Vidu is an AI video generation service from the Chinese company Shengshu, founded by researchers from Tsinghua University. Like most rivals, it makes short clips from text or a single photo, but its distinguishing feature is video from "several reference images": you provide a photo of a character, a product and a place separately, and Vidu combines them into one scene with a consistent look. Its newer generations also create sound, dialogue and music along with the picture.
What is Vidu good for?#
- Ads featuring a real product: give it your store's product photo along with a photo of a model and a setting, and it makes a clip where the product still looks like the real one.
- Short stories with a consistent character: a content creator shows a cartoon character in several different scenes with the same face.
- Anime and illustration styles: it gives good results for anime-style and illustrated clips.
- Bringing photos to life: turn an old photo or illustration into a short animated clip.
- Scene previews for filmmakers: try a combination of character and location before shooting.
Strengths and weaknesses#
Strengths:
- Building from several reference images, with consistent character and product appearance
- Good generation speed
- Synchronized sound in newer generations
- Daily login credit for free users, and a mobile app
Weaknesses:
- Clips are short, so for a full video you have to edit pieces together
- In busy scenes and complex motion, body parts or objects sometimes get mixed up
- Speech and text rendered inside images are unreliable outside the main languages
- Its tutorials and user community are smaller than those of Western rivals
Free plan and pricing#
Vidu is freemium: new users get initial credit and a small amount more with daily login. Free output usually has a watermark and limited quality, and its generation queue is slower. Paid plans give higher quality, watermark removal and commercial rights. Check the official site for current prices.
Prompts and language#
English prompts give the most precise results, and other languages are understood only to a degree. If a video has dialogue, do not rely on generated speech in less common languages; do the voiceover separately.
Getting started in three steps#
- Sign up on the site or mobile app with an email or Google account.
- Open the reference-image section and provide two or three clean photos with simple backgrounds.
- In an English prompt, say what role each reference plays in the scene and what motion it should perform.
Vidu alternatives#
- Kling: high realistic quality and more control and editing tools.
- Hailuo: natural body motion and a decent free allowance.
- Pika: simpler and more fun, with ready-made effects.
To know what to expect from these tools, read making and editing video with AI: what is possible and what is not, and before commercial release, see copyright and commercial use of AI-generated images.
