Text-to-image editing using generative AI with cross attention control

Wong, Kai Jun (2023) Text-to-image editing using generative AI with cross attention control. Project Report. Melaka, Malaysia, Universiti Teknikal Malaysia Melaka. (Submitted)

	Text (24 Pages) Text-to-image editing using generative AI with cross attention control.pdf - Submitted Version Download (552kB)
	Text (Full Text) Text-to-image editing using generative AI with cross attention control.pdf - Submitted Version Restricted to Registered users only Download (1MB)

Abstract

Recently, text-to-image models, quickly garnered attention for their incredible generating potential in both semantics and composition. It can save a significant amount of time and resources in generating realistic and detailed images using the text-to-image model instead of manual artwork creation. However, editing is challenging for these generative models, in the text-based models, even a small modification of the text prompt often leads to a completely different outcome. Hence, this project proposes a prompt editing insights solution for text-to-image editing using cross attention control. The cross-attention maps associated in this solution empower users to grasp the connection between the text prompt and the generated image. This helps users pick an accurate word in image generation and editing. At last, the developed tool is able to provide meaningful editing insight and edit the image accordingly for word within noun word class.

Actions (login required)

View Item