We introduce Edify Image, a family of diffusion models capable of generating photorealistic image content with pixel-perfect accuracy. Edify Image utilizes cascaded pixel-space diffusion models trained using a novel Laplacian diffusion process, in which image signals at different frequency bands are attenuated at varying rates. Edify Image supports a wide range of applications, including text-to-image synthesis, 4K upsampling, ControlNets, 360 HDR panorama generation, and finetuning for image customization.
@article{arxiv.2411.07126,
title = {Edify Image: High-Quality Image Generation with Pixel Space Laplacian Diffusion Models},
author = {NVIDIA and : and Yuval Atzmon and Maciej Bala and Yogesh Balaji and Tiffany Cai and Yin Cui and Jiaojiao Fan and Yunhao Ge and Siddharth Gururani and Jacob Huffman and Ronald Isaac and Pooya Jannaty and Tero Karras and Grace Lam and J. P. Lewis and Aaron Licata and Yen-Chen Lin and Ming-Yu Liu and Qianli Ma and Arun Mallya and Ashlee Martino-Tarr and Doug Mendez and Seungjun Nah and Chris Pruett and Fitsum Reda and Jiaming Song and Ting-Chun Wang and Fangyin Wei and Xiaohui Zeng and Yu Zeng and Qinsheng Zhang},
journal= {arXiv preprint arXiv:2411.07126},
year = {2024}
}