11 image-to-video models · one studio

Animate any photo with Seedance, Kling and VeoImage to Video AI

Drop in a product shot, a portrait or a frame you designed, describe the motion, and get a clip back, with sound on the models that make it. Every leading video model runs on one credit balance.

Model · Seedance 2.5
Try

You add your photo as the start frame in the video studio.

  • Clips from 75 credits
  • Up to 30s per clip
  • Up to 1080p
  • Start and end frames
  • Commercial use
Made in Perfect Photo AI
Made with Kling 3.0 Pro
Made in Perfect Photo AI

How to turn an image into a video

No editing timeline and no keyframes. The photo sets the scene; your prompt sets what happens next.

  1. 01

    Add your photo

    Upload a JPG, PNG or WebP image, or pick one you made in the image studio. It becomes the first frame of the clip.

  2. 02

    Describe the motion

    Say what moves and how the camera moves. The model already sees the photo, so you don't need to describe what's in it.

  3. 03

    Pick a model and length

    Choose the model, resolution and clip length, and add an end frame if you want the shot to land somewhere exact. The credit cost shows before you generate.

  4. 04

    Generate and compare

    Keep the take you like, tweak the prompt, or run the same photo on a second model and keep the better one.

Every image-to-video model, compared

All of these take your photo as the start frame and share one credit balance. Lengths, resolutions and prices come straight from the studio, so they match what you'll see when you generate.

ModelLengthResolutionSoundEnd frameFromOpen in studio
Veo 3.14s–8s720p, 1080pGeneratedYes, 8s clips300 crTry
Veo 3.1 Fast4s–8s720pGeneratedYes, 8s clips250 crTry
Seedance 2.55s–30s480p, 720p, 1080pGeneratedYes225 crTry
Seedance 2.05s–10s480p, 720p, 1080pGeneratedYes150 crTry
MiniMax H3 Max5s–15s480p, 768p, 1080pYour trackYes, or on its own75 crTry
MiniMax H3 Max Turbo5s–15s480p, 768p, 1080pSilentYes, or on its own120 crTry
Wan 3.05s–30s480p, 720p, 1080pGeneratedYes150 crTry
Wan 2.75s–15s720p, 1080pGeneratedYes225 crTry
Kling 3.0 Pro5s–15s720pSilentYes250 crTry
Kling O3 Pro5s–15s1080pGeneratedYes300 crTry
Happy Horse5s–10s720p, 1080pGenerated—250 crTry

“From” is the shortest clip at the lowest resolution. Longer and sharper clips cost more; the studio shows the price before you generate.

Which model to pick

  1. Longest clips

    Seedance 2.5 and Wan 3.0 run up to 30 seconds in one generation, long enough for a full ad beat.

  2. Lowest cost per take

    MiniMax H3 Max starts at 75 credits for a 5s 480p clip. Test the motion there, then re-run your favorite on a premium model.

  3. Sound in the same pass

    Veo 3.1, Veo 3.1 Fast, Seedance 2.5, Seedance 2.0, Wan 3.0, Wan 2.7, Kling O3 Pro and Happy Horse generate audio with the picture, so the clip is usable without a separate sound step.

  4. Movement-heavy shots

    Dance, sport and fast camera moves: Kling is usually the stronger pick when the action is the point of the shot.

  5. Polished, on-brief framing

    Veo 3.1 is good at doing what the prompt says, with clean lighting. It makes 4s or 8s clips in 16:9 or 9:16.

  6. Exact start and end

    Every model here takes a start frame; MiniMax H3 Max and MiniMax H3 Max Turbo can also work from an end frame alone. Most add an optional end frame for controlled transitions.

What you can make from a single photo

A few of the jobs image to video gets used for most, and where to start for each.

Sample clips and images made in Perfect Photo AI.

Image to video prompts that work

An image-to-video prompt is shorter than a text-to-video one. The picture is already there; you're directing the next few seconds.

  1. Describe the motion, not the picture

    Your photo already sets the subject, colors and framing. Spend the words on what moves, how fast, and in which direction.

  2. One camera move per clip

    Push in, pull back, orbit, pan, tilt or handheld. Stacking three moves into five seconds usually looks muddled.

  3. Keep the action small

    "She looks up and smiles" beats a whole storyline. On longer clips (up to 30s) you can add a second beat.

  4. Say what should stay put

    Phrases like "background stays still" or "the label doesn't move" help products, logos and faces hold their shape.

  5. Add sound cues on audio models

    Name the ambience, effects or a short line of dialogue. Models that don't make sound simply ignore it.

  6. Use an end frame to land the shot

    When the last frame matters, like a product turn or a before-and-after, give the model both images and let it fill the motion in between.

Example prompts. Each opens in the studio with Seedance 2.5 selected; add your photo as the start frame.

Questions, answered

New accounts get 10 free credits. Image-to-video clips are paid in credits by model, resolution and length, starting at 75 credits for a 5s 480p clip on MiniMax H3 Max. The free credits are enough to make a start frame in the image studio (images from 15 credits), but a video clip needs a credit pack. You see the exact cost before you generate.

From 4 to 30 seconds per clip, depending on the model. Seedance 2.5 and Wan 3.0 go up to 30 seconds. To go longer, use "Continue from last frame" on a finished clip: its final frame becomes the start frame of the next one.

Veo 3.1, Veo 3.1 Fast, Seedance 2.5, Seedance 2.0, Wan 3.0, Wan 2.7, Kling O3 Pro and Happy Horse generate audio with the video: ambience, effects and speech you describe in the prompt. MiniMax H3 Max Turbo and Kling 3.0 Pro make silent clips you can score yourself. MiniMax H3 Max can use your own audio file as the soundtrack.

JPG, PNG and WebP. You can upload a photo or pick any image you've generated in the studio. Seedance 2.5, MiniMax H3 Max, MiniMax H3 Max Turbo, Wan 3.0 and Wan 2.7 follow your photo's aspect ratio; on the others you pick the ratio in the studio, so crop your photo to the frame you want first.

The start frame is the photo your clip begins on. An end frame is a second image the clip should finish on, and the model fills in the motion between the two. Every model here except Happy Horse accepts an end frame (on Veo 3.1 and Veo 3.1 Fast, only for 8s clips). MiniMax H3 Max and MiniMax H3 Max Turbo can also work from an end frame alone.

For a finished clip with sound, start with Seedance. For dance, sport and fast camera moves, try Kling. For a polished shot that follows the prompt closely, use Veo 3.1 (4s or 8s clips in 16:9 or 9:16). For many cheap takes while you explore, MiniMax H3 Max costs the least. All of them share one credit balance, so you can run the same photo on two and keep the better take.

The clip starts on your exact image, so the composition, colors, product and people come from it. What happens after the first frame is generated, and details like faces, hands and small text can drift during big or fast motion. Small movements and a steady camera keep the result closest to the original.

Yes. H3 Max Lip Sync takes a clear face photo and a voice or vocal track and makes the portrait speak or sing along, for up to 60 seconds. The clip runs as long as the audio and is billed by its length. It doesn't use a prompt.

Yes. You need a free account to generate, so your clips, uploads and credits are saved in your library.

Yes. Videos you generate can be used for personal and commercial work, including ads and client projects, under our Terms of Service. Only animate photos you own or have permission to use.

Keep exploring

Got a photo? Give it a few seconds of motion.

Open the video studio, add your image as the start frame and describe the move.

Animate a photo
Image to Video AI: Animate Any Photo | Perfect Photo AI