Build the pieces around a Canvas renderer

Canvas supplies drawing operations, but a game needs more than a drawing surface. Scene state, asset loading, animation scheduling, input mapping and hit testing belong in your application or in a framework you deliberately choose.

Keep a small game’s model separate from its renderer. A player object can store position and velocity without storing the drawing context. That separation makes collision tests and replayable input easier to reason about.

Build only the systems your prototype needs. A simple asset loader and one scene transition are more useful than a general engine built before the first playable interaction. Add reusable abstractions after two real uses reveal what they need to share.

A practical first milestone

• Load one sprite and show a loading error if it fails.
• Move it with a keyboard and a touch button.
• Pause and resume without a large movement jump.
• Restart from a known initial state.