• Hussain Hamid
• Dr Manisha Bharti

Recent advances in generative AI and facial landmark detection have enabled sophisticated face-swapping techniques, but challenges remain in achieving seamless blending and user-controlled generation. This paper proposes FaceMorpher, a novel framework that integrates Stable Diffusion-based image synthesis with dlib-powered geometric morphing to generate high-fidelity face-swapped images. Our method first leverages a pretrained Stable Diffusion 2.1 model to create realistic facial images from text prompts, then employs a landmark-based alignment pipeline to warp and blend facial features between source and target images. Key innovations include: (1) a multi-feature convex hull masking technique to isolate facial regions (eyes, nose, mouth) for precise blending; (2) an affine transformation optimized via Singular Value Decomposition (SVD) for pose-invariant warping; and (3) Gaussian blur-based color correction to enhance realism. Experiments on diverse facial datasets demonstrate that FaceMorpher outperforms traditional GAN-based face-swapping methods in perceptual quality (measured via SSIM and user studies) while maintaining computational efficiency. The system’s modular design supports applications in entertainment, privacy preservation, and digital art, offering a flexible balance between automation and user control.

Face morphing, Stable Diffusion, landmark detection, image warping, generative AI, dlib.

"FaceMorpher: A Hybrid Deep Learning and Geometric Approach for Realistic Facial Image Generation and Morphing", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 6, page no.g405-g409, June-2025, Available :http://www.jetir.org/papers/JETIR2506650.pdf

"FaceMorpher: A Hybrid Deep Learning and Geometric Approach for Realistic Facial Image Generation and Morphing", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 6, page no. ppg405-g409, June-2025, Available at : http://www.jetir.org/papers/JETIR2506650.pdf