Skip to content

Video

Vidu Q3

Video

Available on Ofrai — chat, images, and video in one place with shared credits.

196 creditsModel details

Video studio

Preview

No clip yet

Your latest clip preview will show here after generation.

Open studio

Example prompts

Choose an example to load its prompt and recommended settings into the Vidu Q3 creator above.

How it works

Create with Vidu Q3 in three steps

  1. 01

    Choose your inputs

    Keep Vidu Q3 selected, then add a prompt and any supported reference assets.

  2. 02

    Set the output

    Choose the aspect ratio, resolution, and available Video options before generating.

  3. 03

    Generate and refine

    Review the result, adjust the prompt or settings, and generate another version from the same place.

Vidu Q3 is the base member of a three-model endpoint, and it is easy to assume that means it is the weakest of the three. The documentation says otherwise in fairly specific terms: among the Q3 models it offers the best multi-camera consistency, intelligent shot switching, and multi-reference generation. Everything else about it follows from that positioning. Reference images are required rather than optional, up to seven of them, so this is not an entry you can use for pure text-to-video.

Duration starts at three seconds rather than one. And audio is on by default across the whole Q3 series, which means a clip from the base model arrives with a soundtrack unless you say otherwise — a very different default from the silent standard tier of the Kling entries. The pricing line up is blunt as well: the high-quality model costs about twice the turbo one.

The base Q3 is the consistency member, not the cheap one

Three models share this endpoint and they are separated by what they optimise. The turbo model is described as the fastest and the best value, with intelligent scene switching; the pro model is the high-quality one, with audio-visual synchronisation and storyboard video generation; and the base model, this one, is the one with the strongest multi-camera consistency, intelligent shot switching, and support for multi-reference generation.

Read together, those three descriptions describe a genuine set of trade-offs rather than three marketing tiers, and the base model sits in the middle on price and at the top of the family on structural consistency.

The practical meaning is that base Q3 is the member to reach for when the same subject has to look right from more than one angle inside a single clip — which is a requirement that a faster, cheaper, less consistent model will quietly fail, and fail in a way that is not obvious until the clip is finished.

Where it visibly fails, and why

The required reference list is the first thing to understand, because it removes a capability rather than degrading it. For the pro and turbo models, passing no images is a valid text-to-video request; for base Q3 the image list is mandatory, up to seven entries. Any workflow that assumed it could fall back to text has to branch before submission. Second, the duration floor is three seconds, where the siblings accept one, so ultra-short inserts are not available here.

Third, intelligent shot switching cuts both ways: it is the feature that makes the consistency model useful across a longer clip, and it is also the feature that may introduce a cut you did not author and had not budgeted time for, so cut points should be chosen knowing the model is allowed to add its own. Fourth, audio defaults to true across the series, so a silent deliverable requires saying so explicitly, and a clip rendered with sound has a rhythm your edit has to account for.

Prompt structure this model responds to

The prompt is required and the documented ceiling is 5000 characters, the largest in this group, which leaves room for a subject defined once and then referenced throughout. The documentation's guidance is to describe subject, action, scene, camera and style, and the example given follows exactly that order. What makes base Q3 different from its siblings at the prompt level is the subject mechanism: when you declare subjects, they are referenced inside the prompt by name using an @name token rather than being re-described each time.

That matters most precisely when you are using the feature this model is built for. With several reference images in play, a named subject addressed consistently gives the model something stable to carry across angles, whereas re-describing the same character in every clause gives it four slightly different characters to average. Name each reference, address it by name, and keep style words at the end.

Parameters and where the money actually goes

Resolution takes 540p, 720p or 1080p, defaulting to 720p — the presence of a 540p draft tier is worth planning around, because a consistency model is usually iterated on more than once and drafts do not need delivery resolution. The Q3 series accepts arbitrary aspect ratios rather than a fixed set of three, and when you pass reference images the ratio is typically determined by the images themselves, so the source material becomes a framing decision.

Duration is three to sixteen seconds for base Q3, against one to sixteen for the other two members. On cost, the documented relationship is that the pro model is roughly twice the turbo price, which places base Q3 between them. Audio defaulting to true across the series also means sound is inside the base cost profile here, unlike the Kling standard tier where silence is the default and sound is a separate tier — that is a genuine part of the comparison, not a footnote.

When to pick a sibling instead

The siblings are not diluted versions of this model; each is the right answer to a different brief. Choose turbo when turnaround and price dominate, when you may not have reference imagery at all, when you want intelligent scene switching rather than multi-camera consistency, or when you need a one-second insert that base Q3 cannot produce — the turbo model accepts durations from one second and treats images as optional.

Choose pro when the priority is raw output quality, audio-visual synchronisation, or generating storyboard video from a planned sequence, and when the roughly twofold premium over turbo is justified by the shot being a hero asset. Choose base Q3 when consistency across angles matters more than either, when the brief genuinely provides several reference images, or when you are deliberately trading a little speed for a result that does not need identity repair in post.

Moving a turbo or pro job onto base Q3

The migration is small in code and large in expectations, because three defaults change at once. Reference images become mandatory, so a call that previously ran text-only now fails submission and needs up to seven images attached; treat that as a scheduling dependency, because sourcing those images is often the long pole. The duration floor rises from one second to three, so any one-second micro-insert in the old pipeline has to be moved to turbo rather than stretched.

And framing changes character: the series accepts arbitrary ratios, and with images attached the ratio is usually inherited from them, so source files chosen for identity reasons may also be silently choosing your aspect ratio. The one thing that does not change is audio, which stays on by default. Plan for the cut points too, since the intelligent shot switching this model is valued for is allowed to add a transition you did not plan.

FAQ

Can base Vidu Q3 generate a video without any reference images?
No. For the base model the image list is required and accepts up to seven references. Omitting it is valid for the pro and turbo models, where an empty list means text-to-video, but it is not valid here.
What durations are available?
Three to sixteen seconds for the base model. The pro and turbo members of the same endpoint accept one to sixteen seconds, so anything shorter than three has to be rendered on one of those.
Is audio included by default?
Yes. All Q3 models default to audio enabled, and setting it to true makes the system generate speech and background sound from the prompt. A silent deliverable therefore has to be requested rather than assumed.
Why choose the base model over turbo?
Base Q3 is documented as the strongest for multi-camera consistency, with intelligent shot switching and multi-reference generation from up to seven images. Turbo is faster and cheaper but built around scene switching and optional images.

FAQ

Can I use Vidu Q3 directly on this page?
Yes. The creator above is the working Ofrai studio with Vidu Q3 selected by default.
Where are my finished results saved?
Signed-in generations are kept in your Ofrai library so you can download or revisit them.
Which prompts work best with Vidu Q3?
Describe the subject, setting, visual style, lighting, and camera or motion clearly. The examples above provide practical starting points.
Can I edit the example prompts before generating?
Yes. Load any example, then change its wording, aspect ratio, references, or other supported settings in the creator.

Compare other Video models that can handle similar creative work to Vidu Q3.

Commercial use depends on the selected model and model terms.