Metraipolis › archive › dc_db98461e05bf

Adversarial Training for Pixel Diffusion — arXiv:2609.38170

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.38170
License (as archived): CC0-1.0
Archived: 2026-10-07T18:42:02.113Z
SHA-256 of the archived text: 18a627970eef9ddfe8c8d7772f97026595af5150a8932f4316c0d924766e3680
Stable URL of this page: https://metraipolis.com/documento/dc_db98461e05bf

How to cite

Adversarial Training for Pixel Diffusion — arXiv:2609.38170 — arXiv (en), https://arxiv.org/abs/2609.38170 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_db98461e05bf (Metraipolis archive dc_db98461e05bf, sha256 18a627970eef9ddfe8c8d7772f97026595af5150a8932f4316c0d924766e3680)

BibTeX

@misc{dcdb98461e05bf,
  title = {Adversarial Training for Pixel Diffusion — arXiv:2609.38170},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38170},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_db98461e05bf; sha256 18a627970eef9ddfe8c8d7772f97026595af5150a8932f4316c0d924766e3680},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38170. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Pixel diffusion models generate RGB images directly, avoiding the bottleneck of an autoencoder, yet their outputs still systematically underrepresent fine-scale natural-image statistics. We show that adversarial learning provides an effective post-training correction for this deficiency. Starting from a pretrained model, we retain its original diffusion or flow-matching objective and add an adversarial loss to the predicted output at non-high-noise timesteps, leaving the model architecture and sampling procedure unchanged. To our knowledge, this is the first systematic study of adversarial post-training for pixel diffusion. Across two pixel backbones, the method jointly improves distribution fidelity, coverage, prompt alignment, and perceptual quality. We further investigate why it works. Frequency-band and power-law analyses show that the original models systematically underproduce natural-image high-frequency content, while adversarial post-training restores this missing spectral power. In contrast, perceptual loss also increases high-frequency content but sacrifices distribution fidelity and prompt alignment. Nearest-neighbor, recall, and matched no-GAN SFT controls further rule out memorization, mode dropping, and additional optimization as simple explanations. Finally, we examine the boundary of this effect. Under the tested latent diffusion configurations, the same procedure does not produce comparable joint gains and adds almost no decoded high-frequency power. These results identify direct output access to the image statistics being corrected as a key factor governing when adversarial post-training succeeds.

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/2609.38170 — 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 2026-10-07T18:04:24.958Z 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).