Text to video turns a written scene into a short clip. You describe what should appear. Nothing is filmed. If you already have a still you want to animate, use image to video instead, so the subject stays the one in the photograph.
How to write the scene and generate
- Open AI text to video.
- Describe the shot. Subject, setting, light, camera move. "A red handbag on a table, soft side light, slow push-in" is more useful than "cinematic luxury ad."
- Pick engine, ratio and length. Grok caps at 6 seconds and bills 6 credits a second. Kling bills 3. Seedance engines run 34 to 95 depending on settings, and can go to 30 seconds. The studio shows the estimate next to the picker.
- Generate per placement. 9:16 for Reels, TikTok and Stories. Square or landscape for feed and catalogue. Cropping one master loses the framing the format was composed for.
What to check
Colour and label text are generated, not taken from a product photo. If the listing and the clip have to agree, start from a real still in image to video, or from a presenter in the UGC studio. Video needs a paid credit pack. Disclose AI where required.
Frequently asked questions
How long can the video be?
It depends on the engine. Grok caps at 6 seconds. Seedance can run to 30. All of them cover lengths Reels, TikTok and Shorts accept.
How much does it cost?
Grok is 6 credits a second. Kling is 3. Seedance engines run 34 to 95 depending on duration and resolution. The studio shows the estimate before you generate.
Can I describe a specific product?
Yes, but the product is generated from the description. Colour and label text can drift. For a real SKU, animate a photograph in image to video.
Should I generate one master and crop it?
No. Generate per placement. A crop loses the framing the format was composed for.