Pixel Farm updates PFTrack with new machine learning features and scene reconstruction improvements

Pixel Farm has released PFTrack version 24.12.19, introducing several updates to its 3D scene reconstruction and matchmoving software. One of the most notable changes is the overhaul of the Photo Survey Node, now featuring a machine learning-based feature-matching engine that improves accuracy and significantly reduces processing time, particularly for large datasets.

The update also brings improvements to the Photo Cloud, improving distortion estimation and delivering faster, more accurate depth map creation. New cleanup tools also allow users to tidy up point clouds before meshing scenes.

Texture creation capabilities have been updated, providing more accurate results even with lower-quality input meshes. Additionally, the Match Camera Node has been optimized for faster, more precise performance when solving multiple cameras within a single scene.

PFTrack’s USD support has also been expanded to include point-based geometry and locator access. The release features a new integrated help menu, consolidating training materials, user forums, and support services with AI-assisted search tools for easier navigation.

This version includes various UI and performance improvements across the application. More details can be found on Pixel Farm’s website.