Vultorch Tutorial

A step-by-step guide through Vultorch, one example at a time.
Each chapter maps to a runnable script in the examples/ folder.

Chapter	Topic	Key concepts
01 — Hello Tensor	Minimal display	View, Panel, Canvas, bind, run
02 — Multi-Panel	Multiple panels & canvases	Layout, side, multi-canvas
03 — Training Test	Fit a tiny network to a GT image	custom dock layout, create_tensor, per-pixel optimization
04 — Conway's Game of Life	GPU cellular automaton	create_tensor for simulation, filter="nearest", sidebar, buttons, color pickers
05 — Image Viewer	Load, transform & save images	imread, imwrite, Canvas.save, combo, input_text, filter toggle
06 — Pixel Canvas	Interactive drawing on a GPU tensor	mouse interaction, screen→pixel mapping, backing store pattern
07 — Multi-Channel Viewer	RGB + depth + normal + alpha in one window	multiple zero-copy tensors, turbo colormap, ray-sphere intersection
08 — GT vs Prediction	Live training comparison with error heatmap	error heatmap, PSNR, loss curves, error mode switching
09 — Live Hyperparameter Tuning	Change LR, optimizer, loss at runtime	step()/end_step(), log-scale LR, optimizer hot-swap
10 — 2D Gaussian Splatting	Differentiable 2D Gaussian rendering	nn.Parameter, alpha compositing, cumprod transmittance
11 — 3D Surface Inspector	Orbit camera with Blinn-Phong lighting	SceneView, Camera, Light, MSAA, procedural textures
12 — Neural Rendering Workstation	Capstone: 6-panel workstation with dual-head MLP	Two-head MLP, six panels, pause/resume, snapshot, optimizer hot-swap
13 — Snake RL	DQN learns to play Snake	RL visualization, DQN, ε-greedy, Q-value heatmap, manual mode