Haitian Zheng

Haitian Zheng is a Research Scientist 2 at Adobe Research, which he joined full-time in 2023 after three research internships at Adobe (2020–2022). He is the first author and original developer of the CM-GAN image inpainting model, which Adobe shipped in Content-Aware Fill in Photoshop (2022), Content-Aware Remove in Lightroom (2022) and the Remove Tool in Photoshop (2023), and which also powered the 2023 launch of Generative Fill. He is also the first author and original developer of PixPerfect, the pixel-space refinement model used in the updated Generative Fill and Generative Expand models in Photoshop and Adobe Firefly (2024–2026). Most recently, he has led the research on high-resolution image restoration and upscaling (ZipIR) behind Generative Upscale (Firefly Upscaler 2) in Photoshop (currently in Photoshop Beta). A list of his shipped features is available on his personal website.

He received his Ph.D. degree in computer science from the University of Rochester in 2023, advised by Prof. Jiebo Luo. Prior to that, he obtained his M.S. degree in electronics and communication engineering from the University of Science and Technology of China in 2016, advised by Prof. Lu Fang. His research interests include computer vision, computer graphics, and generative models, with a focus on image generation, editing, enhancement, and restoration for computer-aided digital art creation. He has published at CVPR, ICCV, ECCV, NeurIPS, SIGGRAPH, AAAI, and IEEE TPAMI, and is a named inventor on 12 issued U.S. patents.

More information about his publications and profile can be found at: Google Scholar. For more information, please visit his personal website.