PhotoAgent: A Robotic Photographer with Spatial and Aesthetic Understanding
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908909051052032 |
|---|---|
| author | Che, Lirong Gan, Zhenfeng Chen, Yanbo Tan, Junbo Wang, Xueqian |
| author_facet | Che, Lirong Gan, Zhenfeng Chen, Yanbo Tan, Junbo Wang, Xueqian |
| contents | Embodied agents for creative tasks like photography must bridge the semantic gap between high-level language commands and geometric control. We introduce PhotoAgent, an agent that achieves this by integrating Large Multimodal Models (LMMs) reasoning with a novel control paradigm. PhotoAgent first translates subjective aesthetic goals into solvable geometric constraints via LMM-driven, chain-of-thought (CoT) reasoning, allowing an analytical solver to compute a high-quality initial viewpoint. This initial pose is then iteratively refined through visual reflection within a photorealistic internal world model built with 3D Gaussian Splatting (3DGS). This ``mental simulation'' replaces costly and slow physical trial-and-error, enabling rapid convergence to aesthetically superior results. Evaluations confirm that PhotoAgent excels in spatial reasoning and achieves superior final image quality. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_22796 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PhotoAgent: A Robotic Photographer with Spatial and Aesthetic Understanding Che, Lirong Gan, Zhenfeng Chen, Yanbo Tan, Junbo Wang, Xueqian Computer Vision and Pattern Recognition Artificial Intelligence Robotics Embodied agents for creative tasks like photography must bridge the semantic gap between high-level language commands and geometric control. We introduce PhotoAgent, an agent that achieves this by integrating Large Multimodal Models (LMMs) reasoning with a novel control paradigm. PhotoAgent first translates subjective aesthetic goals into solvable geometric constraints via LMM-driven, chain-of-thought (CoT) reasoning, allowing an analytical solver to compute a high-quality initial viewpoint. This initial pose is then iteratively refined through visual reflection within a photorealistic internal world model built with 3D Gaussian Splatting (3DGS). This ``mental simulation'' replaces costly and slow physical trial-and-error, enabling rapid convergence to aesthetically superior results. Evaluations confirm that PhotoAgent excels in spatial reasoning and achieves superior final image quality. |
| title | PhotoAgent: A Robotic Photographer with Spatial and Aesthetic Understanding |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2603.22796 |