StableMorph: High-Quality Face Morph Generation with Stable Diffusion

TL;DR

StableMorph tackles the challenge of realistic, hard-to-detect face morphs by combining diffusion-based synthesis with a four-stage pipeline that fine-tunes a Stable Diffusion model per subject using LoRA, merges identity embeddings via SLERP, and generates full-head morphs conditioned with an IP-Adapter. The approach achieves superior visual fidelity, maintains robust attack potential, and outperforms baselines on FIQA and IQA metrics while delivering competitive MAP, enabling more rigorous MAD evaluation. By providing high-quality morph datasets and a controllable generation framework, StableMorph facilitates the development of more robust biometric defenses and realistic operational testing. This work advances biometric security research by raising the bar for morph realism and realism-based evaluation in MAD research and practical deployments.

Abstract

Face morphing attacks threaten the integrity of biometric identity systems by enabling multiple individuals to share a single identity. To develop and evaluate effective morphing attack detection (MAD) systems, we need access to high-quality, realistic morphed images that reflect the challenges posed in real-world scenarios. However, existing morph generation methods often produce images that are blurry, riddled with artifacts, or poorly constructed making them easy to detect and not representative of the most dangerous attacks. In this work, we introduce StableMorph, a novel approach that generates highly realistic, artifact-free morphed face images using modern diffusion-based image synthesis. Unlike prior methods, StableMorph produces full-head images with sharp details, avoids common visual flaws, and offers unmatched control over visual attributes. Through extensive evaluation, we show that StableMorph images not only rival or exceed the quality of genuine face images but also maintain a strong ability to fool face recognition systems posing a greater challenge to existing MAD solutions and setting a new standard for morph quality in research and operational testing. StableMorph improves the evaluation of biometric security by creating more realistic and effective attacks and supports the development of more robust detection systems.

Figures (10)

• Figure 1: Sample morphed images generated by StableMorph. The 1st and 3rd columns are the bona fide images of the subjects. Images from the FRLL dataset DeBruine2021.
• Figure 2: Comparison of morphed images from the different methods. Images from FRLL are in the top two rows, and from FRGC in the bottom two ones. A nice feature of StableMorph is that it is not influenced by the accent color in the original images.
• Figure 3: Morph Attack Potential results on FRLL assessed with three verification attempts and four FRSs with a fixed FMR at 0.1%.
• Figure 4: Morph Attack Potential results on FRGC assessed with three verification attempts and four FRSs with a fixed FMR at 0.1%.
• Figure 5: Quality value distributions of the two FIQA measures: sharpness and UQS (MagFace) on FRLL and FRGC. StableMorph clearly has the best sharpness and face image quality of all other methods, including the SoTA MorDiff method.