EDoF-NeRF: extended depth-of-field neural radiance fields using a coded aperture camera

A new imaging technique called EDoF-NeRF extends the depth-of-field in neural radiance fields by using a coded aperture camera, according to a preprint posted to the arXiv repository on June 17, 2026 [1][2]. The method, formally named extended depth-of-field neural radiance fields, addresses a fundamental trade-off between depth-of-field and light quantity that affects both conventional cameras and the NeRF models that rely on their images [1][2]. NeRF, an emerging technique for rendering photorealistic novel views from multi-view image datasets, is built on implicit neural representations [2]. The research was submitted by Yoshiyuki Shirasaki [1]. To overcome the limitation, the approach places a coded aperture at the camera pupil. This preserves spatial frequency components even when the scene is defocused, information that is typically lost with a standard circular aperture [2]. The team developed a camera model that incorporates the coded aperture directly into the NeRF pipeline, allowing the system to accept coded images as input and generate novel views with an extended depth-of-field [2]. The authors validated EDoF-NeRF through both simulations and physical experiments, reporting that it demonstrated superior performance compared to conventional aperture cameras [1][2]. The submission, a 5,418 KB preprint, was posted under the physics.optics category on arXiv, an open-access repository for electronic preprints that has been operating since August 1991 and now receives about 24,000 articles per month [1][6]. arXiv itself is not a peer-reviewed journal; papers are moderated before posting but do not undergo formal peer review [6]. The platform hosts work across mathematics, physics, computer science, and related fields, and by the end of 2021 had surpassed two million articles [6]. The EDoF-NeRF paper appears alongside experimental community tools developed under the arXivLabs framework, a program launched in 2020 that allows third-party collaborators to build features such as citation explorers and code linkers on top of the repository's article pages [4][5].