Image to Video AI: How to Animate a Photo (Models & Costs)

You have a portrait or a product shot and want it moving for a Reel. Image-to-video AI uses that picture as the first frame, reads a sentence about what should move, and generates the frames that follow. The picture fixes the look; the prompt decides what changes.

In Spicy Studio this happens in the Video Lab, where five mainstream models can animate a still. A 5-second clip costs 35 to 195 credits, depending on model and resolution.

The model only moves what's already in the frame, so solve problems in the still.

Crop to the final shape: 9:16 for TikTok, Reels and Shorts, 16:9 for YouTube. Seedance 2.5 and Kling Omni O3 take the shape from your image; Kling 3.0 and Veo 3.1 make you pick a ratio, so crop to match.

Clear distractions: a stray object in the still shows up in every frame. The Eraser removes it for 8 credits (AI magic eraser guide).

No image yet? Generate one with Nano Banana 2.1 (12 credits at 2K) or GPT Image 2.5 (9 credits at 1K, medium quality), or follow the Playbook's first-frame workflow, free with an account.

Sound or a long clip: Seedance 2.5 makes 4–30 seconds at up to 1080p, with native audio on by default, up to 10 reference images and an optional reference video or audio track. References work only without a first or last frame (once a frame is set they're ignored), so animating your image uses frame mode. Five seconds without a reference video: 50, 110 or 195 credits at 480p, 720p or 1080p (a reference video's length is added to the billed seconds, at a lower rate; Seedance 2.5 guide).

Cheap tests: Seedance 2.0 makes 5–15 seconds at 480p or 720p for 35 or 75 credits per five seconds, with optional audio and a Camera Fixed switch (Seedance 2.0 guide).

Controlled motion toward a set ending: Kling 3.0 requires a start frame (JPEG or PNG, at least 300 px, up to 10 MB) and takes an optional end frame. 5 or 10 seconds at Standard (720p) or Pro (1080p): 60 or 80 credits for five seconds, 90 or 120 with sound (Kling 3.0 guide).

Several actions in a row: Kling Omni O3 splits a 3–15 second clip into segments of 3 seconds or more, each with its own prompt; the lengths must add up to the clip. Five seconds: 70 credits Standard, 95 Premium, or 95 and 120 with sound (Kling Omni O3 guide).

Top resolution: Veo 3.1 always makes 8 seconds, for 80, 110 or 200 credits at 720p, 1080p or 4K (Veo 3.1 guide).

Moves copied from a real video: use Kling 3.0 Motion Control in Character Swap: a character image plus a 3–30 second driving video, at 20 credits a second (Standard) or 27 (Pro), billed for at least 3 seconds (motion transfer guide).

18+ images: mature models are for users 18 or older (or the local age of majority, if higher) and for fictional or consenting adults only. See Uncensored AI Image to Video.

The model can already see the image, so spend the words on what changes, in this order:

- Camera: what it does, or “The camera stays still,” which helps keep the frame steady. - Action: one short sequence; five to ten seconds fits two or three small actions. - Finish: a short note such as “Natural movements, realistic.”

Portrait (Seedance): “The camera stays still. The woman looks up from her coffee, spots someone off-screen and smiles. Steam rises from the cup. Natural movements, realistic.”

Three beats (Kling Omni O3, 10 seconds): 4 seconds of “The baker dusts flour over the dough,” 3 of “She folds it and presses it flat,” then 3 of “She looks up at the camera and smiles.”

On Seedance 2.5, describe the sound as well (“rain on the window, a kettle clicking off”). The Seedance prompt guide has a full template.

1. Open the Video Lab, pick a model, attach the image as the first frame, paste the prompt and set length, aspect ratio, resolution and sound. Made the image in Spicy Studio? Open it in your history and click Animate: it opens in the Video Lab as the first frame, with Seedance 2.5 selected.

2. Draft five seconds at 480p on Seedance 2.0 or 2.5 and check hands, faces and edges. Then change one thing per try: a sharper first frame and less motion if the face drifts, an end frame if the clip ends in the wrong place (all five models accept one).

3. Render the keeper at 720p or 1080p. A new run won't repeat the draft exactly, so the draft tests the prompt, not the final clip. Video Upscale can then take it to 1080p, 2K or 4K for 16, 28 or 56 credits (when upscaling helps).