renderer integration

Manual 3D modeling remains labor-intensive and time-consuming, failing to meet the rapidly growing demands of XR and metaverse applications. Method: This paper presents a systematic survey of state-of-the-art methods for static 3D object and scene generation, introducing— for the first time—a multidimensional classification and cross-comparative framework that jointly considers representation evolution (e.g., point clouds, meshes, NeRFs) and generative paradigms (e.g., supervised learning, diffusion models, 2D foundation model priors, procedural modeling). Contribution/Results: We identify key challenges—including geometric-semantic consistency and scalability—and establish a reusable, multi-axis evaluation framework. Our analysis clarifies technological trajectories and performance boundaries, providing both theoretical foundations and practical guidelines for industrial-grade 3D content generation, thereby advancing the paradigm shift in 3D content creation.

This work addresses the opacity in understanding how rendering artifacts in 3D Gaussian Splatting (3DGS) reconstructions relate to the evolution of Gaussian parameters, compounded by the lack of effective diagnostic tools. To bridge this gap, we propose Vis4GS—the first multi-view visual analytics system that enables structured correlation among artifacts, Gaussian attributes, and optimization history. Vis4GS integrates interactive Gaussian selection, attribute timelines, densification lineage trees, and viewpoint coverage analysis, facilitating fine-grained attribution of blur or needle-like artifacts and multi-scale exploration of their generative lineage. User studies demonstrate that Vis4GS significantly enhances users’ ability to comprehend and diagnose artifacts compared to native viewers, transcending the limitations of final rendered images and global metrics.

Existing 3D portrait reconstruction methods rely on costly hardware or implicit representations (e.g., NeRF), limiting compatibility with industrial rendering pipelines such as those in real-time game engines. This work proposes an explicit, consumer-grade 3D virtual portrait generation framework that reconstructs high-fidelity mesh models and synthesizes dynamic, lighting-adaptive textures from only a few input images. Our core contribution is the novel “dynamic texture-driven” paradigm, which jointly optimizes differentiable rendering, neural texture synthesis, classical mesh refinement, and dynamic UV mapping—yielding explicit geometry and texture representations fully compliant with standard PBR shading and real-time rasterization. The method achieves 60 FPS rendering on a single GPU, attaining visual fidelity comparable to NeRF-based approaches. It has been successfully integrated into Unity and Unreal Engine, enabling real-time interactive control over facial expressions, illumination, and viewpoint.

Standard 3D Gaussian splatting lacks native uncertainty estimation and adaptive complexity control, making it difficult to identify weakly supported geometry or actively select informative novel viewpoints under sparse-view conditions. This work proposes the first Bayesian 3D Gaussian splatting framework, modeling Gaussian geometry via a Normal-Inverse-Wishart posterior and employing a Dirichlet process prior to enable adaptive component selection. The method incorporates rendering-aware surrogate summarization for efficient inference while explicitly delineating the boundaries between closed-form and approximate inference. It supports native uncertainty quantification and Bayesian active view selection, achieving a PSNR gain of 0.453 dB and an LPIPS reduction of 0.0146 in the 16→32 view task. The 95% coverage error is reduced by approximately 17× compared to surrogate models, with training costs only one-third of those required by deep ensembles.

This work addresses the geometric and photometric degradation in 3D Gaussian Splatting (3DGS) under sparse-view settings by proposing a unified inpainting framework leveraging video diffusion models. The authors construct a large-scale video dataset comprising 107.5K paired samples and introduce an isomorphic dual-model architecture featuring fine-grained 3DGS artifact classification and an Artifact-Aware Triplet Fusion mechanism guided by artifact heatmaps. For the first time, intensity-aware restoration is integrated into the self-attention structure, enabling precise spatiotemporal consistent inpainting. The proposed method significantly outperforms existing approaches in sparse novel-view synthesis and robust 3D reconstruction, effectively enhancing multi-view consistency and generalization capability.

Existing 3D Gaussian splatting methods rely on manual cloning/segmentation and sensitive initialization, resulting in unstable rendering quality and poor controllability over Gaussian count. This work reformulates the 3D Gaussian point set as a Monte Carlo Markov Chain (MCMC) sampling process of the scene’s physical representation. We replace heuristic Gaussian splitting/cloning with stochastic gradient Langevin dynamics (SGLD), and recast densification and pruning as deterministic MCMC state transitions. A probabilistic density relocation mechanism supplants cloning, while an unused-Gaussian regularizer enables flexible control over Gaussian count and robust initialization. Evaluated on standard benchmarks, our method achieves significantly improved rendering quality, reduces reliance on manual hyperparameter tuning, and unifies neural rendering with principled probabilistic modeling.

This work addresses the unclear trade-offs among reconstruction quality, model compactness, and rendering speed in dynamic 3D scene reconstruction. It presents the first systematic taxonomy and empirical evaluation of dynamic 3D Gaussian splatting methods, categorizing them into two paradigms: structure-guided approaches (e.g., deformation fields, canonical spaces, and meshes) and Gaussian-centric approaches (e.g., continuous functions and 4D representations). A comprehensive benchmarking study on the D-NeRF dataset reveals that structure-guided methods achieve superior reconstruction fidelity and model compactness, whereas Gaussian-centric methods enable faster rendering—often reaching real-time performance—at the cost of reduced quality stability and higher storage overhead. This study elucidates the fundamental trade-offs inherent in dynamic Gaussian splatting techniques and establishes a clear benchmark for future research.

Existing neural rendering methods are predominantly evaluated based on image visual quality, often overlooking geometric surface accuracy—a critical requirement for applications such as robotic grasping. To address this gap, this work proposes the first systematic evaluation framework specifically designed to assess the geometric fidelity of neural rendering techniques. Leveraging mainstream approaches including NeRF and Gaussian Splatting, the study evaluates surface and shape accuracy across 19 diverse 3D scenes using established geometric reconstruction metrics. By establishing a reproducible benchmark for geometric precision, this framework complements conventional visual-quality assessments and provides a reliable foundation for selecting and optimizing neural rendering methods in high-accuracy applications.

This work addresses floating artifacts, flickering, and blurriness in 3D Gaussian splatting reconstructions of wild scenes, which arise from camera pose errors, insufficient coverage, and noisy geometric initialization. To resolve these issues, the authors propose a geometry-guided video-to-video generation approach that refines rendered outputs with temporal consistency. Their method introduces, for the first time, a geometry-aware video generation framework that constructs a Gaussian primitive video buffer using depth, normals, opacity, and covariance. Combined with a synthetic data training strategy capable of simulating diverse degradation patterns, this approach significantly enhances generalization. The method achieves state-of-the-art performance on novel view synthesis benchmarks, with an efficient variant running at 21 FPS, enabling interactive applications.

Existing synthetic datasets suffer from insufficient photorealism and temporal coherence, limiting the applicability of generative inverse and forward rendering in real-world scenarios. This work proposes a large-scale dynamic video dataset derived from AAA-grade games, employing a dual-screen capture method to simultaneously acquire RGB frames and five-channel G-buffers, thereby disentangling geometry from material properties. Building upon this dataset, we introduce a ground-truth-free evaluation protocol leveraging vision-language models (VLMs) and develop a text-prompt-driven framework for G-buffer style editing. Experiments demonstrate that our approach significantly outperforms existing methods in cross-dataset generalization and controllable generation, with VLM-based evaluations showing strong alignment with human judgments. We also release an open-source toolkit enabling real-time, high-fidelity style transfer for game rendering.

This work addresses the challenge of high-fidelity, editable 3D reconstruction of large-scale indoor scenes from multi-view RGB images. The proposed method decomposes the scene into spatially overlapping local blocks and extends object-level generative priors—such as Trellis-2—to the full scene scale for the first time. It introduces a view-invariant, spatially anchored 3D conditional generation mechanism that fuses multi-view image features through projective conditioning to guide the generative model toward outputs consistent with observed data. This approach substantially improves reconstruction quality, outperforming state-of-the-art methods by 16% on indoor scenes, while producing multi-view consistent, physically based rendering (PBR)-ready, and editable meshes.