EvoMakeup: High-Fidelity and Controllable Makeup Editing with MakeupQuad

TL;DR

EvoMakeup introduces a high-quality dataset and a unified framework for realistic, controllable makeup editing that preserves identity and detail, outperforming prior methods on real-world benchmarks.

Contribution

The paper presents MakeupQuad, a large-scale dataset, and EvoMakeup, a novel training framework enabling high-fidelity, multi-task makeup editing with controllability and generalization to real-world images.

Findings

EvoMakeup achieves superior makeup fidelity and identity preservation.

The method generalizes well to real-world images despite training on synthetic data.

Supports multi-task, controllable makeup editing including text-driven and partial edits.

Abstract

Facial makeup editing aims to realistically transfer makeup from a reference to a target face. Existing methods often produce low-quality results with coarse makeup details and struggle to preserve both identity and makeup fidelity, mainly due to the lack of structured paired data -- where source and result share identity, and reference and result share identical makeup. To address this, we introduce MakeupQuad, a large-scale, high-quality dataset with non-makeup faces, references, edited results, and textual makeup descriptions. Building on this, we propose EvoMakeup, a unified training framework that mitigates image degradation during multi-stage distillation, enabling iterative improvement of both data and model quality. Although trained solely on synthetic data, EvoMakeup generalizes well and outperforms prior methods on real-world benchmarks. It supports high-fidelity, controllable, multi-task makeup editing -- including full-face and partial reference-based editing, as well as text-driven makeup editing -- within a single model. Experimental results demonstrate that our method achieves superior makeup fidelity and identity preservation, effectively balancing both aspects. Code and dataset will be released upon acceptance.