Photo to video: turn any photo into a TikTok-ready clip

You have a photo. You want a video. The gap between those two things is one browser tab — but the quality of what comes out is decided before you ever press generate, by the photo you picked and the mode you picked it for. This guide covers both halves.

What "photo to video" means here

The pipeline is image-to-video generation: a diffusion model reconstructs a clean image from noise in a compressed latent space (Rombach et al., 2022), and a temporal layer extends that from stills to motion so consecutive frames stay consistent (Blattmann et al., 2023). In practice: your photo supplies the subject; a preset script supplies the motion, scene and camera work. You never write a prompt.

What photo works best

The photo is the biggest quality lever you control. The short version of the rules:

• Face the camera. Preset motion scripts assume a starting pose; side profiles get reconstructed and lose detail.
• Chest-up framing. Keeps the face large enough to stay sharp after social-app compression.
• Even light, plain background. The model separates you from the background to re-render it — clutter and darkness leak artifacts into hair edges.
• One person per slot. Group photos swap faces.
• Rights to the photo. Borrowed faces get reported.

The full version with examples lives in the photo guide.

One-photo vs two-photo templates

Most templates are one-photo: your face goes in, the template invents everyone else in the scene. Two-photo templates exist when the clip has a second cast member the user supplies — the other half of a duet, a partner, a friend, or a second pet. Examples: the Rumpelstiltskin barn dance (you + a wildcard), the zombie love story (you + your partner, or two cats), the kiss-cam, the rain dance.

The mechanic is slots, not face detection: one clear photo per slot. A photo of two people in one slot will not split into two subjects.

Beyond video: GIF and image modes

Not every asset needs to be a video. Each template can offer up to three modes with prices shown separately:

• Video — the full scripted clip, silent vertical MP4, roughly 8 to 26 seconds depending on the trend.
• GIF — a short loop for reactions and comments. The kiss-cam, pet-dance and street-groove templates all have GIF versions that cost fewer credits and render faster.
• Image — a restyled still. The image generators turn your photo into a poster-paint, pixel-diorama, comic-ink or blueprint version of itself, useful for covers and profile art.

The studio shows the price for whichever mode you select before you generate.

Export specs

The output is built for where it is going:

• 9:16 vertical (16:9 selectable on some templates)
• Silent MP4 — sound gets added in TikTok from the licensed library
• No watermark; commercial use OK with rights to the photo
• Resolution selectable before generating; render time about a minute

Step by step

1. Pick your template in the AI video studio — trend board or category pages make it fast.
2. Upload the photo (or photos, for two-slot templates), following the rules above.
3. Choose the mode — video, GIF or image — and check the price shown before generating.
4. Generate and download. Failed renders are refunded automatically.
5. Post and add the sound in TikTok.

FAQ

Can I use a photo of two people? Only in a two-slot template, and only as one photo per slot. A single group photo will not split into two subjects.

What is the best photo of a pet? Daylight, face visible, pet filling the frame. Fur detail collapses in dim light.

Can I turn a photo into a GIF instead of a video? Yes — several templates have GIF modes priced separately, shown before you generate.

Do I write a prompt? No. The template carries the scene and motion; you carry the photo.

Related reading

Start in the AI video studio, study the photo guide before uploading, or restyle a still in the AI image studio.