Drop a .json or .lottie file and see exactly what is in it: composition size, frame rate, in and out points, the real duration, every layer with its type and timing, markers, embedded assets and file size. Layers inside precompositions are listed too, indented, because a file that shows three layers at the top level often holds forty once the precomps are opened — and that is where the weight of a heavy animation usually is. The duration shown is (op − ip) / fr rather than op / fr, which is the calculation most often got wrong. Everything is read in your browser and the file is never uploaded.
| Field | Meaning | Note |
|---|---|---|
| w / h | Composition width and height | In pixels, at the design size |
| fr | Frame rate | Frames per second |
| ip | In point | First frame that plays, in frames |
| op | Out point | Last frame, in frames |
| Duration | (op − ip) / fr | Not op / fr — the in point counts |
| ty | Layer type | 4 shape, 2 image, 5 text, 0 precomp |
| hd | Hidden | Does not render, still costs bytes |
| markers | Named segments | Used to play part of the timeline |
Duration is (op − ip) / fr — out point minus in point, divided by frame rate. If the in point is not zero, dividing the out point alone overstates the length. This page shows both points so you can see which case you are in.
The in and out points, in frames, not seconds. They define which part of the timeline actually plays. Most exports start at zero, but not all, and the ones that do not are where timing bugs come from.
Because layers inside precompositions are listed too, indented by depth. The top-level list shows one entry for an entire nested composition, which is misleading when you are trying to work out why a file is heavy.
The layer has Lottie's `hd` flag set, so it does not render. It still takes up space in the file and still gets parsed, so hidden layers are usually worth deleting before export rather than hiding.
They name segments of the timeline so a player can be told to play just that part. An animation with markers is generally meant to be driven by code rather than looped end to end.
Yes. A .lottie is a ZIP containing the animation and its assets; it is unpacked in the browser and the animation inside is read the same way.
No. It is read in the page and never sent anywhere.
A Lottie is JSON, so every property this page reports is read straight from the file rather than inferred. The header fields are terse and easy to misread: `w` and `h` are the composition size, `fr` the frame rate, and `ip` and `op` the in and out points in frames. Duration is `(op − ip) / fr`, and that subtraction is the part people skip — an animation whose in point is not zero is shorter than dividing the out point alone suggests, which is how a fifteen-frame lead-in silently becomes half a second of extra runtime in someone's timing calculation. The layer list is walked recursively rather than read off the top level, because `assets` holds both image assets and precompositions, and a precomposition carries its own `layers` array. A file that looks like three layers from the outside routinely contains forty once the precomps are opened, and that is exactly where the weight of a heavy animation is hiding, so nested layers are listed with their depth rather than hidden behind one entry. Distinguishing the two kinds of asset matters too: an entry with a `layers` array is a precomp and an entry with a string `p` is an image, and treating them alike either recurses into something that has no layers or misses the nesting entirely. Each layer's `ty` is mapped to its name — shape, image, text, solid, null, precomposition, camera — and the `hd` flag is surfaced, since a hidden layer is invisible but still costs bytes in the file and work in the parser. Markers are listed with their frame and duration, because they are how a player is told to play a named segment, and an animation with markers is usually meant to be driven rather than merely looped.