Abstract:While current makeup transfer methods have made progress in realism and color fidelity, they struggle with capturing texture details and producing high-resolution images, limiting their practical utility. To address these challenges, we propose a Fine Rendering High-resolution Makeup Transfer (FRHMT) network, which leverages a powerful style-based generator and introduces a novel inversion-editing strategy tailored for makeup transfer. Concretely, in the inversion phase, considering the semantic decoupling properties in the latent space, we design a Hierarchical Residual Inversion (HRI), which projects the image onto high-dimensional feature maps in coarse layers and low-dimensional style codes in fine layers. This design effectively restores the content information of the image while maintaining flexibility in editing the makeup styles. In the editing phase, the Makeup Modulation Module (MMM) learns two mapping networks to adjust the latent variables of the source image based on those of the reference image. This modification occurs in fine layers to transfer the makeup information and preserve the content information. With new network structures and customized loss functions, our training eliminates cumbersome pseudo-paired data synthesis and unstable adversarial learning. Extensive experiments have demonstrated that our method outperforms existing methods in both image quality and makeup similarity through quantitative and qualitative analysis. Additionally, we address the lack of high-resolution data by collecting a dataset of 9716 face images with a resolution of 1024 × 1024. In conclusion, our framework offers a novel artificial intelligence (AI) implementation of makeup transfer in engineering, with the collected dataset holding substantial value for further advancements in AI research.

Reference-conditional Makeup-aware Discrimination for Face Image Beautification

MakeupDiffuse: a double image-controlled diffusion model for exquisite makeup transfer

A Fine Rendering High-Resolution Makeup Transfer network via inversion-editing strategy

BeautyREC: Robust, Efficient, and Component-Specific Makeup Transfer

Makeup like a superstar: Deep Localized Makeup Transfer Network

BeautyREC: Robust, Efficient, and Content-preserving Makeup Transfer

Facial Makeup Transfer Combining Illumination Transfer

Personalized facial makeup transfer based on outline correspondence

MuNeRF: Robust Makeup Transfer in Neural Radiance Fields

Lipstick ain't enough: Beyond Color Matching for In-the-Wild Makeup Transfer

Stable-Makeup: When Real-World Makeup Transfer Meets Diffusion Model

Local Facial Makeup Transfer via Disentangled Representation

High Fidelity Makeup Via 2D and 3D Identity Preservation Net

DA-GAN: Dual-attention Generative Adversarial Networks for Real-World Exquisite Makeup Transfer

RAMT-GAN: Realistic and accurate makeup transfer with generative adversarial network

DRAN: Detailed Region-Adaptive Normalization for Conditional Image Synthesis

Semi-Supervised Eye Makeup Transfer By Swapping Learned Representation

Composite makeup transfer model based on generative adversarial networks

Content-Style Decoupling for Unsupervised Makeup Transfer without Generating Pseudo Ground Truth

Makeup Removal for Face Verification Based Upon Contrastive Learning