Metraipolis › archive › dc_f940252cbb15

presto: Efficient, Training-free, and Open-world Object Placement via Imaginary Search — arXiv:2608.21543

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.21543
License (as archived)
CC0-1.0
Archived
2026-10-18T07:42:18.180Z
SHA-256 of the archived text
92bafa6d1d8a27c7950613013bebe0c64d11f6136a795325ba7e07055279b53e
Stable URL of this page
https://metraipolis.com/documento/dc_f940252cbb15

How to cite

presto: Efficient, Training-free, and Open-world Object Placement via Imaginary Search — arXiv:2608.21543 — arXiv (en), https://arxiv.org/abs/2608.21543 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f940252cbb15 (Metraipolis archive dc_f940252cbb15, sha256 92bafa6d1d8a27c7950613013bebe0c64d11f6136a795325ba7e07055279b53e)
BibTeX
@misc{dcf940252cbb15,
  title = {presto: Efficient, Training-free, and Open-world Object Placement via Imaginary Search — arXiv:2608.21543},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.21543},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f940252cbb15; sha256 92bafa6d1d8a27c7950613013bebe0c64d11f6136a795325ba7e07055279b53e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.21543. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Object placement is critical in image composition, requiring spatially and semantically coherent positioning of objects within diverse scenes. Existing approaches typically rely on hand-crafted rules or supervised learning on limited datasets, which restricts their generalization and interpretability, especially in open-world scenarios involving novel objects and scenes. In this work, we reformulate open-world object placement as a heuristic search task guided by reasoning from a Multimodal Large Language Model (MLLM). We introduce \textsf{presto}, a zero-shot, training-free framework that operates within an imaginary action space to iteratively refine object position and scale. Our coarse-to-fine search strategy ensures fast convergence, and we evaluate two decision-making variants: Metric-guided Selection and MLLM-as-a-judge. Experiments across multiple benchmarks show that \textsf{presto}~achieves state-of-the-art performance, particularly in previously unseen, open-world settings. Human studies further reveal that the MLLM-as-a-judge variant produces more perceptually coherent placements than metric-driven approaches, highlighting a gap between standard evaluation metrics and human visual judgment.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-25 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.21543 — 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 2026-10-18T07:37:52.819Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).