---
license: mit
task_categories:
- text-to-image
tags:
- art
size_categories:
- 10K<n<100K
---

# Introduction

MangaZero dataset from paper [DiffSensei: Bridging Multi-Modal LLMs and Diffusion Models for Customized Manga Generation](https://arxiv.org/abs/2412.07589)

Please see [GitHub repo](https://github.com/jianzongwu/DiffSensei) to get the usage

Project page: https://jianzongwu.github.io/projects/diffsensei

# The "type" key in character annotation

Type = 0: Characters with clean faces

Type = 1: Characters with vague faces

Type = 2: Exaggerated, artistic characters with appearances that don't conform to typical standards.