Datawhale X 魔搭 AI夏令营第四期 魔搭-AIGC方向 task01笔记
赛题:可图Kolors-LoRA风格故事挑战赛
赛事链接:https://tianchi.aliyun.com/s/ce4dc8bf800db1e58d51263ff357d28f
- 参赛者需在可图Kolors 模型的基础上训练LoRA 模型,生成无限风格,如水墨画风格、水彩风格、赛博朋克风格、日漫风格…
- 基于LoRA模型生成 8 张图片组成连贯故事,故事内容可自定义;基于8图故事,评估LoRA风格的美感度及连贯性
样例:偶像少女养成日记
文生图基础知识
文生图主要以SD系列基础模型为主,以及在其基础上微调的lora模型和人物基础模型等。
接下来,我们简单了解下提示词、lora、ComfyUI和参考图控制这些知识点。
提示词
提示词很重要,一般写法:主体描述,细节描述,修饰词,艺术风格,艺术家
举个例子:
- 【promts】Beautiful and cute girl, smiling, 16 years old, denim jacket, gradient background, soft colors, soft lighting, cinematic edge lighting, light and dark contrast, anime, super detail, 8k
- 【负向prompts】(lowres, low quality, worst quality:1.2), (text:1.2), deformed, black and white,disfigured, low contrast, cropped, missing fingers
Lora
Stable Diffusion中的Lora(LoRA)模型是一种轻量级的微调方法,它代表了“Low-Rank Adaptation”,即低秩适应。Lora不是指单一的具体模型,而是指一类通过特定微调技术应用于基础模型的扩展应用。在Stable Diffusion这一文本到图像合成模型的框架下,Lora被用来对预训练好的大模型进行针对性优化,以实现对特定主题、风格或任务的精细化控制。
ComfyUI
ComfyUI 是一个工作流工具,主要用于简化和优化 AI 模型的配置和训练过程。通过直观的界面和集成的功能,用户可以轻松地进行模型微调、数据预处理、图像生成等任务,从而提高工作效率和生成效果。
在ComfyUI平台的前端页面上,用户可以基于节点/流程图的界面设计并执行AIGC文生图或者文生视频的pipeline。
参考图控制
ControlNet是一种用于精确控制图像生成过程的技术组件。它是一个附加到预训练的扩散模型(如Stable Diffusion模型)上的可训练神经网络模块。扩散模型通常用于从随机噪声逐渐生成图像的过程,而ControlNet的作用在于引入额外的控制信号,使得用户能够更具体地指导图像生成的各个方面(如姿势关键点、分割图、深度图、颜色等)。
速通指南
Datawhale官方的速通教程链接:Task 1 从零入门AI生图原理&实践
环境配置
参考官方教程即可
体验一站式baseline
- 环境安装
!pip install simple-aesthetics-predictor
!pip install -v -e data-juicer
!pip uninstall pytorch-lightning -y
!pip install peft lightning pandas torchvision
!pip install -e DiffSynth-Studio
- 下载数据集
#下载数据集
from modelscope.msdatasets import MsDataset
ds = MsDataset.load(
'AI-ModelScope/lowres_anime',
subset_name='default',
split='train',
cache_dir="/mnt/workspace/kolors/data"
)
import json, os
from data_juicer.utils.mm_utils import SpecialTokens
from tqdm import tqdm
os.makedirs("./data/lora_dataset/train", exist_ok=True)
os.makedirs("./data/data-juicer/input", exist_ok=True)
with open("./data/data-juicer/input/metadata.jsonl", "w") as f:
for data_id, data in enumerate(tqdm(ds)):
image = data["image"].convert("RGB")
image.save(f"/mnt/workspace/kolors/data/lora_dataset/train/{data_id}.jpg")
metadata = {"text": "二次元", "image": [f"/mnt/workspace/kolors/data/lora_dataset/train/{data_id}.jpg"]}
f.write(json.dumps(metadata))
f.write("\n")
- 处理数据集,保存数据处理结果
data_juicer_config = """
# global parameters
project_name: 'data-process'
dataset_path: './data/data-juicer/input/metadata.jsonl' # path to your dataset directory or file
np: 4 # number of subprocess to process your dataset
text_keys: 'text'
image_key: 'image'
image_special_token: '<__dj__image>'
export_path: './data/data-juicer/output/result.jsonl'
# process schedule
# a list of several process operators with their arguments
process:
- image_shape_filter:
min_width: 1024
min_height: 1024
any_or_all: any
- image_aspect_ratio_filter:
min_ratio: 0.5
max_ratio: 2.0
any_or_all: any
"""
with open("data/data-juicer/data_juicer_config.yaml", "w") as file:
file.write(data_juicer_config.strip())
!dj-process --config data/data-juicer/data_juicer_config.yaml
import pandas as pd
import os, json
from PIL import Image
from tqdm import tqdm
texts, file_names = [], []
os.makedirs("./data/lora_dataset_processed/train", exist_ok=True)
with open("./data/data-juicer/output/result.jsonl", "r") as file:
for data_id, data in enumerate(tqdm(file.readlines())):
data = json.loads(data)
text = data["text"]
texts.append(text)
image = Image.open(data["image"][0])
image_path = f"./data/lora_dataset_processed/train/{data_id}.jpg"
image.save(image_path)
file_names.append(f"{data_id}.jpg")
data_frame = pd.DataFrame()
data_frame["file_name"] = file_names
data_frame["text"] = texts
data_frame.to_csv("./data/lora_dataset_processed/train/metadata.csv", index=False, encoding="utf-8-sig")
data_frame
- lora微调
# 下载模型
from diffsynth import download_models
download_models(["Kolors", "SDXL-vae-fp16-fix"])
#模型训练
import os
cmd = """
python DiffSynth-Studio/examples/train/kolors/train_kolors_lora.py \
--pretrained_unet_path models/kolors/Kolors/unet/diffusion_pytorch_model.safetensors \
--pretrained_text_encoder_path models/kolors/Kolors/text_encoder \
--pretrained_fp16_vae_path models/sdxl-vae-fp16-fix/diffusion_pytorch_model.safetensors \
--lora_rank 16 \
--lora_alpha 4.0 \
--dataset_path data/lora_dataset_processed \
--output_path ./models \
--max_epochs 1 \
--center_crop \
--use_gradient_checkpointing \
--precision "16-mixed"
""".strip()
os.system(cmd)
- 加载微调好的模型
from diffsynth import ModelManager, SDXLImagePipeline
from peft import LoraConfig, inject_adapter_in_model
import torch
def load_lora(model, lora_rank, lora_alpha, lora_path):
lora_config = LoraConfig(
r=lora_rank,
lora_alpha=lora_alpha,
init_lora_weights="gaussian",
target_modules=["to_q", "to_k", "to_v", "to_out"],
)
model = inject_adapter_in_model(lora_config, model)
state_dict = torch.load(lora_path, map_location="cpu")
model.load_state_dict(state_dict, strict=False)
return model
# Load models
model_manager = ModelManager(torch_dtype=torch.float16, device="cuda",
file_path_list=[
"models/kolors/Kolors/text_encoder",
"models/kolors/Kolors/unet/diffusion_pytorch_model.safetensors",
"models/kolors/Kolors/vae/diffusion_pytorch_model.safetensors"
])
pipe = SDXLImagePipeline.from_model_manager(model_manager)
# Load LoRA
pipe.unet = load_lora(
pipe.unet,
lora_rank=16, # This parameter should be consistent with that in your training script.
lora_alpha=2.0, # lora_alpha can control the weight of LoRA.
lora_path="models/lightning_logs/version_0/checkpoints/epoch=0-step=500.ckpt"
)
- 图片生成
torch.manual_seed(0)
image = pipe(
prompt="二次元,一个紫色短发小女孩,在家中沙发上坐着,双手托着腮,很无聊,全身,粉色连衣裙",
negative_prompt="丑陋、变形、嘈杂、模糊、低对比度",
cfg_scale=4,
num_inference_steps=50, height=1024, width=1024,
)
image.save("1.jpg")