Vidu AI: text-to-video, image-to-video and reference workflows
Vidu supports several ways to start a video, including text prompts, a source image and reference-led creation. Its reference workflow is useful when a character, object or visual style needs to remain recognisable across multiple shots.
What you can use it for
- Text-to-video for creating a clip from a scene description.
- Image-to-video for animating a still image or starting frame.
- Reference-to-video for guiding scenes with multiple visual references.
- Reusable references for characters, props and visual styles, subject to the current product options.
How to get started
- Open the official Vidu website and sign in.
- Choose Text to Video, Image to Video or Reference to Video based on the material you have.
- For reference-led work, prepare clear images of the character, object or scene you want to preserve.
- Describe the action, camera movement, scene and visual style in the prompt.
- Choose the available aspect ratio, duration and quality settings, then review the credit cost shown before generating.
- Preview the result and refine the prompt or references if the subject changes between shots.
Example prompt to try
Adapt this prompt to your own subject and output requirements:
A cinematic medium-wide shot of the same friendly orange delivery robot moving slowly through a sunlit modern office lobby. Preserve the robot design, orange body panels and blue status light. The camera tracks gently from left to right. Natural daylight, realistic reflections, calm movement, polished commercial video style.
Things to check before using it
- Vidu states that users receive a set amount of free credits; the allowance and generation cost can change. Check the live balance before creating.
- Reference-to-video can help with subject continuity, but generated results still need visual review.
- Check export resolution, watermark settings, usage rights and commercial terms before publishing.
Official resource
Visit the official Vidu AI website ↗
This guide is an independent editorial overview, not a product endorsement. Interface labels, models, availability, credits and terms may change. Confirm current details in the official service.