AR Fluid Sim That Collides with Real Objects

A webcam and polarization filter enable a fluid simulation to perceive and collide with physical objects placed in front of a screen. The live video feed is aligned to the simulation so silhouettes act as obstacles, with hands becoming playful, unintended interactors. The system uses a wind-tunnel style model informed by Tidepodious’s resources.

Key Points

• Goal: test a fluid simulation that collides with and responds to real-world objects placed in front of a screen.
• Setup: a webcam above the screen plus a polarization filter prevents feedback loops while capturing object silhouettes.
• Calibration aligns the camera feed with the simulation so the fluid treats captured shapes as obstacles.
• Hands are also detected as obstacles—initially a side effect but ultimately a fun interactive feature.
• The simulation uses a wind-tunnel style fluid model, with Tidepodious cited as a key resource.

Sentiment

Overall sentiment is strongly positive and collaborative, with enthusiasm for the concept and constructive technical suggestions; critiques are mild and center on practical alternatives and the absence of open code.

In Agreement

• The demo is impressive, playful, and ideal for science museum-style installations; scaling it up to large walls would be compelling.
• Using a polarization filter to block screen emit light is a simple, elegant solution to prevent feedback while preserving object tracking.
• Hands-as-obstacles make the interaction engaging and intuitive.
• Performance needs are modest for 2D fluids; it can run on integrated graphics, making the concept accessible.
• The approach fits well within educational/interactive art contexts and aligns with prior work like Dynamicland and AR sand tables.

Opposed

• No shared code was provided, limiting immediate reproducibility or community contributions.
• LCD-based polarization filtering may not work with OLED displays, reducing portability across screen types.
• Some suggested hardware paths (e.g., SUR40/FTIR) are dated or hard to source and can complicate behind-screen camera integration.
• Alternatives (IR floodlights, IR-absorbing film, high-speed IR cameras) may outperform polarization in varied lighting, implying the presented method isn’t universally optimal.