Unlimited Seedance, Kling, Wan and Seedream 5.0, now on KreatorFlow · available now · Fair use

See passes
An instant photo of a mountain lake held up in front of the real lake, perfectly lined up
Prompting

Official BytePlus partnerBytePlus, a ByteDance company

Image to video with AI: how to get consistent results

How to turn a photo into an AI video that still looks like the original: choosing the image, writing the motion prompt and using up to 9 references.

KreatorFlow6 min read

Short answer

Start from a sharp, well-lit image, then describe only what should move and how the camera behaves; do not re-describe what is already in the picture. With Seedance 2.0 you can add up to 9 reference images (for example a character, a product and a location) to one generation. For people you want to reuse, save them as a character first. A 5-second clip costs 15 credits at 720p.

Updated:

Step by step

  1. 1

    Choose the right image

    Sharp, evenly lit and uncluttered. Faces front-facing or three-quarter, products fully in frame. The model can only keep what it can see.

    A good source image for image to video: a sharp, evenly lit, front-facing portrait on a plain background
  2. 2

    Open image to video

    In the video studio, switch to image to video and upload the photo or pick one from your assets.

  3. 3

    Describe the motion, not the picture

    "She turns towards the camera and smiles as the wind moves her hair. Slow push in." The image already shows who she is.

  4. 4

    Keep the move modest

    Small, natural motion keeps the subject recognisable. A full 360 spin from one photo asks the model to invent the sides it has never seen.

  5. 5

    Add references for anything exact

    Use extra references for a product, a logo or a second person. Up to 9 images fit in one generation.

Motion prompts for photos

  • Portrait

    She looks up from her book and smiles at the camera. Soft window light, slow push in, gentle breeze through the curtains.

  • Product

    The perfume bottle rotates slowly on its base as light sweeps across the glass. Static camera, dark studio background.

  • Landscape

    Clouds drift across the mountains and the lake ripples in the wind. Slow pan right.

  • Food

    Steam rises from the bowl of ramen as chopsticks lift a few noodles towards the camera. Close-up, slow push in, warm restaurant light.

  • Pet

    The dog tilts its head, ears perking up, then wags its tail and steps towards the camera. Static camera, soft afternoon light.

  • Old photo

    The couple in the photo turn to each other and smile, the woman's dress moving slightly in the breeze. Gentle film grain, slow push in.

Seedance price per second on KreatorFlow

Seedance 2.0 and 2.5 at 720p and above: price per second on each plan (monthly and annual), credits per 5-second clip, and clips per plan.

Model · resolutionCreditsPer secondUSD per second (monthly)USD per second (annual)Creator · 525 credits/moPro · 1,050 credits/moMax · 2,500 credits/moBusiness (2 seats) · 4,500 credits/mo
Seedance 2.0 · 720p15/5s3/s$0.119–0.145/s$0.099–0.121/s3570166300
Seedance 2.0 · 1080p30/5s6/s$0.238–0.289/s$0.198–0.241/s173583150
Seedance 2.0 · 4K70/5s14/s$0.554–0.675/s$0.462–0.563/sPro and up153564
Seedance 2.5 · 720p24/5s4.8/s$0.190–0.231/s$0.158–0.193/s2143104187
Seedance 2.5 · 1080p50/5s10/s$0.396–0.482/s$0.330–0.402/s10215090
Seedance 2.5 · 4K120/5s24/s$0.950–1.157/s$0.792–0.964/sPro and up82037

Seedance 2.0 and 2.5, 720p and above. Plan columns: 5-second clips a month on the plan's full monthly credits. Longer clips cost the same per second, so a 10-second clip uses twice the credits. Generated audio is included. USD per second: each plan's price over its full monthly credits, lowest to highest plan; monthly is the regular monthly price, annual is the yearly price divided by 12.

Why image to video drifts away from the photo

A photo is one moment from one angle. As soon as the subject moves or the camera travels, the model has to invent what the photo never showed: the side of a face, the back of a product, the room behind the door. The more it has to invent, the more the clip drifts from your original. That is why a modest move holds a likeness and a dramatic one loses it.

The prompt can also pull the clip away. If you re-describe the subject ("a blonde woman in a red dress") and your words do not exactly match the picture, the model tries to satisfy both and the result shifts. Let the picture carry the look and let the prompt carry the motion.

A prompt formula for image to video

Write the motion prompt in three short parts, in this order:

Worked example

He looks up from the laptop and gives a small, confident smile to the camera. Behind him, people walk past the café window and the late afternoon sun flickers through the trees. Slow push in, shallow depth of field.

  • Subject action: one thing the subject does. "She turns her head and smiles." "The bottle slowly rotates."
  • Environment motion: what else moves. "Hair moves in the wind." "Steam rises." "Waves roll in behind."
  • Camera: one plain camera instruction. "Slow push in." "Static camera." "Gentle pan left."
  • Optional mood: lighting or atmosphere only if it should change. "Light shifts from blue to golden."

Photo checklist before you upload

Run through this list. Fixing the photo first saves credits on takes that were never going to hold.

  • Sharp: no motion blur, not a screenshot of a screenshot, large enough to see skin and fabric detail.
  • Even light: no harsh shadow across half the face, no blown-out highlights on the product.
  • Clear subject: one main subject, not cut off at the edges of the frame.
  • Friendly angle: faces front or three-quarter, products from the side you want people to see.
  • Right shape: pick the aspect ratio closest to your photo so the frame does not crop the subject.
  • Clean background: clutter behind the subject tends to move and morph. A simpler background stays steadier.

Troubleshooting

The face changes: use a sharper front-facing photo, reduce the movement, or save the person as a character and mention their @handle so the model has a stronger reference.

The product logo warps: keep the product still and move the camera or the light instead, and add a clean product shot as an extra reference.

Nothing moves: the prompt probably describes the picture rather than an action. Add a verb for the subject and one for the environment.

Too much moves: remove extra actions and use "static camera" or "slow push in" instead of big camera moves.

Hands or text look odd: crop or choose a photo where hands are relaxed and text is large, and avoid prompts that make the subject hold or write things.

Which model to use

Seedance 2.0 takes several references at once, so it suits scenes that combine a person, a product and a place. Seedance 1.5 uses a single image as the opening frame, which is useful when the first frame must match your photo exactly; 5 seconds at 720p costs 5 credits.

Photos of real people

Only animate photos you have the rights and consent to use. Photo references are registered in BytePlus's trusted asset library before the render, and the Acceptable Use Policy applies to every generation.

Templates for this guide

Each template has its own page with a preview, the credit cost and a link to the studio.

Frequently asked questions

Why does the face change during the clip?

Usually because the photo is small, blurred or at a steep angle, or because the prompt asks for a big movement. Use a clearer photo, a smaller move, or save the person as a character.

Can I animate an image I made with AI?

Yes. Images from the image studio (2 credits each with Seedream 4.5) can go straight into image to video.

How much does it cost?

A 5-second Seedance 2.0 clip costs 15 credits at 720p or 30 at 1080p. Seedance 2.0 Fast costs 12 credits for 5 seconds at 720p. The studio shows the exact cost before you generate.

How do I animate a photo with AI?

Open the video studio, switch to image to video, upload the photo and write what should move and how the camera behaves. Keep the motion small for the most faithful result, then generate.

Should I describe the person in the photo in my prompt?

No, or only very briefly. The photo already shows what they look like. A long description that does not match the picture exactly can pull the clip away from it. Spend the words on the motion instead.

Can I animate several photos into one video?

Yes, two ways. Use up to 9 images as references in one Seedance 2.0 generation, for example a person, a product and a place, or animate each photo as its own clip and join them in an editor.

What aspect ratio should I pick?

The one closest to your photo, so nothing important is cropped. Seedance 2.0 supports 16:9, 4:3, 1:1, 3:4, 9:16, 21:9.

Try it in the studio

New accounts get 25 starter credits for images and video. Free accounts can make video on Seedance 2 Fast and Seedance 2.0 Mini at 720p, queued and rate limited. A plan adds every video model, 1080p and above, and no queue. The cost is shown before every generation.

More guides