# Vidu AI: ShengShu's video model with sound — Q3 Pro, Turbo and Q4

> Vidu AI is a family of video models from the Chinese company ShengShu Technology. The current Vidu Q3 lineup (since late January 2026) generates clips with sound — speech, effects and music — up to 16 seconds long at 1080p and switches shots on its own. Vidu Q4 Preview arrived in October 2026 with 2K and 4K. Vidu is available in the AI2Face generator.

Developer: ShengShu Technology · Released: 2026-01-27 · Updated 2026-10-27
URL: https://ai2face.com/models/vidu

## Key facts

| | |
|---|---|
| Developer | ShengShu Technology (China) |
| Current versions | Vidu Q3 (Pro, Turbo, Mix, Drama, Ad), Vidu Q4 Preview (October 7, 2026) |
| What it does | Video with sound from text, a frame, first and last frames, or references |
| Duration | Q3 Pro, Turbo and Mix — 1–16 seconds; Drama and Ad — 3–15 seconds |
| Resolution | Q3 — 540p, 720p, 1080p; Q4 Preview — up to 2K and 4K, 24 fps |
| Aspect ratio | 16:9, 9:16, 4:3, 3:4, 1:1 |
| References | Q3 — 1–7 images; Q4 Preview — up to 15 images plus voice references |
| Sound | Dialogue, voice-over, effects and music in one generation |
| In AI2Face | Yes, in the generator |

## What is Vidu AI

Vidu is a family of video models from the Chinese company ShengShu Technology. Since November 2025, Vidu models generate video with sound included: character dialogue, voice-over, sound effects and music.

Two lineups are current:

- **Vidu Q3** — the main one, since late January 2026. Pro reached the API on January 27, the fast Turbo on February 11, and reference-based generation on April 13.
- **Vidu Q4 Preview** — the next generation, in preview since October 7, 2026: 2K and 4K, up to 15 references and voice references.

Vidu runs on vidu.com, in mobile apps and through a developer API. Beyond video, ShengShu also offers models for images, voice-over, voice cloning and real-time avatars.

## What Vidu Q3 can do

**Up to 16 seconds with sound.** Dialogue, voice-over, sound effects and music are generated with the video in a single pass. You can also turn sound off and get a silent clip.

**Automatic shot switching.** Q3 changes shot size and angle inside a clip without a separate storyboard — ShengShu calls it "intelligent shot switching." Camera moves and pacing can be controlled down to individual frames.

**Multi-character dialogue.** The model keeps a conversation between several characters going, syncs lips to the lines and places the characters in the frame.

**Strong anime and comic dramas.** This is the main use case — ShengShu even released a dedicated Q3 Drama for it: stylized animation, cinematic shots, short series and story-driven ads.

**Multiple references.** Up to 7 images: characters, a location, props. ShengShu's own example is "three characters and one scene" combined into one animated clip.

![Three full-body anime characters on a white background: a young swordsman in a green coat, a girl with a silver braid in a teal cloak and a small orange three-tailed fox](https://ai2face.com/_astro/characters.Dk6mASwL_1Mw9wx.jpg "A character sheet for reference mode: three characters, front view, on a clean background. Vidu Q3 accepts up to 7 images like this. Image made with Nano Banana Pro")

![An anime-style tea house interior: paper lanterns, low tables with cushions, open sliding doors to a garden with maple trees, steam above a kettle](https://ai2face.com/_astro/scene.D6UQcsVB_1cVyqe.jpg "A separate location image with no characters — the second type of reference. Together with the character sheet, it sets where the action happens and who's in it. Image made with Nano Banana Pro")

**Off-peak mode.** In the API you can queue a task as off-peak: it costs less but is delivered within 48 hours. If it can't finish in time, it's canceled and the credits are refunded.

**Tools around the model.** Extending a clip by 1–7 seconds, "smart multi-frame" with up to 9 keyframes for longer scenes, motion transfer from a video onto a character, lip sync to text or audio, and upscaling to 1080p, 2K and 4K.

## Vidu Q3 Pro vs. Turbo vs. Mix vs. Drama vs. Ad

| | Q3 Pro | Q3 Turbo | Q3 Mix | Q3 Drama | Q3 Ad |
|---|---|---|---|---|---|
| Best for | top quality | speed and cost | balanced, general use | comic dramas and film | short 5–8 s ads |
| Resolution | 540p, 720p, 1080p | 540p, 720p, 1080p | 540p, 720p, 1080p | 720p, 1080p | 720p, 1080p |
| Length | 1–16 s | 1–16 s | 1–16 s | 3–15 s | 3–15 s |
| Text to video | yes | yes | — | — | — |
| Image to video | yes | yes | — | — | — |
| First and last frame | yes | yes | — | — | — |
| References | — | yes | yes | yes | yes |

ShengShu calls Q3 Drama the best in the series for dialogue, character positioning, motion and cinematic camera language. Q3 Ad builds short ads with sensible pacing and keeps product details consistent.

## Vidu Q4 Preview: what we know

Q4 Preview launched on October 7, 2026. According to ShengShu:

- **2K and 4K** instead of Q3's 1080p, with the same 1–16 seconds;
- **up to 15 references** instead of 7;
- **voice references** — the same voice for a character across clips;
- **action** — fights, running and clashes without jerky motion;
- **effects** — explosions, fireworks, particles and energy effects with more detail;
- **modes** — image to video and reference-based generation.

ShengShu hasn't announced pricing or a date for the full release.

## What Vidu is good for

- **Anime and comic dramas** with several characters and dialogue — Q3 Drama's strength.
- **Short series** with the same characters episode to episode, using a character sheet as the reference.
- **Story-driven 5–15 second ads** — Q3 Ad.
- **Quick drafts** on Q3 Turbo before the final on Pro.
- **Music videos:** a separate ShengShu tool builds a video from a track and 1–7 images.

## Vidu vs. alternatives

| | Vidu Q3 | [Kling 3.0](/models/kling-3-0) | [Seedance 2.5](/models/seedance-2-5) | [Hailuo](/models/hailuo) |
|---|---|---|---|---|
| Developer | ShengShu | Kuaishou | ByteDance Seed | MiniMax (H3) |
| Length per generation | 1–16 s | up to 15 s | 4–30 s | 4–15 s |
| Max resolution | 1080p (Q4 — 4K) | 4K | 1080p | 2K |
| Sound | yes | yes | yes | yes |
| References | up to 7 images | up to 7 images or elements | 30 images, 10 videos, 10 audio | 9 images, 3 videos, 3 audio |
| Strength | anime, comic dramas | storyboards, characters with voices | long stories | speech in 11 languages, open weights |
| In AI2Face | yes | yes | yes | yes |

People pick Vidu for stylized animation and multi-character dialogue. For photorealism and step-by-step storyboards, Kling 3.0 is more convenient; for one long continuous story, Seedance 2.5; for speech in more languages, [MiniMax H3](/models/minimax-h3) from the Hailuo family.

## How to write prompts for Vidu

1. **Lead with style and genre:** "anime, cinematic shot, soft sunset light."
2. **Describe the action in order.** Q3 splits it into shots on its own, but you set the sequence of events.
3. **Put dialogue in quotes with the character's name.** Speech works in English, Japanese or Chinese.
4. **For references, say who is who:** "the swordsman from image 1, the braided girl from image 2, the location is image 3."
5. **Test the idea on Turbo,** then render the final on Pro or Drama.

```
Anime style, cinematic, warm dusk light. The swordsman from image 1 and the silver-braided girl
from image 2 enter the tea house from image 3, the fox spirit from image 1 trots behind them.
The girl sits down and pours tea, steam rises. The swordsman says: "We rest here tonight."
The fox curls up by the kettle and yawns. Sound: rain outside, a kettle whistling softly.
```

## How to use Vidu in AI2Face

Vidu is in the AI2Face generator next to [Kling 3.0](/models/kling-3-0), [Seedance 2.5](/models/seedance-2-5) and [Veo 3.1](/models/veo-3-1).

1. Sign up — new accounts get starter credits.
2. Open the generator, switch to video and pick Vidu.
3. Upload a frame or references, write the prompt, set the length and resolution.
4. The credit cost is shown before you start. Credits don't expire.

You can create character and location references in the same place, for example with [Nano Banana Pro](/models/nano-banana-pro): every image on this page was made that way.

## Limitations

- **1080p on Q3.** 2K and 4K are only in Q4 Preview and the separate upscaler.
- **Not every mode in every version.** Text to video, image to video and first-last frame are in Pro and Turbo; Mix, Drama and Ad work from references only.
- **Speech languages.** Officially English, Japanese and Chinese.
- **Video as a reference.** In reference mode, Q3 accepts images only; video input is supported by the older Q2 Pro.
- **Q4 is a preview.** Pricing and the full release date haven't been announced.

## Bottom line

Vidu is the pick for anime, comic dramas and story-driven ads where multi-character dialogue and automatic shot switching matter. Q3 Turbo works well for drafts, Q3 Pro and Drama for the final cut, and Q4 Preview promises 4K and consistent character voices. For photorealistic scenes and step-by-step storyboards, compare the result with [Kling 3.0](/models/kling-3-0); for a 30-second continuous story, with [Seedance 2.5](/models/seedance-2-5) — all in the same AI2Face generator.

## FAQ

### What is Vidu AI?

A video generation model from the Chinese company ShengShu Technology. It makes clips with sound from text, photos and sets of references. The current lineup is Vidu Q3, and the new Q4 entered preview in October 2026.

### How long are Vidu videos?

Q3 Pro, Turbo and Mix generate 1 to 16 seconds per run. The specialized Q3 Drama and Q3 Ad generate 3 to 15 seconds.

### How is Vidu Q3 Pro different from Turbo?

Pro has the best quality — depth, expressiveness, accurate sound. Turbo is faster and cheaper with stable results. Length, resolution and sound are the same for both.

### What is Vidu Q4?

The next generation, released in preview on October 7, 2026. It adds 2K and 4K, up to 15 references, voice references, smoother high-action motion and better visual effects.

### Which languages does Vidu speak?

Officially, Vidu generates speech in English, Japanese and Chinese.

### Is Vidu available in AI2Face?

Yes. Vidu is in the AI2Face generator, and you see the cost before you start a generation.

## Sources

- https://www.vidu.com/vidu-q3
- https://platform.vidu.com/docs/overview/model-map
- https://platform.vidu.com/docs/api-reference/updates
- https://platform.vidu.com/docs/api-reference/video-models/vidu-q3/text-to-video
- https://www.vidu.com/blog/create-ai-animation-3-characters-1-scene-vidu-multi-reference