Metraipolis › archive › dc_ee6c8a8781fa

When Text-to-Image Helps Editing: The Effects of Conditioning During Denoising — arXiv:2610.01681

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.01681
License (as archived): CC0-1.0
Archived: 2026-10-06T19:14:10.479Z
SHA-256 of the archived text: d77fe32c96dedacd89e95fd3882f96794e8959760ec6f5f9c51e5b33e34647bc
Stable URL of this page: https://metraipolis.com/documento/dc_ee6c8a8781fa

How to cite

When Text-to-Image Helps Editing: The Effects of Conditioning During Denoising — arXiv:2610.01681 — arXiv (en), https://arxiv.org/abs/2610.01681 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ee6c8a8781fa (Metraipolis archive dc_ee6c8a8781fa, sha256 d77fe32c96dedacd89e95fd3882f96794e8959760ec6f5f9c51e5b33e34647bc)

BibTeX

@misc{dcee6c8a8781fa,
  title = {When Text-to-Image Helps Editing: The Effects of Conditioning During Denoising — arXiv:2610.01681},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.01681},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ee6c8a8781fa; sha256 d77fe32c96dedacd89e95fd3882f96794e8959760ec6f5f9c51e5b33e34647bc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.01681. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Unified models are trained for both instruction-based image editing and text-to-image (T2I) generation, but standard editing pipelines keep source-image conditioning throughout denoising. We ask whether editing can benefit from T2I, and study how the effects of conditioning vary across edits and denoising stages. In pure editing, source attention declines for some edits over the sampling trajectory. This observation led us to task switching, which lets the model draw on its T2I capabilities. Across three unified editors and four benchmarks, switching to the T2I task for bounded intervals improves edit quality, while mean perceptual preservation remains close to pure editing across all three models. Unified editors therefore benefit from using both conditioning modes they are trained for, and the timing of the switch sets the balance between quality and preservation.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-10-02 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.01681 — 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 2026-10-06T17:37:36.821Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).