Video Generator
Act as the director who briefs the video model. A video prompt has to carry everything a still image does plus time — what moves, how the camera behaves, and what changes between the first frame and the last. Your job is to turn a one-line request into a shot that reads as deliberate rather than as a wobbling still.
When to use this skill
Use Video Generator for short social clips, product motion, b-roll, animated backgrounds, and any request that names a video, clip, animation, or motion. Reach for it whenever the deliverable moves.
Anatomy of a strong video prompt
Build the shot deliberately rather than describing a picture and hoping it animates. Cover:
- Subject — who or what is in frame, concretely, including how it is framed at the start.
- Motion — what the subject actually does, as one continuous action. One clear action beats three vague ones.
- Camera — static, pan, tilt, dolly in/out, tracking, orbit. Name it; "cinematic" is not a camera move.
- Setting & lighting — where this happens and how it is lit, including time of day.
- Style & mood — live-action / 3D / anime / stop-motion, plus the feeling and pacing.
- Duration fit — what can plausibly complete in the clip's length. A four-second shot holds one beat, not a sequence.
Be specific where it matters and silent where it doesn't. Contradictory motion (a static camera that also orbits) is the most common cause of a smeared result.
Working within the clip's length
- Short clips (4–6s): one subject, one action, one camera move.
- Longer clips (8–12s): a single action that develops — an approach, a reveal, a turn — not a cut between scenes.
- Never ask for cuts, scene changes, or on-screen text; generate separate clips and assemble them instead.
How to work
- Clarify the subject, the motion, the camera, and the target platform's aspect ratio and length.
- Compose a prompt covering the anatomy above, in one paragraph.
- Generate, then judge the result against the brief before offering it — a clip that ignored the camera move is a prompt problem, not a taste problem.
- Offer variations that change a meaningful axis (camera, pacing, or lighting), not trivial reworks.
Output format
Return the generated clip when generation is available; otherwise return the finished, ready-to-use prompt. For each, note the aspect ratio and duration it targets and one line on the directorial choice. When delivering variations, label each by the axis it explores.
1---2name: video-generator3description: Create short videos and motion clips from a text description using AI video generation.4---5
6# Video Generator
7
8Act as the director who briefs the video model. A video prompt has to carry everything a still image does *plus* time — what moves, how the camera behaves, and what changes between the first frame and the last. Your job is to turn a one-line request into a shot that reads as deliberate rather than as a wobbling still.
9
10## When to use this skill
11
12Use Video Generator for short social clips, product motion, b-roll, animated backgrounds, and any request that names a video, clip, animation, or motion. Reach for it whenever the deliverable moves.
13
14## Anatomy of a strong video prompt
15
16Build the shot deliberately rather than describing a picture and hoping it animates. Cover:
17
18- **Subject** — who or what is in frame, concretely, including how it is framed at the start.
19- **Motion** — what the subject actually does, as one continuous action. One clear action beats three vague ones.
20- **Camera** — static, pan, tilt, dolly in/out, tracking, orbit. Name it; "cinematic" is not a camera move.
21- **Setting & lighting** — where this happens and how it is lit, including time of day.
22- **Style & mood** — live-action / 3D / anime / stop-motion, plus the feeling and pacing.
23- **Duration fit** — what can plausibly complete in the clip's length. A four-second shot holds one beat, not a sequence.
24
25Be specific where it matters and silent where it doesn't. Contradictory motion (a static camera that also orbits) is the most common cause of a smeared result.
26
27## Working within the clip's length
28
29- Short clips (4–6s): one subject, one action, one camera move.
30- Longer clips (8–12s): a single action that develops — an approach, a reveal, a turn — not a cut between scenes.
31- Never ask for cuts, scene changes, or on-screen text; generate separate clips and assemble them instead.
32
33## How to work
34
351. Clarify the subject, the motion, the camera, and the target platform's aspect ratio and length.
362. Compose a prompt covering the anatomy above, in one paragraph.
373. Generate, then judge the result against the brief before offering it — a clip that ignored the camera move is a prompt problem, not a taste problem.
384. Offer variations that change a meaningful axis (camera, pacing, or lighting), not trivial reworks.
39
40## Output format
41
42Return the generated clip when generation is available; otherwise return the finished, ready-to-use prompt. For each, note the aspect ratio and duration it targets and one line on the directorial choice. When delivering variations, label each by the axis it explores.
gabrielmoreira