Metraipolis › archive › dc_59ec45c8173b

DMA$^2$: Pixel-space Distribution Matching with Adversarial and Anchor Losses — arXiv:2609.38156

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38156
License (as archived)
CC0-1.0
Archived
2026-10-07T18:42:00.622Z
SHA-256 of the archived text
6ad2a4bfdad003fa06f71e667cb3ef6c094485b2c20c26520dfe8e828af19336
Stable URL of this page
https://metraipolis.com/documento/dc_59ec45c8173b

How to cite

DMA$^2$: Pixel-space Distribution Matching with Adversarial and Anchor Losses — arXiv:2609.38156 — arXiv (en), https://arxiv.org/abs/2609.38156 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_59ec45c8173b (Metraipolis archive dc_59ec45c8173b, sha256 6ad2a4bfdad003fa06f71e667cb3ef6c094485b2c20c26520dfe8e828af19336)
BibTeX
@misc{dc59ec45c8173b,
  title = {DMA$^2$: Pixel-space Distribution Matching with Adversarial and Anchor Losses — arXiv:2609.38156},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38156},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_59ec45c8173b; sha256 6ad2a4bfdad003fa06f71e667cb3ef6c094485b2c20c26520dfe8e828af19336},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38156. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Distribution matching distillation (DMD) provides a general framework for few-step diffusion generation, but its modern text-to-image instantiations have been developed primarily around latent diffusion. It therefore overlooks key properties and design opportunities of native RGB. We revisit two DMD interfaces for pixel-space teachers. On the teacher-matching side, diagnostics show low-noise RGB matching is dominated by a local-texture cue, motivating a fixed high-noise matching band. On the real-data side, native clean-RGB outputs allow guidance from an external visual representation without traversing a decoder or sharing the heavy fake-score critic. DINO-Adv removes this critic from the adversarial gradient path and supplies local parametric patch guidance. For distribution-level guidance, we introduce AF-Loss, a parameter-free auxiliary semantic distribution-field objective designed for text-to-image DMD. It operates on detached rolling real and generated supports in the shared DINOv2 space while preserving prompt-conditioned teacher supervision. AF-Loss adds no learnable parameters or inference-time computation. Together these designs form DMA$^2$. Across DPG-Bench, GenEval, VQAScore, and COCO30K, the four-step DMA$^2$ student performs better than the 25-step teacher and evaluated few-step distillers.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.38156 — 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 2026-10-07T18:04:24.955Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).