Get the latest tech news
New AI diffusion model approach solves the aspect ratio problem
Rice computer scientists have developed a way to improve the performance of artificial intelligence diffusion models.
“One reason diffusion models need help with non-square aspect ratios is that they usually package local and global information together,” said Haji Ali, who worked on synthesizing motion in AI-generated videos before joining Ordóñez-Román’s research group at Rice for his Ph.D. studies. His attire is simple yet dignified” (Image courtesy of Moayed Haji Ali/Rice University)“This approach is a successful attempt to leverage the intermediate representations of the model to scale them up so that you get global consistency,” Ordóñez-Román said. Peer-reviewed paper and project information: ElasticDiffusion: Training-free Arbitrary Size Image Generation through Global-Local Content Separation | The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024
Or read this on Hacker News