How to Turn a Photo into Video with AI

Lead

Turning a photo into video used to mean keyframing in an editor. With AI image-to-video, you upload a still (JPEG, PNG, or WebP), describe the motion you want, and get a short clip you can play and download. On Artgen you can run that flow through Seedance 2.0 Fast or Veo 3.1 Fast — both via unofficial third-party access — and spend credits only when you generate.

What “turn photo into video” means

You start from one image and ask the model to add time: a slow push-in, a head turn, wind in hair, clouds drifting, a product orbit. The subject and framing stay close to the photo; motion and camera notes steer the change. This is different from text-to-video, where the whole scene is invented from words alone.

Before you generate

Pick a sharp still with a clear subject and clean background when you can.
Decide the motion in one sentence (direction, speed, camera).
Choose Seedance 2.0 Fast or Veo 3.1 Fast in the model list and confirm the unofficial-access note.
Set audio on or off if the control is shown for that model.
Generate, play in the browser, then download when you like the take.

Prompt tips that keep the photo intact

Name what must stay fixed: person, outfit, product, logo-free background.
Add one or two motion verbs; avoid stacking five style adjectives.
Prefer locked-off or slow camera moves over hard cuts the model cannot invent from a single frame.
If the face drifts, shorten the motion ask and regenerate.

Example motion note (portrait):
Keep the same person and framing. Subtle head turn toward camera, light breeze in hair, natural blink, locked-off camera

Example motion note (product):
Keep the same product and pedestal. Slow orbital camera move, soft rim light, clean background

Seedance 2.0 Fast for photo → video

Seedance 2.0 Fast (ByteDance) accepts text and image inputs on Artgen via unofficial third-party access. Upload your still, add motion notes, toggle audio if you need sound, then generate. Artgen does not own the Seedance trademark; generation uses credits.

CTA: Open Seedance 2.0 Fast → /en/models/seedance-2-0-fast

Veo 3.1 Fast for photo → video

Veo 3.1 Fast (Google) is the other live Artgen video model for the same still-to-clip path. Same idea: image in, motion notes, optional audio, play and download. Access here is unofficial third-party; Artgen does not claim to be Google.

CTA: Open Veo 3.1 Fast → /en/models/veo-3-1-fast

Which model should you try first?

Try both when quality matters. Seedance 2.0 Fast and Veo 3.1 Fast both do image-to-video on Artgen; pick from the live model list rather than third-party scorecards. If one take softens the face or drifts the product, switch models or tighten the motion note and run again.

Common mistakes

Asking for a new scene the photo does not contain
Writing a novel-length prompt that fights the still
Skipping the unofficial-access note and expecting vendor-official branding
Targeting “image to video AI” tool-head SEO on this blog — that commercial tool URL is backlog; this page is the how-to wedge

FAQ

Q: Can I turn any photo into video?
A: Most clear JPEG/PNG/WebP stills work. Busy crops, heavy compression, or tiny faces need a simpler motion ask.

Q: Do I need an editor after generating?
A: You can play and download the clip from Artgen. Trim or grade elsewhere if your pipeline needs it.

Q: Is this the same as the Artgen image-to-video tool page?
A: This is a how-to blog for “turn photo into video.” Live model entry points today are Seedance 2.0 Fast and Veo 3.1 Fast. A dedicated EN image-to-video tool URL remains backlog.

Q: Are Seedance and Veo official on Artgen?
A: No. Access is unofficial third-party. Artgen does not own those trademarks.

Q: Does generation use credits?
A: Yes. Credits apply when you run a job.

Internal links

/en/models/seedance-2-0-fast
/en/models/veo-3-1-fast
Related later: /en/blog/seedance-2-0-tutorial (A2)