Sphinx is a training‑free hybrid inference system for novel view synthesis that combines fast regression models with selective diffusion refinement. It first generates coarse views via regression, then applies diffusion only where needed to improve quality while reducing computation. An SLO‑aware scheduler dynamically allocates GPU resources, achieving up to 2.7× faster inference with less than 10% perceptual loss and up to 10× higher SLO attainment compared to FCFS scheduling.
Feed-forward 3D Gaussian Splatting enables efficient novel-view synthesis without per-scene optimization, but most existing methods assume a fixed set of context views and process them jointly. This limits their applicability to online scenarios where calibrated views arrive sequentially and the scene must be updated causally.
Spackle is a lightweight residual learning framework designed to improve large-view single-image novel view synthesis (NVS) by mitigating capacity competition in hybrid decoupled systems that combine 3D Gaussian Splatting (3DGS) and diffusion models. It operates in three stages: predicting base 3DGS attributes, automatically identifying poorly reconstructed regions, and learning a residual 3DGS focused on those areas. During inference, Spackle merges the baseline and augmented Gaussians to produce high-fidelity novel views, achieving state‑of‑the‑art performance on large-view-deviation cases.
arXiv:2606. 02068v1 Announce Type: cross Abstract: Recently, novel view synthesis has witnessed remarkable progress, with mainstream methods such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) delivering impressive results.
arXiv:2610.07958v1 Announce Type: new Abstract: Feed-forward 3D Gaussian Splatting (3DGS) reconstructs a scene in a single forward pass, replacing per-scene optimization with a network trained across...
AnyView is a diffusion-based video generation framework designed for dynamic view synthesis, requiring minimal inductive biases or geometric assumptions. It trains a generalist spatiotemporal implicit representation using diverse data sources—monocular, multi-view static, and multi-view dynamic—to produce zero-shot novel videos from arbitrary camera locations and trajectories. The authors evaluate AnyView on standard benchmarks, introduce a new challenging benchmark called AnyViewBench for extreme dynamic view synthesis, and demonstrate that AnyView outperforms baselines in maintaining realistic, plausible, and spatiotemporally consistent videos across diverse real-world scenarios.
arXiv:2610.20254v1 Announce Type: new Abstract: Large view synthesis models (LVSMs) reconstruct scenes from sparse views and render novel views quickly, but they incur unnecessary storage and renderi...
arXiv:2609.23169v1 Announce Type: new Abstract: High-quality texture generation is essential for creating realistic and production-ready 3D assets. Recent multi-view diffusion methods have shown prom...
Turbo4DGen is an ultra‑fast acceleration framework designed for diffusion‑based multi‑view 4D content generation. It introduces a spatiotemporal cache that reuses intermediate attention across denoising steps, along with semantic‑aware attention pruning and an adaptive SCM chain bypass scheduler to cut redundant computations. Experiments on ObjaverseDy and Consistent4D show an average 9.7× speedup without quality loss, making it the first dedicated acceleration solution for 4D generation.
arXiv:2609.06157v1 Announce Type: cross Abstract: 3D Gaussian splatting (3DGS) represents scenes with explicit primitives and supports real-time novel-view synthesis, yet its system efficiency varies...
Large view synthesis models (LVSMs) reconstruct scenes from sparse views and render novel views quickly, but they incur unnecessary storage and rendering costs due to token redundancy. We present GLoD...
arXiv:2607. 17842v1 Announce Type: cross Abstract: Recent breakthroughs in 3D Gaussian Splatting (3DGS) have advanced neural rendering with high fidelity and speed.
NeuralSRender is a sim‑to‑real video translation framework that improves the photorealism of synthetic driving videos while preserving geometric consistency. It builds on a pretrained video diffusion transformer and uses a lightweight projector to inject G‑buffer geometric and material priors, with LoRA for efficient adaptation. The two‑stage training first learns G‑buffer conditioning from synthetic data, then adapts to real‑world appearance, adding only 1.45 % parameter overhead and requiring 24 GB GPU memory. The authors also release GenReal, a large‑scale dataset of synthetic and real driving videos with G‑buffers, and show that training perception models on NeuralSRender‑translated videos boosts detection AP₅₀ on nuImages from 35.1 to 40.6 and segmentation mIoU on Cityscapes from 32.4 to 40.8.