Image Comparisons

Detailed image quality comparisons across methods. Hover over images to zoom and inspect fine details.

Single-View Inputs

Compared to Flash3D, LGTM produces finer details with textured Gaussian prediction. Hover over images to Zoom.

Two-View Inputs

LGTM works with both unposed (NoPoSplat baseline) and posed (DepthSplat baseline) settings. Hover over images to Zoom.