A strong image-to-video prompt describes motion, not a whole new image. Upload a clear source frame to the ClipTrend.ai image-to-video tool, then ask for one believable movement: a slow camera push-in, drifting hair, steam rising, clouds moving, or a product turntable.
Last updated: June 30, 2026 - about 8 min read
The image already gives the model the subject, composition, colors, and lighting. Your prompt should not fight that. It should tell the AI what changes over time.
This guide gives you prompt examples by source type, plus a simple formula you can reuse.
Use this structure:
Keep the original image composition. Add [specific motion] with [speed/camera direction]. Preserve [subject details]. Avoid [distortion risk].
That formula works because it respects the source image.
| Prompt part | Purpose | Example |
|---|---|---|
| Keep composition | Stops the scene from changing too much | "Keep the original framing" |
| Specific motion | Gives the clip a clear action | "slow camera push-in" |
| Speed | Makes motion feel natural | "gentle, slow, subtle" |
| Preserve details | Protects identity or product shape | "preserve the face and logo-free product form" |
| Avoid risk | Reduces common failures | "no warping, no extra fingers" |
Portraits need the least motion. The more recognizable the face, the smaller the motion should be.
Subtle living portrait
Keep the portrait composition unchanged. Add gentle breathing, a soft slow blink, and slight hair movement. Preserve the same face, expression, lighting, and background. No head turn, no new smile, no face warping.
Cinematic headshot
Keep the same headshot and expression. Add a slow camera push-in and very subtle light movement across the background. Preserve facial identity, eyes, mouth, and hairline.
Old photo
Keep the restored vintage portrait faithful to the original. Add one soft blink and a small natural breath. Preserve the original expression and face. No talking, no big smile, no dramatic head movement.
If faces drift, reduce motion. Do not add more instructions. For face-sensitive clips, see the Seedance 2 real-face reference guide and use it responsibly.
Product clips can handle more camera motion than portraits, but the product shape must stay stable.
Simple product reveal
Keep the product exactly the same. Add a slow studio camera push-in with soft light moving across the surface. Preserve the product shape, color, edges, and label-free design. No melting or shape changes.
Turntable feel
Create a subtle product turntable effect, as if the camera slowly moves around the object. Keep the product centered, sharp, and physically consistent. Clean studio lighting, no extra objects.
Ad-style social clip
Keep the product as the hero. Add a slow upward camera move and gentle background light shimmer. Preserve the product silhouette and material. Premium, clean, realistic motion.
For product photos, start from the image-to-video tool rather than a general text-to-video prompt. The source image gives the model the product details.

Landscapes are more forgiving. You can use weather, sky, water, and camera motion.
Travel photo
Keep the original travel photo composition. Add slow cinematic camera drift forward, clouds moving gently, and light wind through trees. Preserve buildings, horizon, and natural colors.
Ocean or lake
Animate the water with gentle ripples and a slow camera push-in. Keep the shoreline and horizon stable. Natural daylight, no dramatic storm, no warped perspective.
City street
Add subtle parallax and slow camera movement through the city scene. Keep architecture straight and realistic. Add small ambient motion only, no new people or vehicles.
The key is horizon stability. If the horizon bends, ask for less camera movement and more environmental motion.
Food motion works well when it focuses on steam, sauce, sparkle, or camera movement.
Hot dish
Keep the plated food composition unchanged. Add gentle steam rising and a slow camera push-in. Preserve the food shape, plate, table, and lighting. No melting or extra ingredients.
Drink
Keep the drink and glass exactly the same. Add small bubbles, light condensation, and a slow camera move. Preserve the glass shape and liquid level.
Bakery or dessert
Add a subtle camera glide and warm light movement across the dessert. Keep texture, toppings, and plate unchanged. No new decorations.
Food prompts fail when the model tries to "cook" the dish. Ask for atmosphere, not transformation.
Pets are harder than landscapes but easier than familiar human faces. Keep motion short.
Dog portrait
Keep the dog portrait composition. Add a small ear flick, soft blink, and gentle breathing. Preserve the same face, fur pattern, collar, and background. No running or body transformation.
Cat photo
Keep the cat in the same pose. Add a slow blink, slight whisker movement, and subtle tail motion if visible. Preserve fur markings and eye color. No extra limbs.
Camera movement is often safer than changing the subject. Use these:
Avoid stacking three camera moves at once. "Pan, zoom, orbit, tilt" in one prompt usually looks unstable.
Write the motion you want, not a full image description. The source image already defines the subject and style. A good prompt says what moves, how fast, and what must stay unchanged.
Use small motion and add a preservation line: "preserve the same face, expression, eyes, mouth, hairline, and lighting." Avoid big head turns, new smiles, talking, or dramatic emotion changes.
A slow push-in is the safest default. It adds life without forcing the subject to change shape. Products and landscapes can handle small orbit or pan moves; portraits usually need less.
Usually the prompt asked for too much motion or the source image was unclear. Reduce movement, crop closer to the subject, and preserve the important details explicitly. Our mistakes and fixes guide covers the common failure patterns.
Open the ClipTrend.ai image-to-video tool, upload your source frame, and write one motion prompt. If the subject stays stable, then you can make the clip more cinematic.