Image-to-video prompts for better meme reactions

Describe the movement you want to add

With image-to-video, your photo supplies much of the visual starting point. Your prompt should explain what changes over the next few seconds. “The dog tilts its head, pauses, and gives the camera a puzzled look” is more useful than repeating every visible detail of the sofa.

Treat the prompt as a direction for a short shot. Keep one primary subject, one action, and a simple ending. If you ask for several location changes, new characters, a transformation, and complicated dialogue in the same short clip, you make the intended joke harder to communicate.

Use a four-part prompt

A practical structure is subject + action + camera + mood. Start with the visible subject, add one readable action, choose a camera movement, and finish with the tone. Add constraints only when they matter to the shot.

Example: “The orange cat at the desk slowly looks from the laptop to the camera, pauses with an unimpressed expression, and blinks once. Gentle camera push-in, warm office lighting, deadpan comedy, one continuous shot, no text or logos.”

• Subject: the visible person, animal, or object.
• Action: one small movement that carries the reaction.
• Camera: static, slow push-in, or a gentle sideways move.
• Mood: deadpan, delighted, confused, or playfully dramatic.

Three original prompts to adapt

Work reaction: “The person at the desk hears an imaginary notification, looks toward the camera, and slowly lowers their coffee cup. A subtle push-in, warm natural light, restrained deadpan expression, no new people or text.”

Pet reaction: “The dog notices something just outside the frame, tilts its head, then looks back at the viewer. Locked camera, soft daylight, gentle comic timing, natural proportions, no text.”

Small victory: “The character notices a finished task, smiles, and makes one tiny celebratory fist pump. A slow confident camera move, bright friendly light, humorous heroic mood, one uninterrupted shot, no lettering.”

Let the editor handle the written joke

Keep the scene prompt focused on the action, then add exact meme captions in the editor. This gives you direct control over the words, spelling, size, and placement. The scene can do the acting while the caption explains why the audience should find it funny.

Avoid putting private information into a prompt. You do not need a person’s employer, address, account name, or full identity to describe an expression or camera movement. Use a photo you have permission to edit.

Change one thing when a result misses

If the subject moves too much, simplify the action. If the joke feels flat, make the reaction more specific. If the camera distracts from the expression, ask for a locked camera. Changing one instruction at a time makes it easier to see which choice affected the result.

Generative output varies. A prompt is a direction, not a guarantee of identity consistency, exact timing, or a particular expression. Preview before downloading and keep the version that makes the joke clearest.

Know which workflow you are using

These prompts are for the AI image-to-video service when it is enabled. The browser studio uses pan, zoom, float, or shake on a still image; it does not interpret a written prompt to create new subject motion.

If AI access is unavailable, use the same joke idea with a strong still image and an edited caption. You can still produce a real animated video with the browser tools, while keeping the distinction between animation and generation clear.