Personalizing Text-to-Image Generation to Individual Taste

Anne-Sofie Maerten Juliane Verwiebe, Shyamgopal Karthik Ameya Prabhu Johan Wagemans Matthias Bethge · 2026

Read Personalizing Text-to-Image Generation to Individual Taste by Anne-Sofie Maerten Juliane Verwiebe, Shyamgopal Karthik Ameya Prabhu Johan Wagemans Matthias Bethge (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-07427.txt.

Description

Maerten and five colleagues build PAMELA, a dataset and model for tailoring image generation to personal taste. Average-appeal models miss how subjective beauty is, so the dataset holds 70000 ratings on 5000 images, with 15 raters per image. A personal reward model trained on these ratings plus prior aesthetic data predicts individual liking better than most current methods predict group taste. Simple prompt tweaks then steer generations toward each user's own preferences.

Catalog record

CREATORAnne-Sofie Maerten Juliane Verwiebe, Shyamgopal Karthik Ameya Prabhu Johan Wagemans Matthias Bethge
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48755
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.07427

Browse the full library · About the Tower of Babel