Metraipolis › archive › dc_2bf769f327e9
UltraDiff: Transferring High-Fidelity Priors to Compressed Latent Spaces for High-Resolution Image Generation — arXiv:2610.21202
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.21202
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T10:35:11.529Z
- SHA-256 of the archived text
49a2e1de0f7b64a4edfb54d2c95af93f35410fc82d2c9dfbbad28343a07eff46- Stable URL of this page
https://metraipolis.com/documento/dc_2bf769f327e9
How to cite
UltraDiff: Transferring High-Fidelity Priors to Compressed Latent Spaces for High-Resolution Image Generation — arXiv:2610.21202 — arXiv (en), https://arxiv.org/abs/2610.21202 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2bf769f327e9 (Metraipolis archive dc_2bf769f327e9, sha256 49a2e1de0f7b64a4edfb54d2c95af93f35410fc82d2c9dfbbad28343a07eff46)
BibTeX
@misc{dc2bf769f327e9,
title = {UltraDiff: Transferring High-Fidelity Priors to Compressed Latent Spaces for High-Resolution Image Generation — arXiv:2610.21202},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.21202},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2bf769f327e9; sha256 49a2e1de0f7b64a4edfb54d2c95af93f35410fc82d2c9dfbbad28343a07eff46},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21202. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
High-resolution synthesis with Latent Diffusion Models (LDMs) faces a fundamental trade-off between computational cost and generative fidelity. This trade off is largely governed by the latent compression factor f (e.g. f8 denotes f=8 and f32 f=32). While aggressive compression (f32) significantly improves efficiency, it suffers from severe information loss and ambiguity compared to standard f8 models. We propose UltraDiff, a framework that reformulates this challenge as joint latent diffusion learning followed by marginal score distillation. We introduce MixTeacher, a dual branch DiT that models the joint f8-f32 distribution, utilizing shared attention to resolve the ambiguity of the compressed space with high fidelity f8 priors. Crucially, we propose vector field distillation in the native f32 space. Theoretically, this compels the student to learn the conditional expectation of the teacher trajectory, effectively filtering out compression artifacts. Across the evaluated 1K to 4K settings, UltraDiff approaches f8 level quality while preserving f32 efficiency. With 20 sampling steps on a single A100 GPU, it generates a 2K image in 6.2 seconds over 40 times faster than our f8 baseline. Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2610.21202 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T08:52:02.638Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).