Photogrammetry and 3D Gaussian Splatting for 4D plant growth capture and greenhouse digital twins

This paper presents a practical workflow that combines photogrammetry (Structure-from-Motion and Multi-View Stereo) with 3D Gaussian Splatting to digitise plants at scales from single flowers to greenhouse environments, using consumer-grade cameras. Tested on a 12-timepoint fern growth series and a greenhouse digital twin, it reports photorealistic rendering quality (PSNR 21.99–26.35 dB) and shows that leaf movement from wind markedly reduces reconstruction fidelity.