top of page

In-Depth Analysis of AI-Generated Portraits:
A Comparative Study of Text-to-Image Models

In my Master's thesis, I explored the effectiveness of text-to-image generation models in creating realistic portraits, focusing on DALL-E 3, Midjourney Version 6, and Stable Diffusion 2. The research combines a detailed analysis of AI image generation technology with a mixed-methods approach, including a literature review and primary data collected from over 300 survey participants. I delve into the underlying mechanics of AI image generation, examining how these models interpret text prompts to generate visual content. The study also highlights the strengths and weaknesses of each tool, providing a comparative analysis of their ability to produce realistic portraits. By generating 60 images and assessing participants' ability to distinguish between real and AI-generated faces, the research offers valuable insights into the potential and limitations of these models, along with recommendations for future development.

DISCLAIMER:
(a) Displayed brand logos and brand names in the portfolio and CV, represent past client work and are showcased solely for portfolio and resume validation. They do not imply direct endorsement or affiliation. (b) Certain visuals used for the portfolio include copyrighted stock images and videos. All the visual media which includes images, GIFs, videos and texts are strictly for portfolio display and must not be used or reproduced without proper authorization from the copyright owners.

bottom of page