AI video prompts for Alibaba Wan 3.0. Reference images supported, up to 30s at 1080p. Build brand promos, product ads and short films from a single prompt.
2026.08.24
Make “Wan 3.0” exactly the way you wantCreate your own “Wan 3.0”
Just ask in plain language — Carat generates optimized images and videos for you.
Start now with 100 free credits daily
Wan 3.0 is a video generation model in Alibaba's Wan series. It supports three ways of working: generating video from text alone, using a single image as the opening frame, and supplying reference images so a person or object keeps the same appearance. You can pick 480p, 720p or 1080p, set the length anywhere from 2 to 30 seconds, and pass a reference audio track as well. Carat pairs each Wan 3.0 result with the prompt behind it, so you can copy and run it directly.
Wan 3.0 lets you set a length between 2 and 30 seconds in a single generation. Producing a long clip in one pass avoids the character and background mismatches that appear when short cuts are generated separately and stitched together in an editor. Reference image input keeps a cast member or a product looking the same throughout, and the choice of 480p, 720p or 1080p means you can iterate cheaply at a low resolution and re-render only the final version at 1080p.
Set the length and resolution first, then spell out the job each reference image is doing. That is the basic order for a Wan 3.0 prompt. Copy an example below and change only the parts you need.
Copying prompts is free. Generating a video uses credits.
It is a video generation model in Alibaba's Wan series.
Length can be set from 2 to 30 seconds, and the resolution options are 480p, 720p and 1080p.
Yes. You can use an image as the opening frame, or supply reference images that define how a person or object should look.
An opening frame image becomes the first frame of the clip. A reference image tells the model how a person or object should look throughout the video.
You can pass a reference audio track as input.
Longer clips need more direction. Break the prompt into time blocks, describe the framing and camera movement in each, and restate the elements that must stay constant.