Hi, Thanks for your excellent work! After checking both papers, I found that the ScanNet dataset used in SceneSplat (ICCV 2025) and SceneSplat++ (current work) points to the same GaussianWorld/scannet_mcmc_1.5M_3dgs repository. However, the reported metrics for ScanNet are not fully consistent between GaussianWorld/scene_splat_49k and GaussianWorld/scene_splat_7k.
Could you confirm whether the two papers fully share the same ScanNet 3DGS data, and whether the metric difference is simply due to the statistics results in GaussianWorld/scene_splat_7k not being updated to reflect the current dataset?
Thank you!
Hi, Thanks for your excellent work! After checking both papers, I found that the ScanNet dataset used in SceneSplat (ICCV 2025) and SceneSplat++ (current work) points to the same GaussianWorld/scannet_mcmc_1.5M_3dgs repository. However, the reported metrics for ScanNet are not fully consistent between GaussianWorld/scene_splat_49k and GaussianWorld/scene_splat_7k.
Could you confirm whether the two papers fully share the same ScanNet 3DGS data, and whether the metric difference is simply due to the statistics results in GaussianWorld/scene_splat_7k not being updated to reflect the current dataset?
Thank you!