Bracket Diffusion: HDR Image Generation by Consistent LDR Denoising

Computer Graphics Forum, Journal Year: 2025, Volume and Issue: unknown

Published: April 16, 2025

Abstract We demonstrate generating HDR images using the concerted action of multiple black‐box, pre‐trained LDR image diffusion models. Common models are not as, first, there is no sufficiently large dataset available to re‐train them, and, second, even if it was, re‐training such impossible for most compute budgets. Instead, we seek inspiration from capture literature that traditionally fuses sets images, called “exposure brackets”, produce a single image. operate denoising processes generate brackets together form valid result. To this end, introduce consistency term into process couple they agree across exposure range share. versions state‐of‐the‐art unconditional and conditional as well restoration‐type (LDR2HDR) generative modeling.

Language: Английский

Alfie: Democratising RGBA Image Generation with No $$$

Lecture notes in computer science, Journal Year: 2025, Volume and Issue: unknown, P. 38 - 55

Published: Jan. 1, 2025

Language: Английский