{"schemaVersion":1,"requestedSlug":"3d-gaussian-splatting","canonicalSlug":"3d-gaussian-splatting","article":"/effect/3d-gaussian-splatting","resource":"/api/public/atlas/effects/3d-gaussian-splatting","effect":{"slide":152,"kind":"historical","decade":"2020s","funFact":"The SIGGRAPH 2023 study reported real-time views at a minimum of 100 frames per second at 1080p.","body":[{"label":"Identify","text":"Photorealistic spaces with cloudy, floating transitions at uncertain points."},{"label":"Principle","text":"Optimized three-dimensional Gaussian distributions are quickly rendered as overlapping image contributions."},{"label":"Example","text":"Kerbl and team: 3D Gaussian Splatting, 2023."},{"label":"Try it out","text":"Compare a good view with the edge of the recorded space."}],"caption":"3D Gaussian Splatting: Research result Bicycle, 2023.","sourceUrl":"https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/","name":"3D Gaussian splatting","dating":"2023: SIGGRAPH work by Kerbl and team","short":"Gaussian Splatting","image":"/images/fx/s152.webp","slug":"3d-gaussian-splatting","context":["In 2023, Kerbl, Kopanas, Leimkühler, and Drettakis presented a method for high-quality, real-time view synthesis. It begins with calibrated cameras and a sparse point cloud, then optimizes spatial Gaussian distributions with extent and other properties. An efficient visibility-dependent renderer projects these contributions onto the image. Their work reports high-quality rendering at a minimum of 100 frames per second at 1080p for their tests. This performance figure applies to the research environment, not to any subsequent tool. Unlike a simple point cloud, the elements possess a spatial shape and overlapping effect. Edge artifacts indicate where the underlying views provide little information."],"question":"Why can a Gaussian splat look photorealistic even though it doesn't have a closed surface mesh?","answer":"The renderer assembles optimized spatial color and transparency contributions for the viewed view; a closed polygon skin is not necessary for this.","sources":[{"label":"Kerbl et al.: 3D Gaussian Splatting, SIGGRAPH 2023","url":"https://repo-sam.inria.fr/fungraph/3d-gaussian-splatting/"}],"credit":"Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler and George Drettakis / Inria.","title":"3D Gaussian Splatting","font":"Sora","highlights":["cloudy, floating transitions","three-dimensional Gaussian distributions","overlapping image contributions"],"margin":[{"term":"Gaussian distributions","note":"Smooth, bell-shaped mathematical profiles. In this technique, each describes a soft 3D element with position, extent, color, and opacity."},{"term":"rendered","note":"Converted from a scene representation into a viewable image. Many soft elements are combined to form each view."}],"fontWeight":"400","fontCategory":"Sans Serif"},"gallery":[{"src":"https://thumb.wikimedia.org/wikipedia/commons/thumb/a/a2/Polycam_gaussian_splatting_exemplo.png/960px-Polycam_gaussian_splatting_exemplo.png?utm_source=commons.wikimedia.org&utm_campaign=imageinfo&utm_content=thumbnail","page":"https://commons.wikimedia.org/wiki/File:Polycam_gaussian_splatting_exemplo.png","title":"Polycam gaussian splatting exemplo","artist":"Óscar Mirás","license":"CC BY-SA 4.0","note":"A garden fountain and lawn are shown with soft, painterly textures in the foreground. The image demonstrates how optimized Gaussians reconstruct complex organic environments from a sparse point cloud for real-time viewing."},{"src":"https://thumb.wikimedia.org/wikipedia/commons/thumb/e/e2/3D_Gaussian_Splatting_Methodology.png/960px-3D_Gaussian_Splatting_Methodology.png?utm_source=commons.wikimedia.org&utm_campaign=imageinfo&utm_content=thumbnail","page":"https://commons.wikimedia.org/wiki/File:3D_Gaussian_Splatting_Methodology.png","title":"3D Gaussian Splatting Methodology","artist":"Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, George Drettakis","license":"CC BY 4.0","note":"A flowchart illustrates the technical pipeline from point cloud initialization to tile rasterization. It outlines how spatial distributions are optimized through density control and gradient flow to produce the final rendered image."},{"src":"https://thumb.wikimedia.org/wikipedia/commons/thumb/c/c9/Gaussian_Splatting_GIF.gif/960px-Gaussian_Splatting_GIF.gif?utm_source=commons.wikimedia.org&utm_campaign=imageinfo&utm_content=thumbnail","page":"https://commons.wikimedia.org/wiki/File:Gaussian_Splatting_GIF.gif","title":"Gaussian Splatting GIF","artist":"Jurdein","license":"CC BY 4.0","note":"A camera movement through a reconstructed scene shows characteristic foggy artifacts at the edges. These floating, translucent shapes appear where the density of 3D Gaussians is insufficient to form solid surfaces."},{"src":"https://thumb.wikimedia.org/wikipedia/commons/thumb/2/2a/3dgs_viewer_QIBWA66MsC.jpg/960px-3dgs_viewer_QIBWA66MsC.jpg?utm_source=commons.wikimedia.org&utm_campaign=imageinfo&utm_content=thumbnail","page":"https://commons.wikimedia.org/wiki/File:3dgs_viewer_QIBWA66MsC.jpg","title":"3dgs viewer QIBWA66MsC","artist":"Jurdein","license":"CC BY-SA 4.0","note":"The user interface displays a reconstructed roof using a splatting viewer. Performance metrics and optimization parameters are visible, illustrating the real-time nature of the 2023 methodology."},{"src":"/images/gallery/3d-gaussian-splatting-ci2.webp","page":"/images/gallery/3d-gaussian-splatting-ci2.webp","title":"AI illustration: A photorealistic street scene transitions into cloudy, floating splats at","artist":"Illustration generated for the Atlas (AI)","license":"Atlas illustration","note":"A photorealistic street scene transitions into cloudy, floating splats at the frame edges. These peripheral elements illustrate the underlying Gaussian primitives used to render three-dimensional volumes beyond the optimized reconstruction zone."},{"src":"/images/gallery/3d-gaussian-splatting-di2.webp","page":"/images/gallery/3d-gaussian-splatting-di2.webp","title":"AI illustration: A clear central garden view transitions into blurred, semi-transparent el","artist":"Illustration generated for the Atlas (AI)","license":"Atlas illustration","note":"A clear central garden view transitions into blurred, semi-transparent ellipsoidal shapes at the periphery. This illustrates how optimized Gaussian distributions represent scene data at the edges of recorded space."}],"prompt":"Apply a 3D Gaussian Splatting look with a photorealistic central subject surrounded by softly overlapping translucent splats in uncertain background areas. Preserve convincing illumination and surface color while adding sparse floating flecks, cloudy depth transitions, and gently dissolving peripheral edges. Keep the original subject and composition."}