Cr3ate

Cr3ate is a 3D reconstruction platform from Ydrive Inc., a Delaware company, built around the claim that a Gaussian splat can carry survey-grade metric accuracy rather than trading one against the other. The pipeline combines learned models with classical photogrammetry in a single system, running ingestion, camera calibration, feature extraction, pose, georeferencing, depth, mesh and Gaussian training as stages of one process.

Every intermediate result is computed in the same coordinate frame, so an exported mesh, a rendered splat and a point-cloud measurement all agree. Measurement of points, lines, areas and volumes is taken directly on the reconstructed model. The pipeline accepts drone, phone, 360 and aerial footage in the same project, takes stills and video together, and optionally ingests LiDAR, existing point clouds and known poses. Scenes run from a single room to city blocks, and the vendor states its input ceiling is not the 10,000 image limit common elsewhere. An AI pass removes people, cars and other moving objects during reconstruction.

Output is hierarchical Gaussians in a proprietary .rad container viewed in a Cr3ate web viewer, plus mesh, point cloud and Gaussian exports. An Unreal Engine plugin handles playback with relighting and actor support. A normal map viewer exposes the underlying geometry for accuracy checking in the browser.

The platform sells as three products. Cr3ate Desktop reconstructs locally with nothing leaving the machine, priced at $950 per year for one machine and scheduled to ship in Q4 2026, with three months free for anyone signing up at Intergeo. Address to 3D takes a street address and returns a model from Cr3ate servers. The Reconstruction SDK is a Python interface to the same pipeline, shipped as a Docker image for Linux with the NVIDIA Container Toolkit, restartable stage by stage, and scaling across single GPU, multi-GPU and distributed nodes. The free tier is limited to personal non-commercial use under terms effective November 1, 2025.

Pixomondo has the platform in its production pipeline, using it for scouting, techvis and previs in pre-production, for environment capture feeding in-camera VFX and car process on LED volumes during production, and for import into Unreal Engine in post. A Cr3ate reconstruction of Vancouver streets rendered live on a Pixomondo LED volume at CES 2025.