Nodes
0
Relations
0
Scenes
0
Duration
00:00
Output
1920×1080
SCENE 1 / 1
CAM 1.00×
3
00:00.000 / 00:00.000
Space: play / pause ← →: change scene R: record J/K/L: previous, pause, play
Write people and relations in JSON, and get a relation chart video with a moving camera
It fits any story with too many people to explain out loud. Show the whole chart first, then move the camera onto whoever you are talking about and grow the relation lines one at a time, and the viewer never loses track of where they are. Case summaries, org structures, project setups and character introductions all work.
The chart itself comes from nodes (people) and edges (relations); the pacing comes from shooting.scenes. A sample project is already loaded in the JSON tab, so the fastest start is to swap in your own names and relation labels and press play.
Keep one idea per scene. camera.targetNode picks who to move to and camera.zoom decides how close, while focusEdges and traceEdges say which relation is being talked about by animating the line. Wrap an area with box, add a note with callout, and put the spoken line into comment, which becomes the subtitle at the bottom of the frame.
| Resolution | Ratio | Typical use |
|---|---|---|
| 1920×1080 | 16:9 | Normal landscape video, embedding in slides |
| 1080×1920 | 9:16 | Short-form video and vertical social posts |
| 1080×1080 | 1:1 | Square posts in a feed |
| 2560×1440 | 16:9 | When the text needs to read large |
The running time here is the sum of the scene lengths (plus the intro and outro seconds). So "make it 30 seconds" would normally mean redistributing every scene by hand — and starting over each time a scene is added.
In the Record tab, "Total video length" takes a target and "Fit to this length" rewrites the scene lengths so the whole video runs for it. There are two ways to fit.
| Mode | What it does | Good for |
|---|---|---|
| Keep the current balance | Stretches or shrinks every scene, keeping their proportions | When you already have a sense of which beats should run longer |
| Equal length for every scene | Gives every scene the same length | Getting inside the running time first, then tuning by hand |
Only the scene lengths change. The number of scenes, their order and their contents stay as they are, and introSeconds / outroSeconds are left alone. Rewriting presentation choices to hit a running time would blur which numbers are yours and which the tool's.
A scene cannot go below 0.5s. A target under "intro + outro + scenes x 0.5s" is raised to that shortest workable length, and the panel says it was raised. If the proportional split would put a single scene under 0.5s, that scene is pinned to 0.5s and the remaining seconds are shared out again among the others (clamping each one in place would push the sum past the target).
For a vertical video (1080×1920), cut back on wide shots and lead with zooms. Holding a wide frame with many people makes the names too small to read. The preflight check in the Record tab flags scenes whose subtitle runs long and scenes shorter than 2.5 seconds.
Only the canvas is captured, so the controls and the side panel never appear in the recording. A music or narration file you load becomes the audio track of the video as it is.
An explainer video over a relationship diagram, where the camera zooms in and pulls back as the story goes. Each scene can set its own zoom target, highlighted people, red box, callout and subtitle, and the result is recorded as WebM.
In the JSON tab you write nodes (people), edges (relations) and shooting.scenes (the shots). A sample is loaded from the start, so you can begin by replacing the names and the relation labels. The Scenes tab lets you edit the same data through the screen.
Yes. Put the target in "Total video length" in the Record tab and press "Fit to this length"; the scene lengths are rewritten so the video runs for that long. You can keep the current balance between scenes or make them all equal. The number of scenes, their order and their contents, and the intro and outro seconds, do not change.
Yes. In the Record tab you can load a music or narration file, and you can record your microphone at the same time. Music and mic levels are set separately.
No. Parsing the JSON, drawing the diagram and recording the video all happen inside your browser. Nothing leaves your machine.