How to inspect sprite animation timing
Before measuring an animation, check whether the download contains a sequence at all. This walkthrough uses the available Blue Coin from Super Mario 64. Its prepared sprite contains one frame, not a reconstructed coin-spin animation. That makes it a useful example of the difference between a visible sprite and evidence of in-game timing.

1. Check the asset and its frame manifest
Open the Blue Coin page, then inspect its frame manifest. The frames array has one entry with width and height both set to 31. The source record describes a planar exported GLB pose composed from geometry, UVs, material and vertex color/alpha. It explicitly says this is not a reconstructed runtime animation.
2. Record what the data does and does not establish
The manifest records timingVerified as false. It does not give a frame duration or a multi-frame loop. Do not invent a rotation speed, duplicate this image into a sequence and call it original timing, or interpret a static preview as proof that the original object never animated.
Save the PNG and manifest together when documenting your inspection. Keep the source identifier and hashes so you can distinguish this prepared view from a later export. The original GLB is a separate representation, available from the asset page.
3. Apply the same checks to a multi-frame resource
When a resource does contain several frames, pause its preview and inspect every frame in order. Preserve full canvas sizes, trim offsets and origins where recorded. Check the wrap from last frame to first for extra holds or duplicates. Use nearest-neighbor scaling to examine pixel edges without smoothing.
Keep three values separate: a preview player’s chosen delay, a stored source default, and timing verified against the game. A GIF’s delay may only describe a prepared preview. A default speed may be overridden by game logic. Neither independently proves the original runtime behavior.
4. Measure only with appropriate evidence
For a sequence with documented equal frame durations, multiply the number of frames by the duration to calculate one loop. For unequal durations, add them. If the resource has no verified durations, report timing as unknown rather than supplying a plausible frame rate.
Exact reconstruction needs relevant runtime logic or a measured reference of the specific action, including speed changes, repeated frames and holds. If you choose timing for your own project, label it as your choice. For this Blue Coin export, the defensible result is one prepared image with unverified animation timing.