Accelerating Redshift-Conditioned Galaxy Image Synthesis with One-step Generative Modeling
Tianyue Yang, Sandro Tacchella, Xiao Xue · 2026
Read Accelerating Redshift-Conditioned Galaxy Image Synthesis with One-step Generative Modeling by Tianyue Yang, Sandro Tacchella, Xiao Xue (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-17546.txt.
Description
Yang, Tacchella and Xue study redshift-conditioned galaxy image synthesis, comparing diffusion samplers with one-step pixel-MeanFlow on GalaxiesML-64 data. Standard DDPM sampling gives the best distributional fidelity at high cost, while second-order samplers improve efficiency over DDIM. One-step pixel-MeanFlow matches many morphology statistics at far lower cost but trails many-step DDPM on fine structure. The work opens a path to efficient conditional simulators for large cosmological surveys.
Catalog record
| CREATOR | Tianyue Yang, Sandro Tacchella, Xiao Xue |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39840 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.17546