Altus Collab
Architecting an ultra-low latency collaborative vector whiteboard engine to allow concurrent high-fidelity design work across global networks.
Stuttering whiteboards.
The Altus whiteboard engine was built on traditional database-synced components. Pointers stuttered, vector paths shifted after completion, and concurrent editing sessions with over 5 users became visually unusable due to conflicting state locks.
They needed a canvas renderer that separated visual presentation calculations from transactional network pushes, achieving zero latency perception.
Local-first drawing.
We designed a dual-buffer canvas engine. Mouse coordinates are tracked and drawn locally at a full 60fps immediately, bypassing all database handshakes.
Concurrently, coordinate datasets are serialized as light WebSocket packets and pushed to peer clients. Peer cursor movements are smoothed using mathematical bezier interpolation, resolving jitter on unstable networks.
Clean architecture.
We styled the workspace to minimize layout clutter, maximizing canvas space. Pointers show peer avatars and carry clear, legible text overlays indicating editor roles.
60fps Collaboration.
The redesigned workspace achieved full 60fps rendering, even during sessions with over 50 concurrent designers. Whiteboard lagging errors fell to zero, and user retention metrics increased by 85% within three months of launch.
The canvas library is currently being integrated into all of Altus's sub-applications.