Neural Flow Dynamics

Interactive visualization of neural networks, where digital neurons connect and evolve based on real-time user input.

Algorithmic Core

Neural Flow uses neural networks trained on hand-drawn sketches to generate dynamic path structures. The network's weights define the flow field patterns you see here, visualized using WebGL shaders.

Neural Input:


network.input([
    Math.sin(time * 0.002),
    pointer.x,
    pointer.y
])

Shader Output:


float strength = pow(
    sin(input.x * 3.1415),
    (output.y + 0.7)
);

Interactive Controls

🖱

Mouse Move

Drags create new neural pathways between nodes

💡

Double Click

Triggers synaptic plasticity - network grows new connections

🔄

Spacebar

Resets network to base activation pattern