Digital twins for plants: how ordinary cameras are helping preserve living collections

Researchers at the University of Tartu have shown that everyday cameras can create detailed, explorable 3D models of plants, giving botanical gardens and museums an affordable way to document and preserve living collections. The study, by Toivo Aukusti Ylinampa and Urmas Kõljalg, was published in the open-access Biodiversity Data Journal.

Plants are three-dimensional and constantly changing, which makes them hard to capture in flat photographs or pressed specimens. The team combined two techniques: photogrammetry, which reconstructs measurable 3D shapes from overlapping photographs, and 3D Gaussian Splatting, a newer method that produces photorealistic models in real time and handles thin, tangled foliage well. Working mostly with a compact Olympus TG-6 camera, supplemented by a macro-lens camera, a smartphone and a drone, the researchers digitised subjects ranging from a single orchid flower to a tree.

Two datasets stand out. In the first, about 8,500 photographs of a fern, taken on twelve days across a fifteen-day period, were aligned together using stable background features rather than fixed cameras. Each day’s model shares the same scale and orientation, allowing the fern’s growth to be compared directly over time. In the second, about 20,000 photographs of the University of Tartu Botanical Garden’s tropical greenhouse were processed in fourteen overlapping sections and assembled into an interactive virtual walkthrough.

 “After running thousands of photogrammetric scans I realised that not all the images have to be directly related. While two pieces of fern won’t share many comparable feature points, the surrounding area has lots of them — for free!” 

- said the lead author, Toivo Ylinampa of University of Tartu.

However, the team noted several limitations. Even gentle wind noticeably reduced model quality, and the greenhouse sections were assembled manually, so the walkthrough is suited to visual exploration rather than precise measurement. On the other hand, in a macro test, a measurement taken on the model (13 mm) closely matched a caliper reading (13.2 mm), supporting the approach for scaled measurements. The authors have identified automated stitching of sections and independent quantitative validation as next steps.

Because the workflow relies on consumer cameras and widely available software, it can be adopted by botanical gardens, museums and research groups without specialised systems. In the paper, the authors write that the workflow “provides practical guidance for generating scalable 3D datasets” for plant documentation, living collections, digital exhibitions and future quantitative analyses. The 3D models and demonstration videos are available online through links in the paper.

Original source:
Ylinampa T, Kõljalg U (2026) Photogrammetry and 3D Gaussian Splatting for 4D plant growth capture and greenhouse digital twins. Biodiversity Data Journal 14: e191083. https://doi.org/10.3897/BDJ.14.e191083