arXiv:2511.15102v1 Announce Type: new 
Abstract: The recent introduction of 3D Gaussian Splatting (3DGS) has significantly advanced novel view synthesis. Several studies have further improved the rendering quality of 3DGS, yet they still exhibit noticeable visual discrepancies when synthesizing views at sampling rates unseen during training. Specifically, they suffer from (i) erosion-induced blurring artifacts when zooming in and (ii) dilation-induced staircase artifacts when zooming out. We speculate that these artifacts arise from the fundamental limitation of the alpha blending adopted in 3DGS methods. Instead of the conventional alpha blending that computes alpha and transmittance as scalar quantities over a pixel, we propose to replace it with our novel Gaussian Blending that treats alpha and transmittance as spatially varying distributions. Thus, transmittances can be updated considering the spatial distribution of alpha values across the pixel area, allowing nearby background splats to contribute to the final rendering. Our Gaussian Blending maintains real-time rendering speed and requires no additional memory cost, while being easily integrated as a drop-in replacement into existing 3DGS-based or other NVS frameworks. Extensive experiments demonstrate that Gaussian Blending effectively captures fine details at various sampling rates unseen during training, consistently outperforming existing novel view synthesis models across both unseen and seen sampling rates.

أدى إدخال تقنية 3D Gaussian Splatting (3DGS) إلى تقدم كبير في تركيب المشاهد الجديدة، لكن الطرق الحالية لا تزال تواجه اختلافات بصرية ملحوظة عند معدلات أخذ عينات غير مرئية أثناء التدريب. تعاني هذه الطرق من تشويش ناتج عن التآكل عند التكبير وعيوب السلم عند التصغير، والتي يُعتقد أنها ناتجة عن القيود الأساسية في مزج ألفا المستخدم في طرق 3DGS. نقترح استبدال مزج ألفا التقليدي بمزج غاوسي جديد يعامل ألفا والشفافية كتواريخ متغيرة مكانيًا.

La introducción del 3D Gaussian Splatting (3DGS) ha avanzado la síntesis de nuevas vistas, pero los métodos existentes aún enfrentan discrepancias visuales a tasas de muestreo no vistas. Esto incluye artefactos de desenfoque al hacer zoom y artefactos en escalera al alejarse, atribuidos a limitaciones en la mezcla alfa. Se propone un nuevo enfoque, la Mezcla Gaussiana, que trata el alfa y la transmitancia como distribuciones que varían espacialmente, lo que podría mejorar la calidad del renderizado.

L'introduction du 3D Gaussian Splatting (3DGS) a fait progresser la synthèse de nouvelles vues, mais les méthodes existantes rencontrent encore des divergences visuelles à des taux d'échantillonnage non vus. Cela inclut des artefacts de flou lors du zoom avant et des artefacts en escalier lors du zoom arrière, attribués aux limitations du mélange alpha. Une nouvelle approche, le Gaussian Blending, est proposée pour traiter l'alpha et la transparence comme des distributions variant spatialement, ce qui pourrait améliorer la qualité du rendu.

The introduction of 3D Gaussian Splatting (3DGS) has advanced novel view synthesis, but existing methods still face visual discrepancies at unseen sampling rates. These include blurring artifacts when zooming in and staircase artifacts when zooming out, attributed to limitations in alpha blending. A new approach, Gaussian Blending, is proposed to treat alpha and transmittance as spatially varying distributions, potentially improving rendering quality.

Gaussian Blending: Rethinking Alpha Blending in 3D Gaussian Splatting

Was this article worth reading? Share it

LucidQuery AI

Blunge

CreativeDevJobs

GPTHumanizer

Golan AI

4o Image Gen

Ready to build your own newsroom?