Abstract. We propose 6DGS to estimate the camera pose of a targetRGB image given a 3D Gaussian Splatting (3DGS) model representing the scene. 6DGS avoids the iterative process typical of analysis-bysynthesis methods (e.g. iNeRF) that also require an initialization of thecamera pose in order to converge. Instead, our method estimates a 6DoFpose by inverting the 3DGS rendering process. Starting from the objectsurface, we define a radiant Ellicell that uniformly generates rays departing from each ellipsoid that parameterize the 3DGS model. Each Ellicellray is associated with the rendering parameters of each ellipsoid, which inturn is used to obtain the best bindings between the target image pixelsand the cast rays. These pixel-ray bindings are then ranked to select thebest scoring bundle of rays, which their intersection provides the cameracenter and, in turn, the camera rotation. The proposed solution obviatesthe necessity of an “a priori” pose for initialization, and it solves 6DoFpose estimation in closed form, without the need for iterations. Moreover,compared to the existing Novel View Synthesis (NVS) baselines for poseestimation, 6DGS can improve the overall average rotational accuracyby 12% and translation accuracy by 22% on real scenes, despite not requiring any initialization pose. At the same time, our method operatesnear real-time, reaching 15f ps on consumer hardware.