Vidu AI Video Generator: Put Your Photos in One Scene with Vidu Q4 Preview and Q3 Pro

• Vidu Q4 Preview

  Turns a photo, or up to 15 reference photos, into a clip up to 4K.

  • 3–16 s·
  • 540P / 720P / 1080P / 2K / 4K·
  • Sound on or off·
  • Up to 15 reference photos

  From 60 credits

  Use this model

• Vidu Q3 Pro

  From a sentence or start and end frames, up to 16 s.

  • 1–16 s·
  • 540P / 720P / 1080P·
  • Sound on or off

  From 20 credits

  Use this model

1. April 2024

   Vidu 1.0

   ShengShu and Tsinghua University unveiled it as China's first large video model, built on their U-ViT design that mixes diffusion with a transformer.

2. November 2024

   Vidu 1.5

   Introduced multiple-entity consistency: hand it separate pictures of people, objects or places and it keeps each one recognizable inside a single clip.

3. January 2025

   Vidu 2.0

   Focused on speed and price; ShengShu said it halved the per-second cost of 1.5.

4. April 2025

   Vidu Q1

   Added transitions between two unrelated first and last frames, plus background music and sound effects from text.

5. Late 2025

   Vidu Q2

   Sharpened facial expressions and camera control, with synchronized dialogue and sound effects.

6. January 2026

   Vidu Q3

   Generated picture and sound together in clips up to 16 seconds, aimed at film and TV style scenes.

7. October 2026

   Vidu Q4 Preview

   Released ahead of the full Q4 model, it works from images only and makes clips with sound.

1. 01

   Pick a Vidu version

   The generator above opens on a Vidu model. Use the version cards to switch between them: Q4 Preview when you have pictures to work from, Q3 Pro when you want to begin with words.
2. 02

   Add your pictures

   For Q4 Preview, drop one photo into the start frame, or switch to reference mode and add up to 15 photos of the people, pets and places that belong in the scene. Q3 Pro takes a start frame with an optional end frame, or no image at all.
3. 03

   Tell Vidu what happens

   A single start frame can go without a prompt. Reference photos need a sentence that explains how they come together, such as “the dog leaps into the woman's arms on the beach”, and a text-only clip on Q3 Pro needs the whole scene written out.
4. 04

   Choose length, resolution and sound

   Set the duration and resolution. Sound starts on, and turning it off doesn't lower the price shown on the Generate button.
5. 05

   Generate and review

   Press Generate once the credit figure looks right. The clip shows up in My creations, where Remake reloads your photos and prompt for another attempt.

• Room for 15 reference photos

  Q4 Preview accepts more reference images than any other video model in Plimo's menu, so separate portraits of a whole friend group can share one scene.

• Vidu beside other video models

  Switch between the two Vidu versions in a click, or run the same photos on Seedance or Hailuo AI without a second login.

• From 540p up to 4K

  Q4 Preview runs from 540p through 1080p to 2K and 4K, so a clip can begin as a low-cost test and finish sharp enough for a big screen.

• Sound at no extra cost

  Both versions can add audio to the clip, and the price is the same with it on or off, so you only decide whether you want it.

• Free credits to get started

  Vidu is a member model, but every new account gets free credits once, which you can spend on Seedance 2.0 Mini or Grok Imagine 1.5 video to try Plimo first.

• Clean clips you can use for work

  Vidu videos made with a membership download without a watermark, may be used commercially and stay in your account longer.

• Kling

  Kling 4.0 runs to 30 seconds and can edit an uploaded clip, while Vidu stops at 16 seconds but Q4 Preview takes up to 15 reference photos to Kling's 7.

• Veo

  Veo 3.1 Fast gives a fixed 8-second shot with a cinematic look, whereas Vidu lets you set the length and switch the sound off.

• Seedance

  Seedance accepts short reference clips alongside photos and can be tried on free credits through its 2.0 Mini, while Vidu guides the scene with pictures only.

• Hailuo AI

  Hailuo AI's MiniMax H3 Max aims at realistic faces with sound in every clip, and takes start and end frames or up to nine photos plus three clips.

• Grok Imagine

  Grok Imagine 1.5 begins at a 6-second clip with built-in sound and can be tried on free credits, and the Grok Imagine series also makes still images.

• Wan

  Wan 3.0 stretches from 2 to 30 seconds and can follow reference clips as well as up to ten photos.

• FLUX

  FLUX 3 Video lets you check a rough draft before the full render and runs up to 20 seconds, working from frames rather than reference photos.

• SkyReels

  SkyReels V4 can end on a chosen frame even without a start frame, which Q4 Preview can't, and it always needs a written prompt.

• HappyHorse

  HappyHorse 1.1 works much like Q4 Preview, from a start frame or a set of reference photos, but caps references at nine and needs a prompt every time.

• A group shot nobody was in together

  Add a separate portrait of each friend, up to fifteen, and describe everyone crowding around one birthday cake as the candles go out. Q4 Preview is built to keep each face recognizable.

• You, your dog and your favorite café

  Use reference photos of a jacket, your dog and a café interior, then ask for you walking in wearing the jacket with the dog trotting at your heels.

• Then and now in one shot

  On Q3 Pro, set a childhood photo as the start frame and a recent one as the end frame. The clip travels from one to the other in a single move.

• A scene from words alone

  Q3 Pro can work without any image. Describe a paper lantern rising over a night market and drifting past lit rooftops, and add the sound of the crowd below.

• A portrait that comes alive

  Put a portrait into the Q4 Preview start frame, leave the prompt empty and see what motion it chooses.

• A product in a styled setting

  Give Q4 Preview a product photo, a model shot and a backdrop, then describe the model lifting the item off a shelf. Made during a membership, the result can go on your shop page.

• One long take with no cuts

  Write a single continuous shot on Q3 Pro, like a cyclist weaving through a morning market from one end to the other, and let it run the full 16 seconds.

• 

  AI Video Generator

  Make a short video clip from a written idea, a photo or both, with a choice of video models.
• 

  Image to Video

  Bring a still photo to life with movement, camera motion and, on models that support it, sound.

All models