Recent Advances in Scene Reconstruction, Simulation, and Generation

Dear Colleagues,

Scene reconstruction, simulation, and generation represent three deeply interconnected pillars of modern 3D computer vision. Together, they form the foundation for digitizing, understanding, and synthesizing the visual world. They are critical to applications ranging from autonomous driving and robotics to immersive media, urban planning, and scientific discovery.

In recent years, scene reconstruction has undergone a paradigm shift driven by neural scene representations. Methods such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting have demonstrated remarkable ability to recover high-fidelity geometry and appearance from sparse or unstructured image collections, enabling photorealistic novel view synthesis and dense 3D modeling at unprecedented quality and speed. These advances have redefined what is achievable in digital reconstruction and opened new possibilities for large-scale, real-time, and dynamic scene capture.

Simultaneously, the growing demand for scalable training environments and digital twins has propelled scene simulation to the forefront of research. Physics-aware and differentiable simulation frameworks now enable realistic modeling of lighting, material properties, object dynamics, and agent interactions within virtual environments. Such capabilities are essential for training and validating embodied AI systems, autonomous vehicles, and robotic manipulation policies in safe, controllable, and reproducible settings before real-world deployment.

Complementing reconstruction and simulation, scene generation has emerged as a transformative research direction fueled by advances in generative models, including diffusion models, large-scale vision-language models, and autoregressive architectures. These approaches have enabled the synthesis of diverse, geometrically consistent, and semantically meaningful 3D scenes from input, such as text prompts, images, or layout specifications, substantially reducing the cost of content creation and accelerating progress in AI-generated 3D content (AIGC), virtual world building, and creative design.

Despite significant progress, many open challenges remain. These include ensuring geometric and physical plausibility in generated scenes, achieving robust generalization across diverse real-world environments, scaling reconstruction and generation to complex and unbounded settings, and seamlessly integrating reconstruction, simulation, and generation into unified and interactive pipelines.

This Special Issue invites original research contributions that advance the state of the art in scene reconstruction, simulation, and generation. We welcome submissions that propose novel methods, address open challenges, or explore new applications across this rapidly evolving landscape. Topics of interest include, but are not limited to, the following:

• Neural scene representations for reconstruction and novel view synthesis;
• Real-time and large-scale 3D scene reconstruction;
• Dynamic scene modeling, reconstruction, and re-rendering;
• Differentiable and physics-based scene simulation;
• Simulation environments for embodied AI, robotics, and autonomous systems;
• Text-, image-, and layout-conditioned 3D scene generation;
• Generative models for 3D content creation and virtual world synthesis;
• Scene editing, composition, and manipulation;
• Unified frameworks bridging reconstruction, simulation, and generation;
• Benchmarks, datasets, and evaluation methodologies for 3D scenes;
• Applications in autonomous driving, AR/VR, digital twins, urban modeling, and creative industries.

Dr. Dan Wang
Dr. Xinrui Cui
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Journal of Imaging is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1800 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• 3D scene reconstruction
• neural scene representations
• 3D scene generation
• physics-based simulation
• novel view synthesis
• differentiable rendering
• embodied AI
• digital twins