About the Caught at the Window
Caught at the Window is an AI commercial generator for your brand, built on the viral window ad. The camera drifts across the street toward one warm, lit apartment window. Inside, someone is enjoying their food, notices the camera, gives it an annoyed caught-in-the-act look, puts the food down and pulls the curtains shut.
Upload a photo of your food or product, or pick pasta, a burger, soup or iced coffee, or describe your own. Add a photo of a person who agreed to appear, or pick one of the built-in adult characters. The way they eat follows the food: pasta twirled on a fork, a burger held and bitten, soup eaten with a spoon and a drink sipped, and you can change it. The ad runs 12 seconds, vertical 9:16 at 720p, with music.
How to make a Caught at the Window ad
- Choose the food or product. Upload a sharp photo of it, or pick a preset or describe it. With nothing chosen, it's a plate of pasta. JPG, PNG or WEBP, up to 10 MB.
- Choose who is at the window. Upload a clear photo of one adult who agreed to appear, or pick a built-in character.
- Customize (optional). Set how they eat or drink it, and turn the music off if you only want the scene's own sound.
- Generate and download. Check the credit cost and generate. The video is made first, then the soundtrack is added. Download the MP4 from your Library.
What happens in the 12 seconds
- 0–3s: the camera approaches the lit window from outside.
- 3–7s: the person eats or drinks, with the action that fits the food.
- 7–9s: they notice the camera and give it an annoyed look.
- 9–10.5s: they put the food down and close the curtains.
- 10.5–12s: a hold on the closed curtains.
Food, products and the eating action
- Presets: pasta (twirled on a fork, the default), burger (held and bitten), soup (eaten with a spoon) and iced coffee (sipped).
- Your own food or product: describe it, or upload a photo. Packaging is kept as closely as the model allows, but small printed text can vary.
- Your choice: under Customize you can set the action yourself or describe it in your own words.
Music and sound
Music is on by default. One mastered soundtrack plays through the approach and the eating, drops when the person notices the camera, carries the room and curtain sounds, and fades out at the 12-second mark. Its sound effects follow the scene's usual timing, so they may not land on every bite. Turn the music off under Customize to keep the scene's own generated sound instead.
Likeness and labels
Keeping a person's face and look, and a product's packaging, are generation goals, not guarantees. Use clear, well-lit photos, only of people who agreed to appear, and check the result before you publish it.
Frequently asked questions
Do I need to upload anything?
No. Without uploads, a built-in adult character eats a plate of pasta. Add a food or product photo and a person photo as you like.
Will the person look like my photo?
The video aims to keep the uploaded person's face, hair, age and look, but an exact match isn't guaranteed. A clear, front-facing photo works best.
Does the eating change with the food?
Yes. Pasta is twirled on a fork, a burger held and bitten, soup eaten with a spoon and a drink sipped. You can pick or describe the action yourself under Customize.
Can I turn the music off?
Yes, under Customize. The video then keeps the sound generated with the scene.
Can I add my logo?
Not yet. The video doesn't include a logo or closing text.
How long is the video?
12 seconds, vertical 9:16 at 720p.
How many credits does it cost?
1,424 credits for the video with the soundtrack.
Discover more templates in the trends gallery or explore AI video effects.
Independent AI template. Trendai.video is not affiliated with or endorsed by the people or brands in the ads the scene is modelled on.