Metraipolis › archive › dc_3aeb967ea601

From Feed-Forward to Flow: Unifying Reconstruction and Generation Is Easier Than You Think — arXiv:2609.32761

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.32761
License (as archived)
CC0-1.0
Archived
2026-10-07T23:51:04.398Z
SHA-256 of the archived text
8b150a011d5f503f3762d54bc0ccc1eb3de21e1934d507bba3e1740114f43096
Stable URL of this page
https://metraipolis.com/documento/dc_3aeb967ea601

How to cite

From Feed-Forward to Flow: Unifying Reconstruction and Generation Is Easier Than You Think — arXiv:2609.32761 — arXiv (en), https://arxiv.org/abs/2609.32761 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3aeb967ea601 (Metraipolis archive dc_3aeb967ea601, sha256 8b150a011d5f503f3762d54bc0ccc1eb3de21e1934d507bba3e1740114f43096)
BibTeX
@misc{dc3aeb967ea601,
  title = {From Feed-Forward to Flow: Unifying Reconstruction and Generation Is Easier Than You Think — arXiv:2609.32761},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32761},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3aeb967ea601; sha256 8b150a011d5f503f3762d54bc0ccc1eb3de21e1934d507bba3e1740114f43096},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32761. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reconstruct where the images provide evidence, and generate where they do not: recent success of spatial world models such as Atlas (World Labs Team, 2026) highlights the value of unifying reconstruction and generation in one model. Yet the two have long lived in separate paradigms with distinctive failure modes: feed-forward reconstruction averages ambiguity into blur, while conditional generation invents plausible but scene-inconsistent detail. In this work, we present a unified flow-based formulation for reconstruction and generation, where a shared clean-target predictor performs direct reconstruction at its single-step endpoint and unfolds conditional generation through multi-step flow. A controlled toy study reveals the mechanism: with a single step, the predictor collapses to the conditional mean just like feed-forward methods, favoring consistency over diversity. With multi-step inference, the fidelity of generated details grows with context richness: closer observations reduce ambiguity and yield better-matched details. We further instantiate the formulation in appearance and geometry 3D tasks. JiT-LVSM improves perceptual and distributional quality in novel view synthesis, while JUSt3R retains competitive single-step geometry prediction with additional multi-step inference capabilities that reduces veil and flying-pixel artifacts, producing cleaner surface structure with greater test-time compute. Together, they show that reconstruction and generation can share both a formulation and a backbone, with their behavior governed by denoising configuration---making unification surprisingly simple.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.32761 — 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 2026-10-07T23:29:27.095Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).