Vidu

Vidu: Reference-Driven AI Video with Consistent Characters

Vidu is an AI video generator built around a problem most text-to-video tools still have: characters and objects that change appearance between shots. Alongside the usual text-to-video and image-to-video modes it offers Reference-to-Video, where you supply three or more images of a subject and Vidu keeps that subject consistent across generations. Clips run 3 to 16 seconds at up to 1080p, up to four at a time, which places it squarely in the short-form and e-commerce end of AI video rather than long-form production.

Key Features

• Reference-to-Video — Feed it three or more images of a character, product or object and generate clips that keep the subject recognisably the same, which is the feature most people arrive for.
• Text, image and reference modes — Text-to-Video, Image-to-Video, Reference-to-Video, plus Text-to-Image and Reference-to-Image for stills.
• Vidu Q3 and S1 models — Q3 is the real-time model with audio generation; S1 targets interactive generation with millisecond-scale response.
• Vidu Agent — An assisted mode that handles more of the prompting and sequencing for you, gated by plan tier.
• Batch generation — Paid plans generate up to four videos at once, with H.265 Pro and H.264 Pro encoding on higher-quality exports.
• Commercial rights on paid plans — Generated content is licensed for commercial use once you are on a subscription.

How Vidu Works

Everything runs on credits. You pick a mode, write a prompt, attach reference images if you are using reference mode, and each generation draws from your monthly credit balance. Vidu splits credits into three kinds with different lifespans: subscription credits that expire after 30 days, purchased credits that last two years, and bonus credits — earned from sign-up, daily logins and events — that also last two years. That distinction matters more than it first appears, because it means an unused month genuinely disappears while topped-up credits do not.

What the Free Plan Actually Gives You

This is the part worth reading carefully. Vidu's free tier is listed as 10 references per month and carries no monthly credit allowance of its own. Reference slots are not generations — so in practice, free usage runs on bonus credits from signing up, logging in daily and taking part in events, not on a predictable monthly budget. It is enough to evaluate the reference-consistency feature, which is the thing worth evaluating, but it is not a free tier you can plan work around.

Plans and Pricing

Paid plans are priced per month billed yearly. Standard is $8/month ($96/year) for 800 credits and roughly 200 videos, with a 50/month subject-creation quota and essential Agent features. Premium is $28/month ($336/year) for 4,000 credits and around 1,000 videos, adding a high-speed generation channel, a 100/month subject quota and early feature access. Ultimate is $79/month ($948/year) for 8,000 credits, a 300/month subject quota, unlimited 1080p image generation and an off-peak mode that accepts up to 200 video submissions a day. Monthly billing is available at a higher rate on every tier, and credit add-ons are discounted 45–55% depending on plan.

How to Generate a Consistent Character with Vidu

1. Collect at least three clear images of the character, product or object you want to keep consistent.
2. Choose Reference-to-Video and upload them as your reference set.
3. Write the prompt describing the shot, not the subject — the references are already carrying the subject.
4. Generate, and use batch mode on a paid plan to produce up to four takes at once.
5. Export at up to 1080p, choosing the Pro encoding formats where output quality matters.

Who It's For

• E-commerce and product marketers who need the same product to look the same in every clip
• Short-form creators producing social video where 3–16 seconds is the native length
• Designers and studios prototyping motion concepts before committing to a full production
• Anyone who has hit the character-drift problem in other text-to-video tools

Conclusion

Vidu's clearest advantage is reference consistency — it is built around keeping a subject stable rather than treating that as a happy accident, and for product and character work that is the difference between usable output and a nice demo. The trade-offs are the short 16-second ceiling, 1080p as the maximum resolution, and a free tier that is a demonstration rather than a working allowance. If consistent subjects across shots is the thing blocking you, it is worth the trial; if you need long-form or 4K, it is not the tool.