With a single product photo and a few generative artificial intelligence tools, it is possible to create about 10-second product ads for TikTok and YouTube at computing costs of around $1, or about 1,500 won.

According to blockchain outlet Decrypt on July 12, combining ChatGPT’s GPT Image 2, Google Flow’s Gemini Omni and CapCut makes it possible to produce ad videos without any separate filming. The process previously required filming crews, lighting and multiple retakes, but can now be replaced with a product photo and AI tools.

The production process starts with preparing a product image. After uploading a product photo with the background and watermark removed to ChatGPT, users create a scene in which a virtual model wears or holds the product. The prompt includes conditions to keep the product’s shape, proportions, color, texture and fit intact and not to change the product.

Users then use ChatGPT to create a 10-second script. The instruction is to grab viewers’ attention within the first 2 seconds, mention a low price and end with a line asking them to tap the shopping cart below. Creating the script in JSON format for use in Google Flow can reflect screen progression, camera movement and dialogue more accurately. If the timeline ends at 8 seconds, the model may repeat actions to fill the remaining time, so the output needs to be checked.

Gemini Omni handles video generation. The model can make videos of up to 10 seconds with audio and also allows reference images to be attached separately. Under the developer API pricing, the cost is $0.10 per second, or about $1 per clip. All videos include an invisible SynthID watermark that identifies them as AI-generated.

Final editing is done in CapCut. Users cut repeated actions or unnecessary phrases and add subtitles. Word-by-word animated subtitles commonly used on TikTok are a paid feature, but manual subtitle input is free. If the first result is not satisfactory, users can remake it several times, and if it can be fixed in editing, there is no need to use additional credits.

AI ads, however, come with disclosure obligations. TikTok and YouTube require labels or AI-use disclosures for realistic AI-generated images, voices and videos, and they also require separate disclosure of whether content is commercial or paid promotion. X, formerly Twitter, bans synthetic or manipulated media but does not explicitly ban AI images, videos and voices used for advertising.

As AI tools replace the entire process from product shoots to scripts, video production and editing, an environment is emerging in which ad videos can be tested quickly with limited capital. Failure to comply with platform-specific AI content disclosure obligations can lead to sanctions, so related policies need to be checked before running ads.