What a Canvas Sprite Sheet Needs
A canvas does not know about frames. It only draws images at coordinates you give it, so everything about animation, from timing to clearing, is your code's job rather than the platform's. A sprite sheet is therefore just one image, and each frame is a source rectangle inside it that you pass to drawImage.
A manifest gives you those source rectangles as numbers, so you do not need an engine or a plugin, only the values and the image itself. Keeping a small game dependency-free is often the whole point. The manifest also states fps and a loop flag per state, which your animation loop reads to decide when to advance and whether to wrap.
Because the layout lives in data, the same render loop can play any sheet. You swap the JSON and the image, and the player keeps working with no code changes, which keeps a small codebase tidy. That separation also means a designer can add a new character without asking a programmer to touch the render loop.
A sheet also reduces network requests, since one image replaces a folder of frames. Fewer requests means faster loading and fewer chances for a missing file. For a browser game, that simplicity is often worth more than any engine feature.

