Generative super-resolution of 3D Gaussian splat sets

Date

2026-9-03

Author

Bal, Batuhan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

145
views

135
downloads

Cite This

3D Gaussian Splatting (3DGS) enables real-time novel view synthesis, but Gaussian representations reconstructed from low-resolution observations are typically sparse and lack fine geometric and appearance details, limiting the quality of high resolution novel-view rendering. Existing 3DGS super-resolution methods either refine each scene through test-time optimization with external 2D priors, or learn feed-forward mappings from low-resolution images to high-resolution splats. Neither line of work learns a reusable cross-scene mapping that operates directly on pre-fit low-resolution Gaussian splat sets. This thesis proposes a generative two-stage framework for super-resolving pre-fit 3D Gaussian splat sets. In Stage I, a Point Transformer-based VQ-VAE learns a discrete latent codebook from high-quality Gaussian splats. In Stage II, a latent diffusion transformer denoises high-resolution latent tokens conditioned on encodings of degraded low-resolution splat sets, while a coordinate prediction head estimates proxy HR geometry for decoding. Instead of directly regressing Gaussian attributes, the proposed approach formulates restoration as generative latent refinement in the Stage~I codebook feature space, enabling feed-forward reconstruction without per-scene optimization or image-space priors. The proposed framework demonstrates the feasibility of generative super-resolution directly in the Gaussian parameter space, providing an alternative to optimization-based and image-driven approaches.

Subject Keywords

3D Gaussian splatting, Super-resolution, Vector quantization

URI

https://hdl.handle.net/11511/120672

Collections

Graduate School of Natural and Applied Sciences, Thesis