With the widespread application of digital intelligent technologies, rebranding has become both a challenge and an opportunity for businesses. This paper first conducts a questionnaire survey to understand the relationship between housing brand image and home-buying decisions. Based on housing brand image design in the context of digital transformation, it innovatively proposes a text-to-image generation method that utilizes a single-stage generative adversarial network (GAN) structure combined with a deep attention mechanism, starting from the application of AIGC technology in brand image design. Analysis reveals that housing brand image significantly influences consumers’ home-buying choices, accounting for over 70% of decisions, while over 80% of consumers perceive a connection between housing brand image and construction quality, underscoring the importance of housing brand image design. The experimental results of the designed text-to-image model, including IS, Accuracy, and FID, outperform the comparison methods. Specifically, the IS value and Accuracy value improved by 13.79% to 28.12% and 3.44% to 62.66%, respectively, while the FID value decreased by 2.63% to 36.52%, demonstrating its excellent performance in generating housing brand image design logo images. Therefore, this method can be applied to housing brand image design to promote the intelligence of brand image design and its display and dissemination on virtual platforms.