import torch from torch import nn import torch.optim as optim import torch.nn.functional as F from torch.utils.data.dataloader import DataLoader from torchvision import transforms from torchvision import utils as vutils import argparse from tqdm import tqdm from models import weights_init, Discriminator, Generator from operation import copy_G_params, load_params, get_dir from operation import ImageFolder, InfiniteSamplerWrapper from diffaug import DiffAugment ndf = 64 ngf = 64 nz = 256 nlr = 0.0002 nbeta1 = 0.5 use_cuda = True multi_gpu = False dataloader_workers = 8 current_iteration = 0 save_interval = 100 device = 'cuda:0' im_size = 256 netG = Generator(ngf=ngf, nz=nz, im_size=im_size) netG.apply(weights_init) netD = Discriminator(ndf=ndf, im_size=im_size) netD.apply(weights_init) netG.to(device) netD.to(device) avg_param_G = copy_G_params(netG) fixed_noise = torch.FloatTensor(8, nz).normal_(0, 1).to(device) optimizerG = optim.Adam(netG.parameters(), lr=nlr, betas=(nbeta1, 0.999)) optimizerD = optim.Adam(netD.parameters(), lr=nlr, betas=(nbeta1, 0.999)) j = 4 checkpoint = "./models/all_%d.pth"%(j*10000) ckpt = torch.load(checkpoint) netG.load_state_dict(ckpt['g']) netD.load_state_dict(ckpt['d']) avg_param_G = ckpt['g_ema'] load_params(netG, avg_param_G) bs = 8 noise_a = torch.randn(bs, nz).to(device) noise_b = torch.randn(bs, nz).to(device) def get_early_features(net, noise): feat_4 = net.init(noise) feat_8 = net.feat_8(feat_4) feat_16 = net.feat_16(feat_8) feat_32 = net.feat_32(feat_16) feat_64 = net.feat_64(feat_32) return feat_8, feat_16, feat_32, feat_64 def get_late_features(net, im_size, feat_64, feat_8, feat_16, feat_32): feat_128 = net.feat_128(feat_64) feat_128 = net.se_128(feat_8, feat_128) feat_256 = net.feat_256(feat_128) feat_256 = net.se_256(feat_16, feat_256) if im_size==256: return net.to_big(feat_256) feat_512 = net.feat_512(feat_256) feat_512 = net.se_512(feat_32, feat_512) if im_size==512: return net.to_big(feat_512) feat_1024 = net.feat_1024(feat_512) return net.to_big(feat_1024) feat_8_a, feat_16_a, feat_32_a, feat_64_a = get_early_features(netG, noise_a) feat_8_b, feat_16_b, feat_32_b, feat_64_b = get_early_features(netG, noise_b) images_b = get_late_features(netG, im_size, feat_64_b, feat_8_b, feat_16_b, feat_32_b) images_a = get_late_features(netG, im_size, feat_64_a, feat_8_a, feat_16_a, feat_32_a) imgs = [ torch.ones(1, 3, im_size, im_size) ] imgs.append(images_b.cpu()) for i in range(bs): imgs.append(images_a[i].unsqueeze(0).cpu()) gimgs = get_late_features(netG, im_size, feat_64_a[i].unsqueeze(0).repeat(bs, 1, 1, 1), feat_8_b, feat_16_b, feat_32_b) imgs.append(gimgs.cpu()) imgs = torch.cat(imgs) vutils.save_image(imgs.add(1).mul(0.5), 'style_mix_1.jpg', nrow=bs+1)