Back to all models

Kuaishou / Video

Kling.
Put your references in motion.

Kling is Kuaishou’s video generation family. Its reference-led workflows can use images of subjects, objects and settings to guide a moving scene. That makes it relevant when a story needs recognisable people and places to return in new situations.

Inside the storyVideo
Fictional characters Anaya and Kabir facing each other across a warmly lit table.
AI Director’s shot briefSame cast. A different kind of tension.
01 / Cast

Attach both approved character reference sets.

02 / Location

Use the established table, doorway and light direction.

03 / Action

Kabir reaches forward. Anaya keeps her hand still.

Fictional story concept. Illustrative artwork, not output from the featured model.

Model capabilities

What Kling brings to the frame

Image-guided generation
Supply visual direction and describe the action you want to develop into a video shot.
Multiple subject references
Supported workflows can combine reference images for characters, objects and settings in a single scene request.
Directed movement
Text instructions can describe character actions and camera movement. The result still needs to be checked against the intended beat.

Capabilities describe the model families, not a guarantee that every version or control is available in Tosheo. Production access depends on provider availability, references, rights and the approved scope.

02 / Kling

Where Kling fits in your story

Kling is a candidate for recurring-cast coverage, character interactions and branded scenes with an approved object. The AI Director can assemble the relevant visual references and specify what changes in this shot, while preserving the story information that should carry across the sequence.

Fictional exampleSame cast. A different kind of tension.

Bring Anaya and Kabir into the same room without losing the faces, costume choices or distance established earlier.

Cast
Attach both approved character reference sets.
Location
Use the established table, doorway and light direction.
Action
Kabir reaches forward. Anaya keeps her hand still.

03 / Kling

Plan the Kling handoff

The AI Director prepares the video request from the scene plan, using AI Skills to specify the framing, performance or sound intention. You can review creative preparation or use Auto mode within your chosen scope. The model receives a bounded job, with the references and constraints that matter to that asset.

For an independent creator, this connects a creative decision to the next production step. For a publisher, it keeps recurring assets attached to the episode plan. Brand buyers can carry approved product references into the shot; production houses can retain the treatment and handoff requirements across the sequence.

04 / Kling

Why reference handling is the deciding factor

In Tosheo a shot does not describe its characters. It references them: an approved identity set, a costume state tied to this story day, a location with approved architecture and lighting, and the geography established in the preceding shot.

That turns a provider capability into a hard routing constraint. If a family cannot accept those inputs in a form that actually binds the output, the shot cannot be routed there, because the result would be a fresh interpretation rather than the character. Headline quality does not compensate for that. A beautiful stranger is still a stranger.

Identity references
Face and body, approved once and inherited. The provider has to take them as input, not as adjectives.
Location references
Approved architecture, layout and lighting for a room the season keeps returning to.
Continuity from the previous shot
Where the doorway sits, who is standing where, what is on the table.
Model restrictions
Some source assets carry a grant that forbids their use as a generation reference. The router has to know before it dispatches.

05 / Kling

How we would judge it, once we have run enough production

The metric that matters is not a score on a public leaderboard. It is acceptance: how often work routed here, for this kind of shot, under this treatment, is accepted by a human without repair.

  • Accepted shots per generation attempt, by shot type
  • Continuity warnings per accepted minute
  • Repair success without collateral change
  • Cost per accepted publish-ready minute under this treatment
  • Whether a repair can reproduce the conditions of the original

06 / Kling

Locked once approved

Provider documentation

Capabilities checked against official sources. Story applications are Tosheo’s editorial examples, not comparative benchmarks.

Kuaishou: Kling multi-image references

Questions before production

Does reference support guarantee a consistent character?

No, and we will not claim it does. It makes consistency achievable and, more importantly, makes a miss detectable - the result is compared against the same approved reference the shot inherited. Without that, drift is an opinion. With it, drift is a mismatch that can be repaired at shot level.

What happens if a shot cannot be routed anywhere that supports its references?

It goes back to the plan. Usually that means re-boarding the shot into something achievable rather than accepting a result that will fail QA, and the cost of that decision lands at the board stage where it is cheap.

Can a Chinese provider be used for an Indian production?

It is a rights, data and terms question answered per deliverable, not a geography question. Provider terms are part of the routing decision and are recorded in the provenance for every accepted shot, so the answer is always inspectable rather than assumed.

Do you use one provider for a whole episode?

Usually one approved treatment per scope, which in practice means a consistent provider and parameters across the episodes that have to match. Different modalities within a shot - picture, voice, music - can still route differently.

How do I know which model made my shot?

It is in the provenance: model and provider per generation attempt, the references and canon version it inherited, the parameters, every attempt rather than only the accepted one, and who accepted it. That record ships in the delivery package.

AI Director + AI Skills + Models

The model makes an asset. Your AI Director connects the story.

  1. 01

    Your direction

    Share the brief, format, episode plan and intended languages.

  2. 02

    AI Director

    Build the storyboard, references and shot-level direction with AI Skills.

  3. 03

    Model fit

    Match the asset to an available model and the approved production scope.

  4. 04

    Story-ready assets

    Check the result in context, repair what needs attention and assemble the edit.

Review creative preparation yourself or use Auto mode within your chosen scope. Cost authorisation, rights clearance and final delivery acceptance remain explicit decisions.

More from the model library

Keep building the story.

Tosheo / AI-native production

Start with the scene.
Let the direction connect it.

See how Tosheo brings the brief, AI Skills, generation and review into one production flow.