Turn an image into video

Upload a picture, say what should move, and get a video back. Twenty three image to video models in one place, from 3 credits a second, no watermark on any plan, and you own the result.

Turn an image into video

Real image to video output

One still image in, one clip out, on five of the models Percify runs. These are each model's own preview, shown so you can see the difference between them before you spend anything.

Pixverse V6, 5s
Seedance 2.0 Mini, 5s
Kling V3 Turbo, 5s
Hailuo 2.3, 10s
Wan 2.7, 5s

How it works

  1. Step 1

    Upload the image

    A photo, a product shot, a drawing or a render. Crop it first: the shape of the image becomes the shape of the video.

  2. Step 2

    Say what moves

    One line is enough. Name the subject's movement, the camera's movement, or both, and choose how long the take runs.

  3. Step 3

    Test cheap, finish strong

    Try the shot at 3 credits a second, then send the same image to Veo 3.1 or Sora 2 for the take you keep. Download the MP4, no watermark.

What image to video costs

Every model is priced per second of video at its own rate, and you pay for what you generate rather than a seat. The five second column is what a five second clip costs at that model's base resolution. Runs that fail are refunded.

ModelPriceBase5 s clipAboutGood for
Pixverse V63 credits a second360p15 credits$0.25 to $0.38The cheapest way to see if a shot works
Seedance V1.5 Pro25 credits per 5 seconds480p25 credits$0.41 to $0.63Steady, natural motion at a low price
Hailuo 2.334 credits per 6 secondsstandard29 credits$0.48 to $0.73Expressive movement and camera work
Seedance 2.0 Mini30 credits per 5 seconds480p30 credits$0.50 to $0.75A newer model that holds detail
LTX-2 Pro36 credits per 6 secondsstandard36 credits$0.59 to $0.90Longer takes without a jump in price
Wan 2.750 credits per 5 seconds720p50 credits$0.83 to $1.25720p out of the box
Luma Ray 3.250 credits per 5 seconds540p50 credits$0.83 to $1.25Cinematic camera moves
Kling V3 Turbo Standard34 credits per 3 secondsstandard57 credits$0.94 to $1.43People and faces that stay themselves
Veo 3.1 Fast60 credits per 4 secondsup to 4K75 credits$1.24 to $1.88Google Veo, and 4K costs the same as 720p
Sora 270 credits per 4 secondsstandard88 credits$1.45 to $2.20OpenAI Sora 2, for the hardest shots

Thirteen more image to video models are live beyond this table, twenty three in all. Credits work out at about 2 cents each depending on the plan: 20 credits is $0.33 on Starter, $0.42 on Creator and $0.50 on a credit pack. Higher resolutions cost a multiple on some models and nothing extra on others, and the playground quotes the exact cost before you run anything. Plan prices are on the pricing page. Checked 22 September 2026.

Which model to use

  • Testing an idea. Pixverse V6 at 3 credits a second. Five seconds costs about 25 cents, so you can try the same shot six ways for the price of one premium take.
  • A person in the frame. Kling V3 Turbo keeps a face recognisable through the movement, which is where cheaper models drift.
  • Lively motion. Hailuo 2.3 moves the subject and the camera more than most, for about 48 to 73 cents a five second take.
  • The best take. Veo 3.1 Fast and Sora 2. Veo also generates its own audio, and its 4K costs the same as its 720p.
  • 720p without a multiplier. Wan 2.7 starts at 720p rather than charging a multiple to get there.

Every model runs on the same credits and the same API key, so changing model is a parameter, not a migration. API pricing.

Make it move, or make it talk

These models animate a picture. They do not make the person in it speak: lip sync is a different family of models, driven by an audio track, where the mouth and jaw follow the voice. If what you want is a photo that says something, start at make a photo talk, which runs from 2 credits a second. If you want the photo to move, stay here.

Watermarks, rights and consent

  • No watermark on any plan, including the free one.
  • You own the content you generate, under Percify's terms. It is not exclusive: similar inputs can give similar output for someone else.
  • Use your own image, or one you have the rights to. A photo of a person needs that person's agreement.

Details, and how to check any other tool for hidden watermarks: AI avatar watermarks and commercial use.

Make a photo talkClone yourselfRemake a shortEvery model we runAPI pricing

Put your picture in motion

One image in, a video out. From 3 credits a second across 23 models, no watermark, and failed runs refunded.

Turn an image into video

Frequently asked questions

Everything you need to know before you start.

It is a model that takes a still image and generates a video from it: the subject, the background and the camera start moving while the picture stays recognisable. You describe the motion you want in a prompt, pick how long the clip runs, and get an MP4 back. It is different from text to video, which invents the picture as well as the motion.

Upload the image, write a short line describing the movement you want, choose the length and resolution, and generate. Percify runs 23 image to video models, so the same photo can be sent to a 3 credit a second model to test the idea and then to Veo 3.1 or Sora 2 for the final take.

There is no single best one, which is why Percify carries 23 of them: Veo 3.1 Fast and Sora 2 are the strongest and the dearest, Kling V3 Turbo holds faces best, Hailuo 2.3 gives the liveliest motion, and Pixverse V6 at 3 credits a second is the one to test an idea on. It is our product, so compare it on the same points against the tools you are considering.

From 3 credits a second. A five second clip is 15 credits on Pixverse V6, 25 on Seedance V1.5 Pro, 50 on Wan 2.7 and 88 on Sora 2. Credits work out at about 2 cents each depending on the plan, so five seconds runs from roughly 25 cents to about $2.20. Runs that fail are refunded.

Percify has a free plan and no watermark on it, and runs that fail are refunded, but generating video costs credits on every plan because every run costs us provider time. The honest version is that you can start without paying and see real output before you decide. What each plan includes is on the pricing page.

Two different jobs. To make a person in a photo speak, use a lip sync model, where the mouth and jaw follow an audio track: that is the talking photo page. To make a photo move without speech, use the image to video models here. Veo 3.1 also generates its own audio track.

No. Image to video output is clean on every plan, including the free one, and there is no paid upgrade to remove a mark because there is no mark. The one Percify tool that marks its output is Replicate a short, whose replicas carry a small corner mark as an AI content disclosure.

Yes. Percify’s terms say you own the AI generated content you create, subject to the terms. Output is not exclusive, and you need the rights to the image you upload.

It depends on the model. Wan 2.7 starts at 720p, Veo 3.1 Fast goes to 4K at the same price as 720p, and models like Seedance and Luma Ray charge a multiplier for higher resolutions: Seedance 2.0 Mini is 2x at 720p and 5x at 1080p over its 480p price. The playground quotes the exact cost before you run anything.

Most of these models generate in takes of a few seconds, which is what a feed post is made of. Length is priced proportionally, so a 10 second clip on Pixverse V6 is 30 credits and the same clip on Wan 2.7 is 100. For something longer, generate the takes and join them.

A sharp image with a clear subject and room around it. Motion is invented from what is visible, so a cropped or blurred subject gives the model less to work with. The aspect ratio of your image decides the shape of the video, so crop it to the shape you want before you generate.

Yes. Every model on this page runs through the same API with one key and one credit balance, so switching from Pixverse to Veo is a parameter, not a new integration. Details are on the API pricing page and in the docs.