Kling 4.0 Reference builds a clip of 3 to 30 seconds from named materials: reference images, reference clips, subjects and voices, each addressed by name in the prompt.
References guide the subjects, setting and sound of a shot; keyframes guide its composition at selected moments. Neither guarantees an exact reproduction. These are separate models here; see the Kling 4.0 Keyframe guide for the frame workflow.
Materials are combined in one task, up to fifteen of them. Images and clips are addressed by their numbered names, subjects and voices by the name you give them, so a single prompt can direct several materials at once.
Attach materials in the reference panel, then describe the shot using their names. Keep original sound appears once a reference clip is supplied, and the shape selector offers automatic, 16:9, 9:16, 1:1 and 21:9.
Up to ten reference images, five reference clips including character clips, seven subjects and three voices, fifteen materials in one task. Up to 15 materials total; up to 3 video subjects. Reference clips and video subjects share a limit of 5 videos and 30 seconds combined.
Build a subject from two to four images or from one continuous character clip, give it a name, and optionally bind a voice to it.
With no materials attached the same model generates from the description alone, and you still choose length, quality tier and shape.
Use two to four pictures of one person or mascot as a named subject to guide its appearance in a new scene. Review the result for identity and detail consistency.
Add up to three voice materials, or bind one to a character subject, and describe in the prompt what each of them says.
Supply a reference clip and switch on keep original sound when the existing audio matters more than a generated track.
Use a button on this page. It opens the right workspace with this model already selected, in the language you are reading now.
Write what should happen in the shot or image. Concrete subjects, framing and action give the model more to work with than short labels.
Choose length or quality tier, shape and sound where offered, then read the credit quote for those exact settings before you generate.
Generation uses account credits. Cost depends on the model, quality and output size or duration. Reference images, video or audio may add charges in some modes. Check the workspace quote for your selected inputs and settings before generating; the pricing page lists credit packages.
See credit packagesReferences guide the subjects, setting and sound of a shot; keyframes guide its composition at selected moments. Neither guarantees an exact reproduction. These are separate models here; see the Kling 4.0 Keyframe guide for the frame workflow.
Fifteen in one task: up to ten reference images, five reference clips, seven subjects and three voices. Reference clips and character clips share the five-clip limit and the 30-second combined limit. Up to 15 materials total; up to 3 video subjects. Reference clips and video subjects share a limit of 5 videos and 30 seconds combined.
Images and clips are numbered, such as @image_1 or @video_1, and subjects and voices use the name you gave them. Every @ mention has to match a material you supplied, and names must be unique.
The quote depends on output duration, quality tier and input video duration, including character clips. Check the displayed credit total after attaching all materials and before generating.
Yes. With nothing attached the model generates from the description alone, and you still pick the length, quality tier and shape.
You can bind a voice to a character subject or supply standalone voice materials. What the model does with them has to be judged from the result: the workspace checks the files, not the performance.
Other models on this platform that do a similar job, or that pair well with this one.
Kling 4.0 Keyframe turns a description or an ordered set of frames into a clip of 3 to 30 seconds that travels between the pictures you supply.
Read the guideSeedance 2.5 creates a clip from a written description or from one starting image, with whole-second lengths from 4 to 30 seconds and three quality tiers.
Read the guideMiniMax H3 turns a description or a single starting image into a clip of 4 to 15 seconds that carries a generated sound track.
Read the guide