✳ SOURCED · KOLDFLUX STRUCTURE · DATASET-V2.1
Per-platform caption render matrix
7 platforms. Each row documents caption support, character limit, single-cue duration, render position, recommended lines, recommended chars per line, source URL, verified-at date, and confidence.
Per-platform matrix
| Platform | Formats | Max chars | Max cue duration | Render position | Recommended lines | Recommended chars / line | Confidence | Source |
|---|---|---|---|---|---|---|---|---|
| TikTok | burn-in, srt, embedded, vtt | 2200 | — | bottom-center | 2 | 42 | high | developers.tiktok.com/docs/en/co… · 2026-09-12 |
| Instagram Reels | burn-in, srt, embedded, vtt | 2200 | — | bottom-center | 2 | 42 | medium | developers.facebook.com/docs/ins… · 2026-09-12 |
| YouTube Shorts | burn-in, srt, vtt | 100 | 60 s | bottom-center | 2 | 36 | high | support.google.com/youtube/answe… · 2026-09-12 |
| burn-in, srt, vtt | 700 | 600 s | bottom-center, no native styling | 2 | 42 | medium | www.linkedin.com/help/linkedin/a… · 2026-09-12 | |
| X (Twitter) | burn-in, vtt | 280 | 140 s | bottom-center, no native styling | 2 | 42 | medium | help.x.com/en/using-x/media-stud… · 2026-09-12 |
| Snapchat | burn-in, srt, vtt | — | 60 s | bottom-center | 2 | 36 | medium | help.snapchat.com/hc/en-us/artic… · 2026-09-12 |
| Spotify Canvas | burn-in, vtt | — | 8 s | no captions (visual loop) | 2 | 36 | high | artists.spotify.com/canvas… · 2026-09-12 |
Methodology
Each row sources from the platform's published creator help (SRT/VTT/ASS upload paths) + the Koldflux Caption QA primitive. Limits are sourced verbatim; recommended-lines + chars/line are derived from render-width / safe-zone geometry. Where the platform doesn't publish a limit, the cell is - (UNKNOWN).