Image-to-video: generate the first frame, then animate it
Make a still with an image model first, fix the composition there, and only then ask the video model to animate it. Cheaper iterations and the look you chose, not the one the video model guessed.
STEP 1 — the still (image model): {subject with concrete details}, {setting}, {shot size and angle}, {lens feel}, {light}, {style and palette}, {aspect ratio 9:16 or 16:9}, sharp focus on {subject}, no text, no watermark. STEP 2 — the motion (video model, with the still as the first frame): Animate this frame. {subject} {one small action}. Camera: {one move from the vocabulary, or static}. Keep the composition, colours and lighting of the reference frame. Natural motion speed. {ambient sound}. 5 seconds.
Paste into the video model as one paragraph; remove the labels once you are happy with the content.
Steps
- Generate 4-8 stills, pick one, fix it there (regenerate, inpaint, upscale). Stills cost cents and seconds; video costs dollars and minutes.
- Upload the still as the first frame (Veo, Kling, Runway and Luma all accept one). Describe only the motion, not the picture: the picture is already decided.
- If the model drifts away from the frame, shorten the clip and reduce the action to one beat.
Example
Still: a ceramic mug on a wooden desk, steam rising, morning light from a window on the left, 85mm look, shallow depth of field, warm film palette, 9:16, no text. Motion: Animate this frame. Steam rises and curls slowly; a hand enters from the right and lifts the mug. Camera static. Keep the composition and lighting of the reference frame. Soft room tone. 5 seconds.
Tips
- For product shots and talking-object formats this is the only workflow that gives repeatable results across a series.
- Keep a folder of your chosen stills: they are your style guide for the next clips (see consistency entries).
| works with | Image models: Midjourney, Flux, Imagen: keyframes and style boards for image-to-video. · Veo: Google text-to-video / image-to-video prompts. · Kling: Kling text/image-to-video prompts. · Runway: Runway Gen prompts and camera controls. · Any LLM: Model-agnostic: Gemini, Mistral, a local model. |
| cost | Free. The video or image model you paste it into has its own pricing. |
| status | Unverified: written from the technique, not yet run end to end by us. |
| license | CC BY 4.0 for the text here; skills are MIT. |
| updated | 2026-11-01 |
| for agents | image-to-video-first-frame-method.md |
Related
Keep the same character across clips: the reference sheet prompt
A character description block you reuse verbatim in every prompt, plus a generated reference sheet, so a face, outfit and build stay the same from shot to shot.
Camera movement vocabulary that video models understand
The 14 camera moves that generation models respond to reliably, with the phrasing that works and the phrasing that produces a wobbling mess.
Hook for a generated video: what must be in frame 1 when a model draws it
The first-frame rules for Veo, Kling and Runway: start mid-action, name the motion already happening, keep text out of the generation, and generate the hook shot as its own 4-second prompt first.
Product hero shot: a 5-second generated clip for an ad or a listing
Prompt for a clean product shot with one motion and one light change, the kind that sits at the top of a listing or opens a 15-second ad.
The 7-part prompt structure for text-to-video
A fill-in template that gives any video model the seven things it needs: subject, action, setting, camera, light, style and sound. Fewer vague words, fewer wasted generations.
Anime style: 2D cel look, limited animation, clean line weight
A filled 7-part prompt for a hand-drawn anime shot: flat cel shading, animation on twos, a held painted background, and the three things the style breaks in current video models.