Laskos has released 360 Gaussian Pro 1.1.0, which densifies the app's COLMAP alignment by keeping points seen in only two views.
Laskos puts the gain at 2.7 times more points and lower reprojection error with the same runtime. Laskos launched 360 Gaussian Pro in August as a paid version of its free 360 Gaussian, preparing 360 footage for gaussian splatting with a custom COLMAP build and SAM3 masking. The app outputs a sparse COLMAP model, pinhole splits and masks, and training runs in Postshot, LichtFeld Studio, Brush or another COLMAP reader.
Two changes target the solve itself. The app caches a camera's lens distortion after its first successful alignment, and lens calibrations are now "only reused when the new solve measures better than the stored one." GPS scaling, which the Spatial matcher uses to set a splat to real world size, now compares the shape of the GPS track against the camera path and ignores bad fixes.
More metadata now travels with the exported images. Split images carry their focal length in EXIF for RealityScan and Metashape, and when the source video has GPS, every extracted frame and split image gets its position written to EXIF. A pinhole set exported from 360 Gaussian Pro can therefore go into either package with lens data, and location data when the video has GPS, already attached.
Insta360 .insv videos and .insp photos can now be batched into one project. The Masking tab gains a picker mode, the extraction timeline preview is smoother, and the release adds simple mesh generation and version rollback. The 1.1.0 changelog does not describe how the picker works, what kind of mesh is generated, or how far back a rollback reaches.
The intervening 1.0.2 release added masking keyboard shortcuts and hover tooltips, extended custom masks to stitched input, and fixed the GUT mask not being written to the 0_train_data folder.
360 Gaussian Pro checks for updates on launch and can update itself from inside the app, and one-time license holders receive every 1.x release free.
It is available now from the 360 Gaussian Pro website.
