reacTIVision Amoeba

Overview

The reacTIVision amoeba fiducial is a topological marker — its identity is encoded not in a grid of bits but in the nesting structure of black and white blobs (its "amoeba" shape) [1, 2]. reacTIVision, the open-source computer-vision framework that reads these markers, tracks each marker's ID, position and rotation and streams them over the TUIO protocol; it was built as the sensing layer for tangible tabletop interfaces [1]. Unlike the pose-from-square fiducials aruco and apriltag, amoeba markers target flat tabletop tracking (2D position + orientation on a surface) rather than full 6-DoF camera pose, but they are the same class of robustly-detectable ID-carrying visual code.

History

reacTIVision was developed by Martin Kaltenbrunner and Ross Bencina at the Music Technology Group, Universitat Pompeu Fabra (Barcelona), as the underlying sensor component of the Reactable — a tangible modular synthesizer that set the standard for tangible multi-touch applications [1]. The amoeba fiducial design and its evolution were described in Bencina & Kaltenbrunner's work around 2005, and the framework was presented as "reacTIVision: A Computer-Vision Framework for Table-Based Tangible Interaction" at TEI 2007 [2]. The project has been maintained for over a decade, with releases continuing into the 2020s [3].

Technical specification

Topological encoding. Each amoeba fiducial is detected by building a region adjacency tree (the topological tree of nested black and white regions) from the thresholded image; the unique sequence of region nesting encodes the marker's ID, while the centroids of the leaf regions give position and 2D orientation [1, 2]. Because identity is purely topological, detection is robust and the markers can be made small and irregularly shaped — the characteristic "amoeba" look results from a genetic / evolutionary algorithm that searched for compact shapes satisfying the required tree structures [1].

Fiducial sets. The default "amoeba" set contains 216 distinct symbols (symbols/default.pdf) [1, 3]. The framework also ships other sets and a beta "yamaarashi" set able to distinguish over one million different markers in a more compact footprint [3]. Multi-touch finger tracking (detecting round white regions as fingertips) is also supported but off by default [3].

Output / TUIO. reacTIVision transmits results using the TUIO protocol as Open Sound Control (OSC) messages over UDP (default port 3333 to localhost) [1, 3]. TUIO assigns a per-object session ID alongside the fiducial ID, so clients can track object lifetimes [1].

Use cases

• Tangible user interfaces — the canonical use; physical objects on a table become controllable tokens [1].
• The Reactable — tangible modular music synthesizer driven entirely by amoeba-tagged pucks [1].
• Multi-touch interactive surfaces — finger tracking on tabletop displays [3].
• Rapid TUI prototyping — its design goal as a toolkit [1].

Implementations

• mkalten/reacTIVision — C; the official cross-platform (Windows, macOS, Linux) framework, GPL v2+; includes the amoeba and yamaarashi symbol sets and TUIO output (~322 stars, active 2026) [3]. TUIO client libraries are LGPL, allowing use in proprietary clients [1].

Comparison

Versus aruco and apriltag: those are bit-grid square fiducials decoded by reading cells and recovering full 6-DoF camera pose; amoeba fiducials are topological (decoded from the region-nesting tree) and are aimed at 2D tabletop position/orientation streamed via TUIO rather than 6-DoF pose [1, 2]. The topological approach tolerates irregular, brand-friendly shapes and small sizes, and the yamaarashi set's >1M ID capacity rivals the large ID spaces of topotag; the trade-off is reliance on a controlled, near-planar tabletop imaging setup. It shares the topological-decoding idea later generalized by topotag.

Fun facts

• The "amoeba" shapes were not hand-drawn — they were evolved by a genetic algorithm to minimize marker size while keeping the required topology [1].
• The whole framework exists because of a musical instrument: the Reactable [1].

Status

Active, mature open source. reacTIVision remains maintained (repository active into 2026) and TUIO is a widely adopted standard for tangible/tabletop interaction [3].

Sources

1. reacTIVision (official site) — Kaltenbrunner & Bencina, accessed 2026
2. reacTIVision: A Computer-Vision Framework for Table-Based Tangible Interaction — Kaltenbrunner & Bencina, TEI, 2007
3. reacTIVision README — mkalten/reacTIVision, GitHub, accessed 2026

Deployments

No country reports mention this standard by name.

Regions / aggregations not mapped to a single country

• Universal