News-to-Meme Generation: Performance Analysis of GPT-3, BLIP and CLIP+GPT Using BBC News Headlines

P. Naik, Pratiksha and Jayabalan, Bhuvana and Priya, Mohana (2026) News-to-Meme Generation: Performance Analysis of GPT-3, BLIP and CLIP+GPT Using BBC News Headlines. INTI JOURNAL, 2026 (30). pp. 251-257. ISSN e2600-7320

	Text ij2026_30.pdf - Published Version Available under License Creative Commons Attribution. Download (205kB)
	Text 922 - Published Version Available under License Creative Commons Attribution. Download (26kB)

Abstract

Making memes has become a powerful way to share cultural and funny commentary. This study assesses the efficacy of AI models in autonomously creating memes from real-world data. The comparative study of three prominent models—GPT-3, BLIP, and CLIP+GPT—utilized 500 BBC news headlines as input prompts. Using a consistent GPT-based evaluation framework, generated memes are rated on five scales: Humor, Relevance, Engagement, Creativity, and Clarity. The findings indicate that GPT-3 excels in Humor and Relevance, CLIP+GPT is superior in Engagement, and BLIP offers balanced results.

Item Type:	Article
Uncontrolled Keywords:	Meme Generation, Multimodal AI, GPT-3, BLIP, CLIP, Text-to-Image
Subjects:	Q Science > Q Science (General) Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software
Depositing User:	Unnamed user with email masilah.mansor@newinti.edu.my
Date Deposited:	03 Sep 2026 08:48
Last Modified:	03 Sep 2026 08:48
URI:	http://eprints.intimal.edu.my/id/eprint/2359

Actions (login required)

View Item