Image to Video: Turn a Still Into a Talking Clip

Image to video is motion first, mouth second

Image to video AI turns a photograph or illustration into a moving clip. The model holds composition and identity, then adds camera and subject motion you describe. When the face stays readable, the next click is a lip sync pass. When the shot is a product flyover, you skip speech and post the motion as is.

This is not Talking Photo. Talking Photo animates a still's mouth and invents a modest amount of life so the JPEG does not look laminated. Image to video is for a push-in, parallax, hair in wind, a slow orbit — motion you asked for. Category tools (Magic Hour, Wireflow, Dzine) often blur those jobs. LipSyncing AI keeps them as separate pages because they spend credits as separate generates.

When to use image to video vs talking photo, motion control, and product video

Use Image to Video when you like the still and need it to move before anyone talks. Use Talking Photo when the job is speech and you do not care about a camera move. Use Motion Control when you have a reference performance to copy onto a character. Use Product Video Generator when the still is a pack shot and the audience is a PDP.

Use Text to Video when you do not have a still yet. Use Frames to Video when you already know the first and last frame. Use Video Lip Sync only after you have footage — including footage this tool just made. Two tools, two bills, if you animate and then talk.

How to turn a still into a talking clip

Plan the mouth before you prompt a dance. A cinematic profile looks expensive and then fails visemes.

If the motion pass hid the mouth — hair, a turn, a super-wide camera — do not lip-sync that export. Go back to the still and use Talking Photo, or regenerate motion with a slower push-in.

• Upload a still with a large, front-facing face if speech is the goal.
• Prompt a gentle camera move. Slow push-in beats a whip-pan around the head.
• Generate the motion clip. Preview locally. Check that identity held and the mouth is still visible.
• Open Talking Photo or Video Lip Sync on that result (or the original still if the motion added nothing useful).
• Add a script or audio and generate the mouth pass.
• Cut the two exports in your editor if you need b-roll plus a talking take; do not expect one button to do both jobs for free.

Quality checklist for motion that can still talk

If you plan to lip sync, the moving clip must still be a talking-head source: mouth visible, face large, light even.

Ask whether the camera move will be visible on a phone in a feed. If not, you paid 150 credits for wind in the hair that nobody sees. Skip straight to Talking Photo next time.

• Identity matches the still. No cousin jaw.
• Hands, if any, are not extra-fingered blobs.
• Label type on products stays type on a slow move.
• The mouth is not occluded by hair, a turn, or a super-wide camera.
• Duration matches the later line. A 6-second orbit cannot carry a 40-second script without a freeze or a second generate.
• You know whether this export is B-roll or a talking source. Mixing those intents in one prompt makes both worse.

Credits: two generates if you talk

Each generate uses the same meter. One credit = $0.01 on a plan. First 30 seconds = 150 credits ($1.50). Extra second = 5 credits ($0.05). 45s = 225 credits. 60s = 300 credits. An 8-second motion pass is still 150 credits. A 20-second talking pass on top is another 150 credits. Budget two first-blocks when you chain the tools.

Do not generate 60 seconds of motion “for safety.” You will pay 300 credits for a loop you will cut to six. Preview locally. No free plan. Packs are $0.08 per credit and never expire. Yearly is 30% off, no auto-renew. Subscription credits spend first, then packs. Unused plan credits roll over while the plan is active and cancel if you cancel.

Rights on stills you animate

Animating a still is still using the likeness and the photograph. Stock licenses do not always include AI video. Check. Faces you do not own stay off-limits even if you only “added wind.”

Some stock licenses forbid AI video or competitive ads. If the still came from a marketplace, read the extra clauses before you animate and before you lip-sync.

• Own the photo, generate it here, or license it for this use.
• Product stills: you need the right to that pack photography.
• Do not animate a competitor's hero image.
• Audio for the later lip-sync pass has its own rights.
• Commercial output requires a paid plan plus those rights.

Common image-to-video failures

Boiling backgrounds, identity drift, and motion that hides the mouth. Name the failure; change the prompt or the crop.

Chaining motion and speech without previewing the motion first is how you lip-sync a cousin. Always watch the motion export muted before you spend the mouth pass.

• Talking Photo would have been enough: you paid for a camera move nobody sees on a phone.
• Profile cinematic shot: visemes will fail. Keep the face front-on.
• Dance prompt on a headshot: use Motion Control with a full-body still.
• Melted product type: slower orbit, or Product Video Generator.
• One 60-second job for motion plus speech: split the pipeline.
• Stolen stills: a rights failure that looks like a successful render.