Prompt Guide

The extension queues a list of prompts and generates a video for each on vibes.ai. Prompts are separated by a blank line — each block becomes one row in the queue.

1. Writing good video prompts

Each clip is short, so describe one clear moment: the subject, the motion, the camera, and the mood. Keep it concrete.

A neon city skyline at night, slow aerial push-in, cinematic

A koi pond in soft rain, gentle ripples, close-up, calm

A white unicorn drinks from a lake, petals drifting, slow motion

2. Start & end frames

Every row can use frame images:

• No frames → text-to-video (generated from the prompt alone).
• Start frame only → the video begins from that image.
• Start + End → a start→end interpolation between the two images.

Upload frames with + Upload, or pick previously-uploaded ones from Library. Each media chip has an editable name — that name is how you reference it.

3. @tags — auto-pick frames in the prompt

Instead of setting the dropdowns by hand, name your media (e.g. forest, lyra) and reference them right in the prompt. The tags are stripped from the text sent to vibes.ai and used to fill the Start/End frame pickers:

@start:forest camera drifts through a misty forest at dawn

@start:lyra Lyra draws her bow, cloak stirring in the wind

@start:forest push through the trees @end:clearing

4. Settings that apply to the whole batch

• Resolution — 480p or 720p.
• Videos per prompt — 1 to 4 results per prompt (fewer uses less quota).
• Auto-download — save every result to a per-run folder, or turn off to preview only.

5. Retry, versions & resume

• Retry a completed job → generates a new version, kept alongside the originals (001_v5, 001_v6…).
• Retry a failed job → retries just the videos that failed, in place.
• Stopped mid-run? → Start again to Resume remaining or Start over.
• All done? → Start again to Regenerate all as new versions.