PicX Studio

VIDEOShengshu (Vidu) · from 18 credits / sec

Shengshu's performance-first video model, in 4K

Vidu Q4 AI Video Generator: Image and Reference to Video

Image or 15 references. 3 to 16 seconds. 540p to 4K. Sound on or off.

Vidu Q4 is built for acting: faces, bodies and voices that move together, fast action that holds its shape, and explosions, smoke and particles that read as real. Start from one still or a set of references.

Artwork: an illustration for this page, not model output
Artwork: an illustration for this page, not model output
  • 3–16s
  • 540p–4k
  • Sound on or off
  • Image · References

Performances, not just motion

Acting that carries the shot

Shengshu pitched Q4 on expressive performance: a face, a gesture and the sound of the scene working as one. Write the emotional beat, not only the movement, and Q4 plays it out over up to 16 seconds.

One scene from up to 15 references

Reference-to-video takes 1 to 15 images: characters, a location, a prop, a style frame. Name them in the prompt by order, such as "the woman in reference image 1", and Q4 places them together and keeps them on-model. Pick 16:9, 9:16, 1:1, 4:3 or 3:4.

Fast action and real-looking VFX

The maker singles out running, martial arts and combat, with fewer warped limbs and stutters, plus explosions, smoke, particles and energy effects. Image mode keeps your still's framing and aspect ratio, so the effect lands where you composed it.

Vidu Q4, from its maker

Official release video by Shengshu (Vidu)

Official release video by Shengshu (Vidu): Meet Vidu Q4 Preview—our next-gen flagship AI video model on Vidu AI Official. Not made on PicX Studio.

Made for the work you ship

  • AI series and short drama Keep a cast consistent across episodes with the same reference set.

  • Ads and product spots Put the product, the model and the set in one reference-driven shot.

  • Action and effects shots Fights, chases, smoke and particle effects from a single key frame.

  • Social clips Vertical 9:16 scenes with sound, ready to post.

  • Animated key art Bring a poster or concept still to life in its own aspect ratio.

Vidu Q4

Why pick Vidu Q4

Try Vidu Q4
  • 15Reference imagesCast, set and style in one shot.
  • 4KTop resolution540p drafts, then 1080p, 2K or 4K.
  • 16sAny length from 3 sPick the exact second count.
  • AudioNative sound toggleEffects and ambience, or silent.
  • 5Reference-mode ratios16:9, 9:16, 1:1, 4:3, 3:4.
  • FrameImage modeKeeps your still's aspect ratio.

What Vidu Q4 costs

Billed per second of video, at the resolution you pick. The same prices the generator shows before you run it.

  • 540p
    18credits / sec
    5s clip 90 creditsOn Starter≈ 40 clips / mo
  • 720p
    38credits / sec
    5s clip 190 creditsOn Starter≈ 18 clips / mo
  • 1080p
    48credits / sec
    5s clip 240 creditsOn Starter≈ 15 clips / mo
  • 2k
    76credits / sec
    5s clip 380 creditsOn Starter≈ 9 clips / mo
  • 4k
    156credits / sec
    5s clip 780 creditsOn Starter≈ 4 clips / mo

Every model on every planCredits never expireSee plans

Pay for Vidu Q4 with crypto

Buy credits with Bitcoin, USDT, USDC, ETH, Solana and more — no card, nothing auto-renews, packs from $5, and 15% extra credits on crypto payments.

How crypto checkout works

Vidu Q4 vs other video models on PicX

How the options compare in the PicX generator. Vidu Q4 is the one for 4K and for big reference sets, but it always needs an image to start from.

Vidu Q4Kling 4.0Seedance 2.5Hailuo 2.3
Text to videoNoYesYesNo
Image to videoYesYesYesYes
Reference to videoYes, up to 15 imagesYesYes, images, video and audioNo
Top resolution4K1080p1080p1080p
Clip length3 to 16 s3 to 30 s4 to 30 s6 or 10 s
Native audioOn or offAlways onOn or offNo

Specs as offered in the PicX generator. Vidu Q4's voice references are not available on PicX yet.

How to use Vidu Q4 in 3 steps

  1. 1

    Choose image or references

    Upload one still to animate it, or switch to reference mode and add up to 15 images of the characters, place and look. Vidu Q4 has no text-only mode.

  2. 2

    Direct the performance

    Say who does what and how they feel, how the camera moves, and what the scene sounds like. In reference mode, point at each input by number.

  3. 3

    Choose seconds, size and sound

    Any whole number of seconds from 3 to 16, any resolution from 540p to 4K, sound on or off. Test a prompt cheaply at 540p, then render the version you keep at 2K or 4K.

Vidu Q4 prompts to copy

Prompts to start from. Copy one, or open it in the generator with the model already picked.

  • Plating up

    The chef slides the food onto a plate and smiles at the customer, lanterns sway

    Try it
  • Fox turns away

    The fox tilts its head, then turns and trots off between the pines, snow falling

    Try it
  • Rooftop fight (image to video)

    The two fighters circle in the rain, then trade a fast flurry of kicks and blocks; one is thrown back into a puddle that bursts into spray, handheld camera tracking the action, thunder rolling

    Try it
  • Two-character scene (reference to video)

    In the café from reference image 3, the man in reference image 1 slides a ring box across the table to the woman in reference image 2; she freezes, then covers her mouth and laughs, slow push-in on her face, quiet café chatter

    Try it
  • Energy blast VFX

    The mage raises both hands and a crackling blue energy sphere grows between them, then fires down the canyon and explodes against the rock wall in fire, smoke and falling debris, camera shakes

    Try it
Browse all templates

Vidu Q4 questions

What is Vidu Q4?

Vidu Q4 is the newest video model in the Vidu family from Shengshu Technology, a generative AI company that came out of Tsinghua University's AI research. Shengshu launched it as "Vidu Q4 Preview" in October 2026, aimed at short-drama studios, ad agencies, social creators and film teams. It makes video from one image or from a set of reference images.

What is Vidu Q4 good at?

Shengshu's launch focused on three things: expressive performances where face, body and voice match, steadier camera moves and cuts, and convincing action and effects such as fights, explosions, smoke and particles. On PicX it renders from 540p up to 4K.

How does Vidu reference to video work?

Add 1 to 15 reference images (people, a location, props, a style) and write a prompt that names them by order, for example "the woman in reference image 2 walks into the room from reference image 4". Vidu Q4 builds one shot that keeps those subjects consistent. Vidu also accepts up to three voice references to match a character's voice, but that is not available on PicX yet.

Can Vidu Q4 make a video from text only?

No. Vidu Q4 works from images: either one start image or a set of references. If you only have an idea, generate a key frame with an image model first, or use Kling 4.0 or Seedance 2.5, which both accept text prompts on PicX.

Does Vidu Q4 generate sound?

Yes. Sound is generated with the picture by default, and you can switch it off for a silent clip if you plan to add your own music. The price is the same either way.

What lengths, resolutions and aspect ratios does Vidu Q4 support?

On PicX you can pick any whole number of seconds from 3 to 16, and 540p, 720p, 1080p, 2K or 4K. In image mode the clip keeps your image's aspect ratio. In reference mode you choose 16:9, 9:16, 1:1, 4:3 or 3:4.

Vidu AI vs Kling: which should I use?

Pick Vidu Q4 when you need 4K output or want to combine many reference images in one scene. Pick Kling 4.0 when you want to start from text alone or need clips longer than 16 seconds. Both run on the same PicX credits, so you can try a shot on each and keep the better take.

Is Vidu AI free?

Not on PicX. Vidu Q4 runs on paid credits, charged per second of video, and the rate rises with resolution, so the 540p setting is the cheapest way to test an idea. The pricing table on this page shows the live rates. New accounts get one free image to try PicX, but video generation needs a plan or a credit pack.

Direct a scene with Vidu Q4

One still or 15 references, sound on, up to 4K.

FREEClaim Free Image Generation