的
use_fp16 = False
5.(原版仓库也要设置)设置 StyleGANv2ADA_Discriminator 的
use_fp16 = False
6. styleganv2ada_method_base.py 优化器要换成SGD：
optimizer = torch.optim.SGD(
itertools.chain(self.model.synthesis.parameters(), self.model.mapping.parameters()), lr=0.001, momentum=0.9
)
...
optimizer = torch.optim.SGD(
self.model.discriminator.parameters(), lr=0.002, momentum=0.9
)
因为Adam更新参数有一定随机性，同样的情况下，跑2次结果不同！！！（但是SGD也有轻微的不同，影响不大。）
第三方实现stylegan2-ada时，不要忘记创建G和D的实例时，都需要设置其的requires_grad_(False)
G = dnnlib.util.construct_class_by_name(**G_kwargs, **common_kwargs).train().requires_grad_(False).to(device) # subclass of torch.nn.Module
D = dnnlib.util.construct_class_by_name(**D_kwargs, **common_kwargs).train().requires_grad_(False).to(device) # subclass of torch.nn.Module
因为第0步训练Gmain阶段时，D的权重应该不允许得到梯度。
而且，除了augment_pipe，其它4个 G.mapping、G.synthesis、D、G_ema 都是DDP模型。
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -c_G G_00.pth -c_Gema G_ema_00.pth -c_D D_00.pth -oc styleganv2ada_32_00.pth
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -c_G G_19.pth -c_Gema G_ema_19.pth -c_D D_19.pth -oc styleganv2ada_32_19.pth
CUDA_VISIBLE_DEVICES=0
python tools/train.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -d 1 -b 8 -eb 1 -c styleganv2ada_32_00.pth
CUDA_VISIBLE_DEVICES=0,1
python tools/train.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -d 2 -b 8 -eb 2 -c styleganv2ada_32_00.pth
CUDA_VISIBLE_DEVICES=0
python tools/train.py -f exps/styleganv2ada/styleganv2ada_32_custom.py --dist-url tcp://192.168.0.104:12319 --num_machines 2 --machine_rank 1 -b 8 -eb 2 -c styleganv2ada_32_00.pth
python diff_weights.py --cp1 styleganv2ada_32_19.pth --cp2 StyleGANv2ADA_outputs/styleganv2ada_32_custom/1.pth --d_value 0.0005
----------------------- 进阶：单卡和多卡进行对齐（用 单卡批大小8 实现 双卡总批大小8 的效果） -----------------------
把 DiscriminatorEpilogue 类的__init__()方法的
# mbstd_num_channels = 0
解除注释，即不使用 MinibatchStdLayer ，因为 MinibatchStdLayer 会有类似BN层的不同图片的数据交流，
使用的话要实现类似同步BN的操作，会加大对齐难度，所以不使用这个层。
除了上述“梯度对齐”的修改外，还需要修改：
1.设置 StyleGANv2ADAModel 的
对下面所有的以if self.align_2gpu_1gpu:开头的代码块解除注释
if self.align_2gpu_1gpu:
xxx
注释掉 accumulate_gradients() 里的：
gen_img, gen_ws = self.run_G(gen_z[:batch_size], gen_c_, sync=sync)
2.(原版仓库也要设置)把 DiscriminatorEpilogue 类的__init__()方法的
# mbstd_num_channels = 0
解除注释
3.把 StyleGANv2ADA_MappingNetwork 类的forward()方法的
self.w_avg.copy_(x.detach().mean(dim=0).lerp(self.w_avg, self.w_avg_beta))
注释掉。把
# bz = x.shape[0]
# gpu0_w_avg = x.detach()[:bz//2].mean(dim=0).lerp(self.w_avg, self.w_avg_beta)
# gpu1_w_avg = x.detach()[bz//2:].mean(dim=0).lerp(self.w_avg, self.w_avg_beta)
# self.w_avg.copy_(gpu0_w_avg)
解除注释。即： 单机多卡训练时，w_avg的更新情况：强制使用 0号gpu 更新后的w_avg 作为整体更新后的w_avg
输入以下命令验证是否已经对齐：
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -c_G G_00.pth -c_Gema G_ema_00.pth -c_D D_00.pth -oc styleganv2ada_32_00.pth
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -c_G G_19.pth -c_Gema G_ema_19.pth -c_D D_19.pth -oc styleganv2ada_32_19.pth
CUDA_VISIBLE_DEVICES=0
python tools/train.py -f exps/styleganv2ada/styleganv2ada_32_custom.py -d 1 -b 8 -eb 1 -c styleganv2ada_32_00.pth
python diff_weights.py --cp1 styleganv2ada_32_19.pth --cp2 StyleGANv2ADA_outputs/styleganv2ada_32_custom/1.pth --d_value 0.0005
----------------------- 进阶：验证每张卡上的训练数据是否不重复 -----------------------
1.exps/styleganv2ada/styleganv2ada_256_custom.py
# 有前16张照片，用来验证每张卡上的训练数据是否不重复
# self.dataroot = '../data/data110820/faces2'
解除注释。
self.print_interval = 10
改成1
self.save_step_interval = 1000
改成2
2.trainer.py下面代码解除注释
# if self.rank == 0:
# logger.info(raw_idx)
# else:
# print(raw_idx)
肉眼观察raw_idx即可(0~15的值)。
单机1卡
python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 1 -b 4 -eb 1 -c styleganv2ada_512_afhqcat.pth
python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 1 -b 4 -eb 1 -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/0_2.pth --resume
单机2卡
CUDA_VISIBLE_DEVICES=0,1
python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 2 -b 4 -eb 2 -c styleganv2ada_512_afhqcat.pth
CUDA_VISIBLE_DEVICES=0,1
python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 2 -b 4 -eb 2 -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/0_2.pth --resume
----------------------- 转换权重 -----------------------
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c_Gema G_ema_afhqcat.pth -c_G G_afhqcat.pth -c_D D_afhqcat.pth -oc styleganv2ada_512_afhqcat.pth
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c_Gema G_ema_metfaces.pth -c_G G_metfaces.pth -c_D D_metfaces.pth -oc styleganv2ada_1024_metfaces.pth
python tools/convert_weights.py -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c_Gema G_ema_ffhq.pth -c_G G_ffhq.pth -c_D D_ffhq.pth -oc styleganv2ada_1024_ffhq.pth
----------------------- 预测 -----------------------
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_128_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_128_custom/48.pth --seeds 85,100,75,458,1500 --save_result --device gpu
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --save_result --device gpu
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --noise_mode random --trunc 0.3 --save_result --device gpu
(trunc==1.0时，StyleGANv2ADA_MappingNetwork的w_avg不参与图像生成；
trunc==0.9时，StyleGANv2ADA_MappingNetwork的w_avg参与图像生成，噪声生成的ws占的比重是0.9，w_avg占的比重是0.1；
trunc==0.0时，StyleGANv2ADA_MappingNetwork的w_avg参与图像生成，噪声生成的ws占的比重是0.0，w_avg占的比重是1.0，即噪声生成的ws不参与图像生成，只用统计的w_avg（ws的平均值，即w_avg妈妈！！！）来生成图像；
其实就是python tools/demo.py A2B那样的插值生成渐变图像，只不过其中1个ws是w_avg
)
(这句代码召唤w_avg妈妈！！！你知道w_avg妈妈有多好看吗！！！她是整个噪声空间经过MappingNetwork之后得到的ws的平均值，倾国倾城都不能形容她的美貌！)
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --noise_mode const --trunc 0.0 --save_result --device gpu
w_avg妈妈！！！
python tools/demo.py image -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --noise_mode const --trunc 0.0 --save_result --device gpu
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_256_uppercloth.py -c StyleGANv2ADA_outputs/styleganv2ada_256_uppercloth/123.pth --seeds 85,100,75,458,1500 --noise_mode const --trunc 0.0 --save_result --device gpu
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 0_1500 --save_result --device gpu
(styleganv3)
python tools/demo.py image -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 0_1500 --save_result --device gpu
(styleganv2ada, uppercloth)
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_256_uppercloth.py -c StyleGANv2ADA_outputs/styleganv2ada_256_uppercloth/123.pth --seeds 0_1500 --save_result --device gpu
(afhq)
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --seeds 85,100,75,458,1500 --save_result --device gpu
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --seeds 0_1500 --save_result --device gpu
w_avg妈妈：
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --seeds 85,100,75,458,1500 --noise_mode const --trunc 0.0 --save_result --device gpu
(afhqv2)
python tools/demo.py image -f exps/styleganv3/styleganv3_r_512_afhqv2.py -c stylegan3_r_afhqv2_512.pth --seeds 85,100,75,458,1500 --save_result --device gpu
(styleganv3_r_1024_ffhq)
python tools/demo.py image -f exps/styleganv3/styleganv3_r_1024_ffhq.py -c stylegan3_r_ffhq_1024.pth --seeds 85,100,75,458,1500 --save_result --device gpu
(styleganv3_t_512_afhqv2)
python tools/demo.py image -f exps/styleganv3/styleganv3_t_512_afhqv2.py -c stylegan3_t_afhqv2_512.pth --seeds 85,100,75,458,1500 --save_result --device gpu
(styleganv3_t_1024_ffhq)
python tools/demo.py image -f exps/styleganv3/styleganv3_t_1024_ffhq.py -c stylegan3_t_ffhq_1024.pth --seeds 85,100,75,458,1500 --save_result --device gpu
(styleganv3_t_1024_metfaces)
python tools/demo.py image -f exps/styleganv3/styleganv3_t_1024_metfaces.py -c stylegan3_t_metfaces_1024.pth --seeds 85,100,75,458,1500 --save_result --device gpu
(metfaces)
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_metfaces.pth --seeds 85,100,75,458,1500 --save_result --device gpu
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_ffhq.pth --seeds 85,100,75,458,1500 --save_result --device gpu
w_avg妈妈：
python tools/demo.py image -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_ffhq.pth --seeds 85,100,75,458,1500 --noise_mode const --trunc 0.0 --save_result --device gpu
----------------------- 渐变，从随机种子A的图像渐变成随机种子B的图像 -----------------------
--frames 表示渐变的帧数。
(动漫头像数据集)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --frames 120 --video_fps 30 --save_result --device gpu
(--A2B_mixing_seed 85 表示这个随机种子生成的ws在每一帧和每一帧的ws进行style_mixing，
--col_styles 0,1,2,3,4,5,6表示A2B_mixing_seed生成的ws提供了动作姿态，每一帧的ws提供了皮肤；
--col_styles 7,8,9,10,11,12,13表示A2B_mixing_seed生成的ws提供了皮肤，每一帧的ws提供了动作姿态；
--A2B_mixing_seed w_avg 表示直接用w_avg妈妈进行style_mixing。
)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed 458 --col_styles 0,1,2,3,4,5,6 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed 458 --col_styles 7,8,9,10,11,12,13 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed 458 --col_styles 2,3,4,5,6,7,8,9,10,11,12,13 --frames 120 --video_fps 30 --save_result --device gpu
(w_avg妈妈提供动作姿态！！！神仙颜值，什么发色瞳色都能驾驭！！！)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed w_avg --col_styles 0,1,2,3,4,5,6 --frames 120 --video_fps 30 --save_result --device gpu
(w_avg妈妈提供皮肤！！！)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed w_avg --col_styles 7,8,9,10,11,12,13 --frames 120 --video_fps 30 --save_result --device gpu
(更低的分辨率使用了w_avg妈妈的潜在因子，人物更像w_avg妈妈！！！)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed w_avg --col_styles 2,3,4,5,6,7,8,9,10,11,12,13 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --seeds 0_20 --frames 120 --video_fps 30 --save_result --device gpu
(styleganv3)
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed 458 --col_styles 0,1,2,3,4,5,6 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed 458 --col_styles 7,8,9,10,11,12,13,14,15 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed 458 --col_styles 2,3,4,5,6,7,8,9,10,11,12,13,14,15 --frames 120 --video_fps 30 --save_result --device gpu
(w_avg妈妈提供动作姿态！！！神仙颜值，什么发色瞳色都能驾驭！！！)
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed w_avg --col_styles 0,1,2,3,4,5,6 --frames 120 --video_fps 30 --save_result --device gpu
(w_avg妈妈提供皮肤！！！)
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed w_avg --col_styles 7,8,9,10,11,12,13,14,15 --frames 120 --video_fps 30 --save_result --device gpu
(更低的分辨率使用了w_avg妈妈的潜在因子，人物更像w_avg妈妈！！！)
python tools/demo.py A2B -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --seeds 85,100,75,458,1500 --A2B_mixing_seed w_avg --col_styles 2,3,4,5,6,7,8,9,10,11,12,13,14,15 --frames 120 --video_fps 30 --save_result --device gpu
(afhq，你会发现stylegan2ada特有的“屏幕粘毛”视觉观感)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --seeds 85,100,75,458,1500 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --seeds 458,1500 --frames 120 --video_fps 30 --save_result --device gpu
(afhqv2，你会发现stylegan3解决了“屏幕粘毛”的问题)
python tools/demo.py A2B -f exps/styleganv3/styleganv3_r_512_afhqv2.py -c stylegan3_r_afhqv2_512.pth --seeds 85,100,75,458,1500 --frames 120 --video_fps 30 --save_result --device gpu
(metfaces)
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_metfaces.pth --seeds 85,100,75,458,1500 --frames 120 --video_fps 30 --save_result --device gpu
python tools/demo.py A2B -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_metfaces.pth --seeds 0_76 --frames 120 --video_fps 30 --save_result --device gpu
----------------------- style_mixing -----------------------
col_styles 0,1,2,3,4,5,6表示的是row_seeds生成的潜在因子的0,1,2,3,4,5,6个被替换成了col_seeds的0,1,2,3,4,5,6个，
即col_seeds提供了动作姿态，row_seeds提供了皮肤。具体解析请看readme_styleganv2ada_note.txt
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_128_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_128_custom/48.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_512_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_512_custom/67.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 7,8,9,10,11,12,13 --save_result --device gpu
(styleganv3)
python tools/demo.py style_mixing -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
(afhq)
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
(afhqv2，如果显存不足)
python tools/demo.py style_mixing -f exps/styleganv3/styleganv3_r_512_afhqv2.py -c stylegan3_r_afhqv2_512.pth --row_seeds 85 --col_seeds 55 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
(metfaces)
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_metfaces.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
python tools/demo.py style_mixing -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_ffhq.pth --row_seeds 85,100,75,458,1500 --col_seeds 55,821,1789,293 --col_styles 0,1,2,3,4,5,6 --save_result --device gpu
----------------------- 获取真实图片的潜在因子ws（真实图片投影到潜在空间） -----------------------
基于优化的方法。从w_avg妈妈开始，渐变成--target_fname指定的图片。
python tools/demo.py projector -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --outdir out --target_fname ./target_imgs/xy.jpg --save_result --device gpu
python tools/demo.py projector -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --outdir out --target_fname ./target_imgs/hqsw.jpg --save_result --device gpu
python tools/demo.py projector -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --outdir out --target_fname ./target_imgs/000138-01.jpg --save_result --device gpu
python tools/demo.py projector -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --outdir out --target_fname ./target_imgs/flickr_cat_000008.jpg --save_result --device gpu
python tools/demo.py projector -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_ffhq.pth --outdir out --target_fname ./target_imgs/xyjy.jpg --save_result --device gpu
(styleganv3)
python tools/demo.py projector -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --outdir out --target_fname ./target_imgs/xy.jpg --save_result --device gpu
python tools/demo.py projector -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --outdir out --target_fname ./target_imgs/hqsw.jpg --save_result --device gpu
python tools/demo.py projector -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --outdir out --target_fname ./target_imgs/000138-01.jpg --save_result --device gpu
----------------------- 训练 -----------------------
后台启动：
nohup xxx > stylegan2ada.log 2>&1 &
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -d 1 -b 8 -eb 1 > stylegan2ada.log 2>&1 &
----------------------- 迁移学习，带上-c（--ckpt）参数读取预训练模型。 -----------------------
后台启动：
nohup xxx > stylegan2ada.log 2>&1 &
迁移学习动漫头像数据集：
CUDA_VISIBLE_DEVICES=0
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 1 -b 6 -eb 1 -c styleganv2ada_512_afhqcat.pth > stylegan2ada.log 2>&1 &
CUDA_VISIBLE_DEVICES=0,1
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 2 -b 8 -eb 2 -c styleganv2ada_512_afhqcat.pth > stylegan2ada.log 2>&1 &
迁移学习上衣256数据集：
CUDA_VISIBLE_DEVICES=0
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_uppercloth.py -d 1 -b 6 -eb 1 -c styleganv2ada_512_afhqcat.pth > stylegan2ada.log 2>&1 &
CUDA_VISIBLE_DEVICES=0,1
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_uppercloth.py -d 2 -b 8 -eb 2 -c styleganv2ada_512_afhqcat.pth > stylegan2ada.log 2>&1 &
----------------------- Linux常用命令 -----------------------
查看日志
cat stylegan2ada.log
查看日志(最后20行)
tail -n 20 stylegan2ada.log
看显存占用、GPU利用率
watch -n 0.1 nvidia-smi
虚拟环境相关：
conda create -n pasta python=3.9
conda activate pasta
export CUDA_VISIBLE_DEVICES=0
export CUDA_VISIBLE_DEVICES=1
export CUDA_VISIBLE_DEVICES=4
----------------------- 恢复训练（加上参数--resume） -----------------------
后台启动：
nohup xxx > stylegan2ada.log 2>&1 &
迁移学习动漫头像数据集：
CUDA_VISIBLE_DEVICES=0
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 1 -b 6 -eb 1 -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/0_1000.pth --resume > stylegan2ada.log 2>&1 &
CUDA_VISIBLE_DEVICES=0,1
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -d 2 -b 8 -eb 2 -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/0_1000.pth --resume > stylegan2ada.log 2>&1 &
迁移学习上衣256数据集：
CUDA_VISIBLE_DEVICES=0
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_uppercloth.py -d 1 -b 6 -eb 1 -c StyleGANv2ADA_outputs/styleganv2ada_256_uppercloth/0_1000.pth --resume > stylegan2ada.log 2>&1 &
CUDA_VISIBLE_DEVICES=0,1
nohup python tools/train.py -f exps/styleganv2ada/styleganv2ada_256_uppercloth.py -d 2 -b 8 -eb 2 -c StyleGANv2ADA_outputs/styleganv2ada_256_uppercloth/0_1000.pth --resume > stylegan2ada.log 2>&1 &
----------------------- 计算指标 -----------------------
转换inceptionv3的权重：
wget https://nvlabs-fi-cdn.nvidia.com/stylegan2-ada-pytorch/pretrained/metrics/inception-2015-12-05.pt
python tools/inception_convert_weights.py
(afhq)
python tools/calc_metrics.py -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth -b 2 -n 50000 --inceptionv3_path inception-2015-12-05.pth --device gpu
(动漫头像数据集)
python tools/calc_metrics.py -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth -b 2 -n 50000 --inceptionv3_path inception-2015-12-05.pth --device gpu
1机1卡，总批大小6，23.pth，
2022-04-29 18:33:53.665 | INFO | __main__:calc_stylegan2ada_metric:230 - total time: 1839.711170s
2022-04-29 18:33:53.665 | INFO | __main__:calc_stylegan2ada_metric:231 - Speed: 0.036794s per image, 27.2 FPS.
2022-04-29 18:33:57.580 | INFO | __main__:calc_stylegan2ada_metric:239 - FID: 14.723749
1机2卡，总批大小8，22.pth，
2022-05-07 10:17:41.038 | INFO | __main__:calc_stylegan2ada_metric:230 - total time: 1744.382321s
2022-05-07 10:17:41.039 | INFO | __main__:calc_stylegan2ada_metric:231 - Speed: 0.034888s per image, 28.7 FPS.
2022-05-07 10:17:45.055 | INFO | __main__:calc_stylegan2ada_metric:239 - FID: 14.092427
1机2卡，总批大小8，32.pth，
2022-05-07 11:08:10.423 | INFO | __main__:calc_stylegan2ada_metric:230 - total time: 1748.136298s
2022-05-07 11:08:10.426 | INFO | __main__:calc_stylegan2ada_metric:231 - Speed: 0.034963s per image, 28.6 FPS.
2022-05-07 11:08:14.463 | INFO | __main__:calc_stylegan2ada_metric:239 - FID: 12.598085
1机2卡，总批大小8，65.pth，
2022-05-09 10:03:12.317 | INFO | __main__:calc_stylegan2ada_metric:230 - total time: 1754.911271s
2022-05-09 10:03:12.317 | INFO | __main__:calc_stylegan2ada_metric:231 - Speed: 0.035098s per image, 28.5 FPS.
2022-05-09 10:03:16.335 | INFO | __main__:calc_stylegan2ada_metric:239 - FID: 8.209035
(styleganv3，1机2卡，总批大小8，40.pth)
python tools/calc_metrics.py -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth -b 2 -n 50000 --inceptionv3_path inception-2015-12-05.pth --device gpu
2022-05-23 15:47:35.642 | INFO | __main__:calc_stylegan2ada_metric:230 - total time: 4849.241369s
2022-05-23 15:47:35.643 | INFO | __main__:calc_stylegan2ada_metric:231 - Speed: 0.096985s per image, 10.3 FPS.
2022-05-23 15:47:39.536 | INFO | __main__:calc_stylegan2ada_metric:239 - FID: 10.792626
1机2卡，总批大小8，77.pth，
2022-05-27 13:59:12.171 | INFO | __main__:calc_stylegan2ada_metric:230 - total time: 5360.057720s
2022-05-27 13:59:12.172 | INFO | __main__:calc_stylegan2ada_metric:231 - Speed: 0.107201s per image, 9.3 FPS.
2022-05-27 13:59:16.578 | INFO | __main__:calc_stylegan2ada_metric:239 - FID: 8.085018
----------------------- 导出为ncnn -----------------------
python tools/demo.py ncnn -f exps/styleganv2ada/styleganv2ada_512_afhqcat.py -c styleganv2ada_512_afhqcat.pth --ncnn_output_path styleganv2ada_512_afhqcat --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv2ada/styleganv2ada_256_custom.py -c StyleGANv2ADA_outputs/styleganv2ada_256_custom/65.pth --ncnn_output_path styleganv2ada_256_custom_epoch_65 --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_metfaces.pth --ncnn_output_path styleganv2ada_1024_metfaces --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv2ada/styleganv2ada_1024_metfaces.py -c styleganv2ada_1024_ffhq.pth --ncnn_output_path styleganv2ada_1024_ffhq --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv3/styleganv3_s_256_custom.py -c StyleGANv3_outputs/styleganv3_s_256_custom/77.pth --ncnn_output_path styleganv3_s_256_custom_epoch_77 --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv3/styleganv3_r_512_afhqv2.py -c stylegan3_r_afhqv2_512.pth --ncnn_output_path stylegan3_r_afhqv2_512 --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv3/styleganv3_r_1024_ffhq.py -c stylegan3_r_ffhq_1024.pth --ncnn_output_path stylegan3_r_ffhq_1024 --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv3/styleganv3_t_512_afhqv2.py -c stylegan3_t_afhqv2_512.pth --ncnn_output_path stylegan3_t_afhqv2_512 --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv3/styleganv3_t_1024_ffhq.py -c stylegan3_t_ffhq_1024.pth --ncnn_output_path stylegan3_t_ffhq_1024 --seeds 0_1500
python tools/demo.py ncnn -f exps/styleganv3/styleganv3_t_1024_metfaces.py -c stylegan3_t_metfaces_1024.pth --ncnn_output_path stylegan3_t_metfaces_1024 --seeds 0_1500
cd build/examples
./stylegan 0 512 16 1.0 seeds/458.bin styleganv2ada_512_afhqcat_mapping.param styleganv2ada_512_afhqcat_mapping.bin styleganv2ada_512_afhqcat_synthesis.param styleganv2ada_512_afhqcat_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin styleganv2ada_512_afhqcat_mapping.param styleganv2ada_512_afhqcat_mapping.bin styleganv2ada_512_afhqcat_synthesis.param styleganv2ada_512_afhqcat_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin styleganv2ada_512_afhqcat_mapping.param styleganv2ada_512_afhqcat_mapping.bin styleganv2ada_512_afhqcat_synthesis.param styleganv2ada_512_afhqcat_synthesis.bin 120 30
./stylegan 0 512 14 1.0 seeds/85.bin styleganv2ada_256_custom_epoch_65_mapping.param styleganv2ada_256_custom_epoch_65_mapping.bin styleganv2ada_256_custom_epoch_65_synthesis.param styleganv2ada_256_custom_epoch_65_synthesis.bin
./stylegan 1 512 14 1.0 seeds/85.bin seeds/100.bin styleganv2ada_256_custom_epoch_65_mapping.param styleganv2ada_256_custom_epoch_65_mapping.bin styleganv2ada_256_custom_epoch_65_synthesis.param styleganv2ada_256_custom_epoch_65_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 14 1.0 seeds/85.bin seeds/100.bin styleganv2ada_256_custom_epoch_65_mapping.param styleganv2ada_256_custom_epoch_65_mapping.bin styleganv2ada_256_custom_epoch_65_synthesis.param styleganv2ada_256_custom_epoch_65_synthesis.bin 120 30
./stylegan 0 512 18 1.0 seeds/458.bin styleganv2ada_1024_metfaces_mapping.param styleganv2ada_1024_metfaces_mapping.bin styleganv2ada_1024_metfaces_synthesis.param styleganv2ada_1024_metfaces_synthesis.bin
./stylegan 1 512 18 1.0 seeds/458.bin seeds/293.bin styleganv2ada_1024_metfaces_mapping.param styleganv2ada_1024_metfaces_mapping.bin styleganv2ada_1024_metfaces_synthesis.param styleganv2ada_1024_metfaces_synthesis.bin 0 1 2 3 4 5 6
./stylegan 0 512 18 1.0 seeds/458.bin styleganv2ada_1024_ffhq_mapping.param styleganv2ada_1024_ffhq_mapping.bin styleganv2ada_1024_ffhq_synthesis.param styleganv2ada_1024_ffhq_synthesis.bin
./stylegan 1 512 18 1.0 seeds/458.bin seeds/293.bin styleganv2ada_1024_ffhq_mapping.param styleganv2ada_1024_ffhq_mapping.bin styleganv2ada_1024_ffhq_synthesis.param styleganv2ada_1024_ffhq_synthesis.bin 0 1 2 3 4 5 6
./stylegan 0 512 16 1.0 seeds/458.bin styleganv3_s_256_custom_epoch_77_mapping.param styleganv3_s_256_custom_epoch_77_mapping.bin styleganv3_s_256_custom_epoch_77_synthesis.param styleganv3_s_256_custom_epoch_77_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin styleganv3_s_256_custom_epoch_77_mapping.param styleganv3_s_256_custom_epoch_77_mapping.bin styleganv3_s_256_custom_epoch_77_synthesis.param styleganv3_s_256_custom_epoch_77_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin styleganv3_s_256_custom_epoch_77_mapping.param styleganv3_s_256_custom_epoch_77_mapping.bin styleganv3_s_256_custom_epoch_77_synthesis.param styleganv3_s_256_custom_epoch_77_synthesis.bin 120 30
./stylegan 0 512 16 1.0 seeds/458.bin stylegan3_r_afhqv2_512_mapping.param stylegan3_r_afhqv2_512_mapping.bin stylegan3_r_afhqv2_512_synthesis.param stylegan3_r_afhqv2_512_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_r_afhqv2_512_mapping.param stylegan3_r_afhqv2_512_mapping.bin stylegan3_r_afhqv2_512_synthesis.param stylegan3_r_afhqv2_512_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_r_afhqv2_512_mapping.param stylegan3_r_afhqv2_512_mapping.bin stylegan3_r_afhqv2_512_synthesis.param stylegan3_r_afhqv2_512_synthesis.bin 120 30
./stylegan 0 512 16 1.0 seeds/458.bin stylegan3_r_ffhq_1024_mapping.param stylegan3_r_ffhq_1024_mapping.bin stylegan3_r_ffhq_1024_synthesis.param stylegan3_r_ffhq_1024_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_r_ffhq_1024_mapping.param stylegan3_r_ffhq_1024_mapping.bin stylegan3_r_ffhq_1024_synthesis.param stylegan3_r_ffhq_1024_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_r_ffhq_1024_mapping.param stylegan3_r_ffhq_1024_mapping.bin stylegan3_r_ffhq_1024_synthesis.param stylegan3_r_ffhq_1024_synthesis.bin 120 30
./stylegan 0 512 16 1.0 seeds/458.bin stylegan3_t_afhqv2_512_mapping.param stylegan3_t_afhqv2_512_mapping.bin stylegan3_t_afhqv2_512_synthesis.param stylegan3_t_afhqv2_512_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_t_afhqv2_512_mapping.param stylegan3_t_afhqv2_512_mapping.bin stylegan3_t_afhqv2_512_synthesis.param stylegan3_t_afhqv2_512_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_t_afhqv2_512_mapping.param stylegan3_t_afhqv2_512_mapping.bin stylegan3_t_afhqv2_512_synthesis.param stylegan3_t_afhqv2_512_synthesis.bin 120 30
./stylegan 0 512 16 1.0 seeds/458.bin stylegan3_t_ffhq_1024_mapping.param stylegan3_t_ffhq_1024_mapping.bin stylegan3_t_ffhq_1024_synthesis.param stylegan3_t_ffhq_1024_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_t_ffhq_1024_mapping.param stylegan3_t_ffhq_1024_mapping.bin stylegan3_t_ffhq_1024_synthesis.param stylegan3_t_ffhq_1024_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_t_ffhq_1024_mapping.param stylegan3_t_ffhq_1024_mapping.bin stylegan3_t_ffhq_1024_synthesis.param stylegan3_t_ffhq_1024_synthesis.bin 120 30
./stylegan 0 512 16 1.0 seeds/458.bin stylegan3_t_metfaces_1024_mapping.param stylegan3_t_metfaces_1024_mapping.bin stylegan3_t_metfaces_1024_synthesis.param stylegan3_t_metfaces_1024_synthesis.bin
./stylegan 1 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_t_metfaces_1024_mapping.param stylegan3_t_metfaces_1024_mapping.bin stylegan3_t_metfaces_1024_synthesis.param stylegan3_t_metfaces_1024_synthesis.bin 0 1 2 3 4 5 6
./stylegan 2 512 16 1.0 seeds/458.bin seeds/293.bin stylegan3_t_metfaces_1024_mapping.param stylegan3_t_metfaces_1024_mapping.bin stylegan3_t_metfaces_1024_synthesis.param stylegan3_t_metfaces_1024_synthesis.bin 120 30
----------------------- 导出为TensorRT -----------------------