AI product video from image: turning one product photo into a video ad

Which free tools turn a product photo into a video?

Three ad managers animate a product photo at no listed price, and each keeps the clip at home. The trade is one platform.

Meta’s route is Add animation, which its help page (Wayback capture of 21 September 2026) describes as turning “eligible images” into “short, engaging videos” with panning, zooming and rotating. It is a single-image enhancement, switched on in the ad’s set-up. Its rights row records the grant as conditional.

Amazon’s generator needs an Amazon Ads account, and its guide says it is available in the US marketplace in English (read 25 September 2026). The clip serves Amazon ads and Product Detail Pages (rights row).

TikTok Symphony makes image-to-video clips inside Ads Manager. Nine of 18 user sources say it is free with a daily credit allowance (YouTube, YouTube); TikTok’s own price is unverified (rights row).

Off-platform, Pippit’s free plan watermarks output (rights row). Runway’s Free plan allows ads on its one-time credits (rights row). Google’s free Flow credits, 50 a day, are unclear for ads: Google’s Terms of Service (30 July 2026, “Your content”) say only that it “won’t claim ownership” of generated content. The broader free list is on the product video hub.

Pippit AI Review: Is It Worth Using? (2026)How to Easy

What one clip costs on each paid route

A 5-second product clip runs from $0.30 on Luma Plus at 720p to $1.47 on Claid.ai Pro. Both are list prices with every credit used, checked 2 October 2026.

• Luma Plus: $0.24 per second of video at 1080p, a quarter of that at 720p. Commercial use covers only outputs made while subscribed (price row).
• Kling: $0.18 a second at 1080p, for a member buying top-ups; 1080p is member-only. Non-members may not use outputs commercially (price row).
• Runway Standard: $0.29 per second of video on Gen-4.5 (price row).
• Google Veo on Google AI Pro: $0.40 for a Veo 3.1 Fast clip of 4 to 8 seconds from monthly credits (price row).
• Claid.ai Pro: 60 credits per 5 seconds of video (price row).

Higgsfield resells several of these models. Seedance is ByteDance’s own model, sold through Dreamina, CapCut and resellers. Our database records no per-second rate for either, so neither gets a figure here. The cost per AI ad calculator works out a clip from your own plan.

Product tools add the store link, avatars or a finished edit, and price by clip or plan instead of by the second. Topview prices each 4 or 5-second clip by model, and Photoroom starts video at its Max plan.

How to prepare the product photo

User reports name labels and logos as the parts that break, so prepare those first.

1. Crop to the output ratio before upload. Meta’s visual touch-ups crop and expand a single image with generated pixels when switched on, so a crop you choose keeps control of the frame.
2. Upload the largest file the tool takes: Claid.ai accepts 10 MB on Essentials and 20 MB on Pro.
3. Use one product per frame on a plain background, label facing the camera, with no other text in the shot.
4. Set an end frame where the tool offers one. Kling’s 3.0 guide lists start and end frames to video, which pins the last frame to your real product.
5. Keep clips short. Kling 3.0 runs 3 to 15 seconds and Veo 3.1 Fast 4 to 8, so plan one camera move per clip.

How to write the motion prompt

Kling’s 3.0 guide builds its example prompts from the subject, the action, the camera move and the lighting. With a photo as input, describe the motion only, since the photo already holds the look.

Patterns to adapt:

• Bottle: “Slow push-in on the bottle, label facing camera, condensation forming, soft window light from the left, product stays still.”
• Garment: “Fabric moves in a light breeze, camera orbits a quarter turn at chest height, daylight, garment shape and print unchanged.”
• Gadget: “Camera tilts down to the device on a desk, screen lights up, shallow depth of field, cool studio light, logo unchanged.”

Name what must stay fixed at the end of the prompt (“label unchanged”, “logo unchanged”). Prompt patterns for Google’s model: Veo prompts for ads.

Where products break, by user report

Labels and logos are where users report damage. 6 of 76 Photoroom sources say generative features alter the product, distorted labels and changed clothing included (Trustpilot, App Store).

2 of 24 Flair.ai sources report wrong logos, patterns or product context, such as lime slices added to a product that is no cocktail (Trustpilot, Shotkit). On Meta, 10 of 35 sources report distorted generated backgrounds and expansions (Meta Advantage+ creative).

Check the last frame against the real product before export.

Labels and disclosure for the finished clip

Meta labels ads made with its own generative tools with “AI info”. TikTok Symphony adds an AI-generated label to every export, and TikTok’s ads policy requires disclosure of fully AI-made media. Google Ads has an AI label setting. The rules by platform: AI disclosure in advertising, Meta’s AI disclosure rules and TikTok’s AI label.