Video Generation
Scene Continuity
How to maintain visual consistency between scenes in Flenly
Scene Continuity
Scene continuity helps your video look like a coherent story rather than a series of disconnected clips. When enabled, the last frame of one scene is used as a visual reference for the next scene.
How It Works
- After a scene finishes rendering, Flenly automatically extracts the last frame of the video.
- This frame is stored and associated with the scene.
- When you render the next scene with continuity enabled, the last frame is passed as a start image to the AI model.
- The AI uses this reference to generate video that visually flows from where the previous scene ended.
Enabling Continuity
When configuring a render, you'll see a continuity toggle. Turn it on to use the previous scene's last frame as the starting point.
Not all models support start images. Models that work well with continuity:
- Wan 2.6 I2V — Designed for image-to-video, excellent continuity
- Hailuo 2.3 Fast — Fast image-to-video rendering
- Seedance 1.5 Pro — Supports end frame control
- PixVerse v5 — Supports end frame
For models without native start-image support, the continuity reference is included in the prompt context instead.
Tips for Better Continuity
- Render scenes in order — Continuity works best when you render sequentially so each scene can reference the one before it.
- Keep prompts consistent — Describe the same characters, setting, and style across scenes.
- Use characters — Link a character profile to maintain consistent character appearance.
- Use style presets — Apply the same style preset across scenes for visual consistency.
Limitations
- Continuity is a visual hint, not a guarantee. The AI may interpret the reference frame differently depending on the prompt.
- Large changes in subject, setting, or style between scenes may reduce the effectiveness of continuity.
- The last frame is extracted automatically — you can't manually select which frame to use.