Description
With AI-generated video, the hardest part is telling whether it actually followed your prompt: which line of dialogue made it to screen, which camera move was dropped, whether continuity drifted. SceneFlow aligns the script and the video on one timeline: during playback it highlights exactly which part of the script is on screen right now, so you can see at a glance whether the prompt instructions were really rendered.
It breaks the script into eight color-coded cue types (dialogue, action, camera, shot, audio, VFX, transition, environment) pinned to video timestamps, and a multi-track timeline lets you check the AI video against the prompt line by line. It supports both traditional screenplay format and state-driven Auteur Script with 5-part staging metadata; projects save as JSON to share and version-control. It is a web app, ready to use. Sync cues are mapped manually or generated externally by multimodal models like Gemini that match video frames to script timecodes.
Script highlights with playback: Highlights the current script segment in real time so you see whether prompt lines reached the screen.
Eight color-coded cues: Dialogue, action, camera, shot, audio, VFX, transition and environment, with a color-blind-safe palette.
Adherence analysis: Compare prompt against footage line by line to pinpoint missed elements, camera drift and continuity deviations.
Multi-track timeline: An NLE-style horizontal timeline lays cue types across lanes, checked against an anticipation playhead.
Dual script formats: Handles both traditional screenplays and state-chained Auteur Scripts.
JSON projects: Projects save as JSON for import/export, URL sharing and version control.
It breaks the script into eight color-coded cue types (dialogue, action, camera, shot, audio, VFX, transition, environment) pinned to video timestamps, and a multi-track timeline lets you check the AI video against the prompt line by line. It supports both traditional screenplay format and state-driven Auteur Script with 5-part staging metadata; projects save as JSON to share and version-control. It is a web app, ready to use. Sync cues are mapped manually or generated externally by multimodal models like Gemini that match video frames to script timecodes.
Features
Script highlights with playback: Highlights the current script segment in real time so you see whether prompt lines reached the screen.
Eight color-coded cues: Dialogue, action, camera, shot, audio, VFX, transition and environment, with a color-blind-safe palette.
Adherence analysis: Compare prompt against footage line by line to pinpoint missed elements, camera drift and continuity deviations.
Multi-track timeline: An NLE-style horizontal timeline lays cue types across lanes, checked against an anticipation playhead.
Dual script formats: Handles both traditional screenplays and state-chained Auteur Scripts.
JSON projects: Projects save as JSON for import/export, URL sharing and version control.
