Estimation of Optimal Prompts for Generating Logos using Image Generation AI

Abstract

This paper estimates the optimal prompts for generating target logos using image generation AI. As AI technology advances, the quality of generated images heavily depends on input prompts. This research utilizes Stable Diffusion and DALL-E 3 to identify key factors for improving logo generation accuracy. It examines how specific prompts generate images similar to existing corporate logos and analyzes their impact on the color, symbol, and style of the output using similarity measurement techniques.