# GaussianVAE: Adaptive Learning Dynamics of 3D Gaussians for High-Fidelity Super-Resolution

arXiv: https://arxiv.org/abs/2506.07897 · HTML: https://mlsanity.com/paper/2506.07897/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 10 of its 12 references that are in the map.
- First posted on arXiv 9 Jun 2025.

## Builds on (most cited first, 10 of 10)

- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Instant Neural Graphics Primitives with a Multiresolution Hash Encoding](https://mlsanity.com/paper/2201.05989/) (2022)
- [A Survey on Evaluation of Large Language Models](https://mlsanity.com/paper/2307.03109/) (2023)
- [PU-GCN: Point Cloud Upsampling using Graph Convolutional Networks](https://mlsanity.com/paper/1912.03264/) (2019)
- [NeRF-SR: High-Quality Neural Radiance Fields using Supersampling](https://mlsanity.com/paper/2112.01759/) (2021)
- [GaussianSR: High Fidelity 2D Gaussian Splatting for Arbitrary-Scale Image Super-Resolution](https://mlsanity.com/paper/2407.18046/) (2024)
- [SRGS: Super-Resolution 3D Gaussian Splatting](https://mlsanity.com/paper/2404.10318/) (2024)
- [SuperGaussian: Repurposing Video Models for 3D Super Resolution](https://mlsanity.com/paper/2406.00609/) (2024)
- [Diffusion-Based Point Cloud Super-Resolution for mmWave Radar Data](https://mlsanity.com/paper/2404.06012/) (2024)
- [A strong baseline for image and video quality assessment](https://mlsanity.com/paper/2111.07104/) (2021)

## Built on by (most cited first, 1 of 1)

- [RobustGS: Unified Boosting of Feedforward 3D Gaussian Splatting under Low-Quality Conditions](https://mlsanity.com/paper/2508.03077/) (2025)

## Abstract (arXiv)

We present a novel approach for enhancing the resolution and geometric fidelity of 3D Gaussian Splatting (3DGS) beyond native training resolution. Current 3DGS methods are fundamentally limited by their input resolution, producing reconstructions that cannot extrapolate finer details than are present in the training views. Our work breaks this limitation through a lightweight generative model that predicts and refines additional 3D Gaussians where needed most. The key innovation is our Hessian-assisted sampling strategy, which intelligently identifies regions that are likely to benefit from densification, ensuring computational efficiency. Unlike computationally intensive GANs or diffusion approaches, our method operates in real-time (0.015s per inference on a single consumer-grade GPU), making it practical for interactive applications. Comprehensive experiments demonstrate significant improvements in both geometric accuracy and rendering quality compared to state-of-the-art methods, establishing a new paradigm for resolution-free 3D scene enhancement.