Published 2026-09-22 · 6 min read
If you have one decent photo of a product, you have enough to make a video ad. This is a straightforward, honest breakdown of what happens at each step, what makes a photo work well, and what to check before you export.
Start with the photo, not the idea
It's tempting to plan out a whole campaign before you've uploaded anything. In practice, the photo you have determines a lot of what's possible, so start there. A single product photo with a plain or simple background, decent lighting, and the product clearly visible is enough for most AI video tools to work from.
Avoid photos where the product is one of several items in frame, where there's a distracting background, or where the lighting is very dark or very yellow. These aren't disqualifying, but they will show up in the generated scenes, since the AI builds the video starting from what's actually in the image.
Let the analysis inform the concept
Once the photo is uploaded, the AI identifies what it's looking at: product category, colours, materials, and visual style. This isn't just metadata — it's what the script generation step uses to write a relevant hook and value proposition instead of a generic one.
This is a good moment to check the analysis is roughly right. If a leather bag gets analysed as a fabric tote, the resulting script may lean on the wrong selling points. It only takes a second to glance at this and correct course before generating a full script around the wrong assumption.
From script to scenes
The script is broken into scenes — typically a hook, one or two scenes demonstrating value, and a closing scene with a call to action. Each scene becomes a short AI-generated video clip, using your product photo as the visual starting point rather than a stock clip or an unrelated AI-generated product.
- You can usually edit, reorder, duplicate or delete individual scenes
- Shorter ads (around 8–16 seconds) tend to perform better for cold audiences on TikTok and Reels
- Longer ads (24–32 seconds) work better once someone already recognises the brand or product
Voice, captions and export
Once the scenes are approved, add a voiceover. Match the tone to how you'd actually talk about the product — energetic for impulse-buy items, calmer and more informative for considered purchases. Captions should be generated automatically and timed to the voiceover; most social video is watched with the sound off, so captions aren't optional if you want the message to land.
Export at the aspect ratio your platform expects: vertical 9:16 for TikTok and Reels, square 1:1 for feed posts, and 16:9 if you're running it as a YouTube pre-roll or a display ad. Getting this wrong is a common, easily avoidable mistake — a video cropped awkwardly for its placement will underperform regardless of how good the content is.