This thesis investigates the use of diffusion-based generative AI for realistic headshot generation, employing Stable Diffusion as the core framework and extending it with conditioning models to improve control and quality. A practical pipeline was developed using Stable Diffusion 1.5 with ControlNet and IP-Adapter FaceID Plus v2 for structure- and identity-aware synthesis, combined with InsightFace-based ReActor for face alignment and swapping.

This thesis investigates the use of diffusion-based generative AI for realistic headshot generation, employing Stable Diffusion as the core framework and extending it with conditioning models to improve control and quality. A practical pipeline was developed using Stable Diffusion 1.5 with ControlNet and IP-Adapter FaceID Plus v2 for structure- and identity-aware synthesis, combined with InsightFace-based ReActor for face alignment and swapping.

AI-Driven Headshot Generation Using Stable Diffusion Models

HEDAYATI, ATEFEH
2024/2025

Abstract

This thesis investigates the use of diffusion-based generative AI for realistic headshot generation, employing Stable Diffusion as the core framework and extending it with conditioning models to improve control and quality. A practical pipeline was developed using Stable Diffusion 1.5 with ControlNet and IP-Adapter FaceID Plus v2 for structure- and identity-aware synthesis, combined with InsightFace-based ReActor for face alignment and swapping.
2024
AI-Driven Headshot Generation Using Stable Diffusion Models
This thesis investigates the use of diffusion-based generative AI for realistic headshot generation, employing Stable Diffusion as the core framework and extending it with conditioning models to improve control and quality. A practical pipeline was developed using Stable Diffusion 1.5 with ControlNet and IP-Adapter FaceID Plus v2 for structure- and identity-aware synthesis, combined with InsightFace-based ReActor for face alignment and swapping.
generative AI
diffusion models
image generation
face recognition
data augmentation
MILANI, SIMONE
Lauree magistrali
File in questo prodotto:
File Dimensione Formato  
Hedayati_Atefeh.pdf

Accesso riservato

Dimensione 7.41 MB
Formato Adobe PDF
7.41 MB Adobe PDF

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/102113