Portrait animation
Add a breath, a blink, or a slight head turn to a still portrait. Keep the expression calm so the face does not drift.
Video generation · Workflow guide
AI Image to Video: Turn a Photo Into Motion
A still photo already contains the composition. Image-to-video should add motion without redesigning the frame. The prompt has one job: tell the model what moves, how the camera moves, and what must not change.
- Independent guide
- No sign-in here
- Prompt runs locally
- Generator opens flashomni.ai
Quick answer
Upload one clear image and describe only the motion, camera move, and atmosphere. Protect the face, wardrobe, and framing. Keep the first version subtle: a head turn, drifting fog, or a slow push-in. Review identity and anatomy before adding a second effect.
At a glance
Best input
One sharp image with visible limbs and a clean edge
Best output
A 5-second clip with one motion cue
Video models
Kling 3.0, Veo 3.1, Sora 2, Gemini Omni Flash, Hailuo
Next step
Edit or extend the clip on Flash Omni
Build your prompt
Pick a preset, adjust the subject and scene, then copy the prompt. The optional photo preview stays in your browser: this page does not upload it. The generator is a separate site with its own terms and pricing.
1. Choose a starting point
No file selected
2. Copy the prompt
Copying the prompt does not create an account or charge credits. The hosted generator opens a separate site with its own terms and pricing.
What you can make
Product turntable
Move the camera around a packshot or rotate the product on a plinth. Protect the label and silhouette.
Living landscape
Animate clouds, water, fog, or traffic in a wide shot. The camera can stay locked while the scene moves.
Costume reveal
Turn a costume still into a short reveal with fabric movement, a head turn, and a slow push-in.
How it works
Pick a frame with clean edges
The model cannot reliably invent missing limbs. Choose an image where hands, feet, and hair are fully visible or clearly out of frame.
Choose one motion
One primary action per clip: turn, blink, drift, pour, or rotate. A second action is a second generation.
Add one camera move
A slow push-in, a gentle orbit, or a locked-off frame. Do not combine a dolly, a whip pan, and a zoom in the same shot.
Review identity and anatomy
Watch the face, hands, and costume seams frame by frame. If they drift, reduce the motion and generate again.
Image-to-video formula
Use this structure as a checklist. Replace the bracketed fields, keep the protection lines, and remove anything that does not apply to your shot.
Use the uploaded image as the exact first frame and visual reference.
Create a continuous [duration]-second cinematic shot.
Action: [one motion].
Camera: [one camera move].
Atmosphere: [fog, light, particles].
Keep the subject's face, costume, and body proportions consistent.
Do not change identity, add characters, or introduce scene cuts.
No text, no logos, no watermarks.- Say “exact first frame”. Without that phrase, some models treat the image as a loose mood reference.
- Keep the first clip short. Five seconds is enough to judge identity and motion.
- Describe atmosphere as a physical event: fog drifts left, rain falls, candles flicker. Avoid abstract mood words.
- If the face drifts, remove the head turn and animate the environment instead.
Model notes
These notes describe where each model tends to fit. They are not a ranking. Model availability changes by provider: WaveSpeed, SeeAPI, FAL, Replicate, and Kie all serve image and video models in this class. The hosted generator shows the available models and credit cost before you run a job.
- WaveSpeed
- SeeAPI
- FAL
- Replicate
- Kie
| Model | Best fit |
|---|---|
| Kling 3.0 | Character motion and cinematic camera moves when the reference already has a strong composition. |
| Veo 3.1 | Detailed scenes with coherent lighting and optional audio cues. |
| Sora 2 | Physically believable motion and atmosphere for short shots. |
| Gemini Omni Flash | Conversational, multi-turn editing after the first clip exists. |
| Hailuo | Expressive character motion for stylized or social-first clips. |
Three prompts you can adapt
A portrait that only breathes
The safest first animation. Small motion protects identity and gives the clip a documentary feel.
Use the uploaded portrait as the exact first frame. Create a continuous 5-second cinematic shot. Add a subtle breath, one natural blink, and a very slight head turn toward the camera. Slow push-in, 85mm lens, eye level. Keep the face, hair, skin tone, and outfit identical to the reference. Do not change identity, add characters, or cut the scene. No text, no logos, no watermarks.A packshot with a controlled orbit
The product stays still; the camera moves. This avoids warping the label and keeps the packshot usable.
Use the uploaded packshot as the exact first frame. Create a continuous 5-second product shot. Rotate the camera slowly around the product while a narrow highlight travels across the surface. Dark studio plinth, soft gradient, controlled reflections. Keep the product shape, label, cap, and logo placement exact. Do not add text or change the packaging. No watermarks.A landscape with moving weather
When the subject is a place, animate the atmosphere and leave the architecture alone.
Use the uploaded landscape as the exact first frame. Create a continuous 5-second cinematic shot. Drift low fog across the foreground and move the clouds slowly behind the horizon. Locked-off wide shot with a very slow push-in. Keep the horizon level and the buildings unchanged. Natural dusk light, realistic motion. No text, no logos, no watermarks.Common mistakes
Asking for a new scene
If the prompt describes a different location, the model may regenerate the image instead of animating it. Protect the reference first.
Too much motion
Fast dances, fights, and camera whips break anatomy. Start small and increase only after a clean result.
Low-resolution source
A soft or compressed image gives the model less identity information. Use the largest clean frame you have.
Cropped limbs
Hands and feet at the edge of the frame are the first things to duplicate or melt. Reframe before generating.
FAQ
What image works best for image-to-video?
A sharp, well-lit frame with a clear subject, visible hands, and clean edges. Portrait, product, and landscape references all work when the composition is readable.
How long should the clip be?
Start with five seconds. Short clips are cheaper to iterate on and easier to judge for identity drift. Extend or edit after the first result is clean.
Can I keep the person's identity?
Yes, when you protect the face, hair, skin tone, and body proportions in the prompt and keep the motion subtle. Review the eyes and hands before publishing.
Does image-to-video add audio?
Some models can generate audio and some cannot. Check the model's capabilities in the generator before relying on sound; add music or voice in your editor if needed.
Which model should I choose?
Use the model whose sample output matches your shot. Start with the default, compare one alternative, and pick based on identity stability and motion quality rather than marketing copy.
Can I edit the clip after generating it?
Yes. Use the video editor workflow to change one layer, such as background or light, while preserving the original motion and timing.
Related workflows
Turn the prompt into a clip
Open the hosted generator, paste the prompt, and run one short test. Keep the first generation small; a clean five-second clip is easier to judge than a long one.
Flash Omni is an independent guide and is not affiliated with Google. The hosted generator is a separate product with its own terms and pricing.