Improving quality checks for 3D scene images without exact comparisons
From Explicit References to Scene Manifolds: Distributional Fidelity and Realism for Radiance Field Quality Assessment
Computer Vision and Pattern Recognition
Summary
Checking the quality of 3D images made from scene data can be tricky, especially when pictures aren't perfectly aligned. The authors created a new method called SCODA that doesn't need a direct side-by-side comparison of images. Instead, it learns what a good scene usually looks like by studying many high-quality views, then judges new images based on how well they fit that pattern. This method uses smart tools to spot both realistic details and overall accuracy, working well even when images come from different techniques or viewpoints.
Radiance field3D Gaussian SplattingNovel view synthesisQuality assessmentFull-reference metricsScene manifoldMultivariate Gaussian distributionDeep feature spacePatch discriminatorNeural Radiance Fields (NeRF)
Authors
Saeed Mahmoudpour, Gi-Mun Um, Hyon-Gon Choo, Peter Schelkens
Abstract
Radiance field representations such as 3D Gaussian Splatting (3DGS) enable high-quality novel view synthesis but can introduce complex, view-dependent artifacts from reconstruction, rendering, and compression. Reliable perceptual quality assessment (QA) is thus essential for evaluating rendered views and guiding the design of perceptually faithful scene representations. Existing full-reference QA metrics require an aligned reference image, while recent cross-reference metrics relax this requirement by comparing a test view with non-aligned references. However, under wide-baseline radiance field settings, selecting a reliable nearby reference can be difficult, particularly when evaluating views along arbitrary trajectories and poses. We propose SCODA, a lightweight scene-conditioned objective QA method that shifts QA from explicit image-to-image comparison to scene-manifold modeling. High-quality observations of each scene are represented as a multivariate Gaussian distribution in deep feature space, producing a semantic fidelity score that measures deviation from the scene distribution. A weakly-supervised distortion-aware patch discriminator provides a complementary realism signal, and both cues are combined through an unsupervised bounded fusion strategy. Experiments on multiple benchmarks show strong agreement with human judgments and robust generalization across GS- and NeRF-generated views and trajectories. Code is publicly available at https://gitlab.com/saeedmp/scoda.