Metraipolis › archive › dc_a64014c8ed9e

Where to Look Is Not How to Fix: Pre-Denoising Diagnostics and Modality-Dependent Control in Diffusion Composition — arXiv:2610.03068

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.03068
License (as archived)
CC0-1.0
Archived
2026-10-06T09:07:44.258Z
SHA-256 of the archived text
bc6958f4cdcc3d652f33c21b223597cfe0b26c0ed7b36397b6bfe72e56632bcb
Stable URL of this page
https://metraipolis.com/documento/dc_a64014c8ed9e

How to cite

Where to Look Is Not How to Fix: Pre-Denoising Diagnostics and Modality-Dependent Control in Diffusion Composition — arXiv:2610.03068 — arXiv (en), https://arxiv.org/abs/2610.03068 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a64014c8ed9e (Metraipolis archive dc_a64014c8ed9e, sha256 bc6958f4cdcc3d652f33c21b223597cfe0b26c0ed7b36397b6bfe72e56632bcb)
BibTeX
@misc{dca64014c8ed9e,
  title = {Where to Look Is Not How to Fix: Pre-Denoising Diagnostics and Modality-Dependent Control in Diffusion Composition — arXiv:2610.03068},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.03068},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a64014c8ed9e; sha256 bc6958f4cdcc3d652f33c21b223597cfe0b26c0ed7b36397b6bfe72e56632bcb},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.03068. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Understanding compositional failures in text-to-image diffusion requires identifying both where stress is detectable and how intervention changes the output. We study these questions through a controlled anchor--stress protocol that jointly evaluates text-encoder diagnostics and denoiser interventions. We introduce a text-only Compositional Stress Index (CSI), which separates common from rare compositions across SD1.5, SDXL, and the SD3 text path and provides an upstream diagnostic coordinate. A matched six-prompt localization study links intervention location to distinct outcomes: residual-minimizing embedding adapters improve representation fit, while downstream cross-attention intervention increases color hit rate (CHR) by 0.0272. Across SD1.5 and SDXL denoiser blocks, the largest positive signed diagnostic-accessibility mean occurs at the deep encoder, whereas selective boost has its largest positive mean CHR response at decoder blocks. Selective subtraction and broad ablation reveal further modality- and architecture-dependent responses, including a substantial CHR decrease when SDXL decoder cross-attention is broadly ablated. We find a diagnosis-control dissociation under our controlled attribute-object composition setting: compositional defects are diagnosable before denoising, but the representation coordinate that exposes risk is not necessarily the coordinate or modality that improves generation.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-10-05 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.03068 — 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 2026-10-06T08:21:44.204Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-10-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).