驾驭AI绘画：《AI魔法绘画》带你秒变顶级画手！

大家好，我是herosunly。985院校硕士毕业，现担任算法研究员一职，热衷于机器学习算法研究与应用。曾获得阿里云天池比赛第一名，CCF比赛第二名，科大讯飞比赛第三名。拥有多项发明专利。对机器学习和深度学习拥有自己独到的见解。曾经辅导过若干个非计算机专业的学生进入到算法行业就业。希望和大家一起成长进步。

今天给大家带来的文章是驾驭AI绘画：《AI魔法绘画》带你秒变顶级画手！，希望能对学习AI绘画的同学们有所帮助。

文章目录

1. 前言
2. 书籍推荐《AI魔法绘画：用Stable Diffusion挑战无限可能》
3. 粉丝福利
4. 自主购买

1. 前言

同学们，抖音买家秀中的那些极具设计感的作品是怎么来的？

你还在羡慕别人的绘画天赋，对着空白的画纸发呆吗？

有没有想过将生硬的代码和灵动的艺术完美融合？

有没有感到现代技术带来了戏剧性改变，从替代马车的汽车，到取代传统影像的数字摄影，再到今天的AI美术创作？

是的，“AI绘画”，它来了。

AI绘画，如雷贯耳，无处不在。

曾几何时，人们认为艺术需要人工创作，需要才情与灵感的瞬间迸发。

然而，技术的高速发展，打破了这种观念，AI绘画正在逐渐崭露头角，成为新的画坛巨擘。

但聊一聊看似高大上的AI技术可能让你觉得高不可攀。那就跟我一起，用Stable Diffusion挑战无限可能，一窥AI绘画的花花世界！

Stable Diffusion WebUI 中的核心组件，人脸图像面部画面修复模型 CodeFormer的核心代码如下所示：

import mathimport torchfrom torch import nn, Tensorimport torch.nn.functional as Ffrom typing import Optionalfrom modules.codeformer.vqgan_arch import VQAutoEncoder, ResBlockfrom basicsr.utils.registry import ARCH_REGISTRYclass CodeFormer(VQAutoEncoder):def __init__(self, dim_embd=512, n_head=8, n_layers=9,codebook_size=1024, latent_size=256,connect_list=('32', '64', '128', '256'),fix_modules=('quantize', 'generator')):super(CodeFormer, self).__init__(512, 64, [1, 2, 2, 4, 4, 8], 'nearest',2, [16], codebook_size)if fix_modules is not None:for module in fix_modules:for param in getattr(self, module).parameters():param.requires_grad = Falseself.connect_list = connect_listself.n_layers = n_layersself.dim_embd = dim_embdself.dim_mlp = dim_embd*2self.position_emb = nn.Parameter(torch.zeros(latent_size, self.dim_embd))self.feat_emb = nn.Linear(256, self.dim_embd)# transformerself.ft_layers = nn.Sequential(*[TransformerSALayer(embed_dim=dim_embd, nhead=n_head, dim_mlp=self.dim_mlp, dropout=0.0)for _ in range(self.n_layers)])# logits_predict headself.idx_pred_layer = nn.Sequential(nn.LayerNorm(dim_embd),nn.Linear(dim_embd, codebook_size, bias=False))self.channels = {'16': 512,'32': 256,'64': 256,'128': 128,'256': 128,'512': 64,}# after second residual block for > 16, before attn layer for ==16self.fuse_encoder_block = {'512':2, '256':5, '128':8, '64':11, '32':14, '16':18}# after first residual block for > 16, before attn layer for ==16self.fuse_generator_block = {'16':6, '32': 9, '64':12, '128':15, '256':18, '512':21}# fuse_convs_dictself.fuse_convs_dict = nn.ModuleDict()for f_size in self.connect_list:in_ch = self.channels[f_size]self.fuse_convs_dict[f_size] = Fuse_sft_block(in_ch, in_ch)def _init_weights(self, module):if isinstance(module, (nn.Linear, nn.Embedding)):module.weight.data.normal_(mean=0.0, std=0.02)if isinstance(module, nn.Linear) and module.bias is not None:module.bias.data.zero_()elif isinstance(module, nn.LayerNorm):module.bias.data.zero_()module.weight.data.fill_(1.0)def forward(self, x, w=0, detach_16=True, code_only=False, adain=False):# ################### Encoder #####################enc_feat_dict = {}out_list = [self.fuse_encoder_block[f_size] for f_size in self.connect_list]for i, block in enumerate(self.encoder.blocks):x = block(x)if i in out_list:enc_feat_dict[str(x.shape[-1])] = x.clone()lq_feat = x# ################# Transformer #################### quant_feat, codebook_loss, quant_stats = self.quantize(lq_feat)pos_emb = self.position_emb.unsqueeze(1).repeat(1,x.shape[0],1)# BCHW -> BC(HW) -> (HW)BCfeat_emb = self.feat_emb(lq_feat.flatten(2).permute(2,0,1))query_emb = feat_emb# Transformer encoderfor layer in self.ft_layers:query_emb = layer(query_emb, query_pos=pos_emb)# output logitslogits = self.idx_pred_layer(query_emb) # (hw)bnlogits = logits.permute(1,0,2) # (hw)bn -> b(hw)nif code_only: # for training stage II# logits doesn't need softmax before cross_entropy lossreturn logits, lq_feat# ################# Quantization #################### if self.training:# quant_feat = torch.einsum('btn,nc->btc', [soft_one_hot, self.quantize.embedding.weight])# # b(hw)c -> bc(hw) -> bchw# quant_feat = quant_feat.permute(0,2,1).view(lq_feat.shape)# ------------soft_one_hot = F.softmax(logits, dim=2)_, top_idx = torch.topk(soft_one_hot, 1, dim=2)quant_feat = self.quantize.get_codebook_feat(top_idx, shape=[x.shape[0],16,16,256])# preserve gradients# quant_feat = lq_feat + (quant_feat - lq_feat).detach()if detach_16:quant_feat = quant_feat.detach() # for training stage IIIif adain:quant_feat = adaptive_instance_normalization(quant_feat, lq_feat)# ################## Generator ####################x = quant_featfuse_list = [self.fuse_generator_block[f_size] for f_size in self.connect_list]for i, block in enumerate(self.generator.blocks):x = block(x)if i in fuse_list: # fuse after i-th blockf_size = str(x.shape[-1])if w>0:x = self.fuse_convs_dict[f_size](enc_feat_dict[f_size].detach(), x, w)out = x# logits doesn't need softmax before cross_entropy lossreturn out, logits, lq_feat