Provider
Vidu AI Models on Martini
Martini hosts both Vidu video and Vidu Image — the model family known for reference-driven generation that locks a subject across multiple shots. Vidu is the easiest model on the canvas for keeping a custom character consistent through a sequence, and it pairs naturally with Kling and Nano Banana 2 for the rest of the workflow.
About Vidu
Vidu is the AI video model family from Shengshu Technology, originally a Tsinghua University spinout. The model is best known for its reference-driven mode — feed it one to seven reference images and it generates video that keeps subjects, props, or scenes consistent across the clip. This makes Vidu one of the easiest models on the canvas for the very specific job of taking a custom character or product and making it appear, in motion, in a brand new scene without losing its identity.
On Martini, Vidu video sits next to the major flagship models (Sora 2, Veo, Kling 3) but is positioned for a different job: when you need character or subject lock, Vidu wins. Vidu Image extends the family into still generation with the same reference-aware behavior, useful for producing a matched still that downstream Vidu video can animate. Teams building IP-driven content (mascots, recurring characters, branded scenes) tend to anchor on Vidu and route prestige hero shots through Sora 2 or Veo.
Pricing is metered per generation and shares the Martini wallet, so there's no separate Vidu.com or Shengshu account to manage. Workspace billing handles attribution. Because Vidu interoperates with every other provider on the canvas, a typical pattern is Nano Banana 2 to design the character, Vidu video to keep it consistent across multiple action shots, and Kling 3 or Sora 2 for the cinematic close-up. The flexibility to mix providers per shot is the main reason teams use Vidu on Martini rather than its standalone product.
Available Vidu models on Martini
Best use cases
- Character consistency across multiple shots driven by reference images
- Branded mascots or recurring characters that must look the same scene to scene
- Product video where the exact product has to appear in new environments
- Multi-shot narrative video built around an IP or character
- Reference-driven still generation in Vidu Image that feeds Vidu video downstream
- Hybrid pipelines combining Vidu character lock with Sora 2 or Veo cinematography
Recommended workflows
AI Character Consistency
Vidu is the easiest model on the canvas for keeping a custom character locked across shots via references.
AI Video Reference Images
Vidu accepts up to seven references and is the canonical model for reference-driven video on Martini.
Multi-Shot AI Video
Vidu keeps subjects identical across multi-shot sequences in a way most flagship models still struggle with.
AI Influencer Video Generator
Lock an influencer or avatar appearance across many clips by driving Vidu with a reference set.
Related features
Related reading
Other providers
Frequently asked questions
How many reference images can Vidu accept?
Vidu accepts up to seven reference images, which can mix subjects, props, and environments. This is the primary reason it leads on character consistency across multi-shot sequences.
How does Vidu compare to Kling O3 and Nano Banana 2 for character consistency?
Vidu wins for multi-reference video where several subjects must stay locked. Kling O3 is strongest for video-to-video restyling. Nano Banana 2 leads on photoreal human character lock in stills. Use them together on the same canvas.
Is Vidu Image the same as Vidu video?
Vidu Image is the still generation companion — same reference-aware behavior, but produces matched stills that feed downstream Vidu video for animation.
Do I need a Vidu.com account?
No. Martini provides hosted access to Vidu and Vidu Image. You pay with Martini credits and never need a separate Vidu subscription.
Can Vidu animate a custom mascot reliably?
Yes — feed Vidu the mascot reference image (or several poses), prompt the new scene, and the model preserves the design across the generated clip. This is its strongest use case.
Can I combine Vidu with Sora 2 in one workflow?
Yes. A common pattern is Vidu for action and continuity shots where the character must stay locked, plus Sora 2 or Veo for cinematic hero close-ups. The whole pipeline lives on one canvas.
Build with Vidu on Martini
Open the canvas and wire Vidu into the rest of your stack in minutes. Free to start — no card required.