Kling Image

Kling's picture models from the makers of the Kling video app: cinematic stills, native 4K on Omni and up to 10 reference pictures.

What it nails

Watch out for

Kling is best known for video, but its makers at Kuaishou also build picture models, and they think like film-makers. Describe a shot the way a director would (who stands where, the lens, the light, the grain) and Kling gives you a still that looks like it came from a movie. Its other gift is keeping things: a face, a product or a whole scene stays the same while you change one detail.

The Lab runs five versions: Image 3.0, 3.0 Omni, O1, 2.1 and 2.0 New.

At a glance: Kling Image. The live spec card, price calculator and Generate button show here on the site.

Day 2: Kling IMAGE O1 is Officially Here! (Kling AI)

What Kling Image is secretly great at

Film stills from a shot description

Kling's Omni guide recreates movie frames from long, careful descriptions: who is left and right, their posture, the light, the lens height and the film grain. Write it like a script supervisor and it follows every line.

The image features two men and a red convertible, styled in a western crime film manner. The man on the left wears a black jacket and pants, with a relaxed posture; the man on the right is in a dark gray hoodie and jeans, with a tense stance. The red Ford convertible has a vintage look, fitting the desert standoff vibe. The two men face each other on opposite sides of the car, with the left man raising his hands in a deceptive gesture, and the right man with hands in his pockets, slightly turned in vigilance. The setting is a desert with red-brown rock hills, sparse vegetation, and a pale blue sky. The lighting is strong midday sunlight, creating sharp contrasts and shadows on the men, car, and ground, emphasizing texture and a sense of tension. The composition centers the car, with the men on either side, framed by desert hills in the background. The color scheme contrasts vibrant red with subdued desert tones, enhancing the crime film's tense mood. The shot is at eye level, with a medium depth of field, clearly capturing details of the men and car, while the distant hills blur naturally. The film quality has a cinematic grain, with sharp contrasts and subtle details, evoking a sense of tension, testing, and isolation.

Example by Kling AI · source ↗

The image features two men in ornate red uniforms with gold embroidery. The man on the left is bald with a beard, exuding authority, while the man on the right has gray hair and a calm demeanor. The left figure sits upright, gazing sternly off-camera, while the right leans back with a slightly tilted head, his expression contemplative. The setting is a formal indoor venue with rows of military or formally dressed figures and classical wooden furniture. The lighting is soft, backlit, creating highlights on their contours and clothing, with a dark, blurred background. The composition is symmetrically centered, with the figures occupying the left and center of the frame, creating depth with the background elements radiating outward. The color palette contrasts the vibrant red uniforms and gold embroidery with the muted greens and blacks of the background. The shot is slightly upward from eye level, emphasizing the figures' authority while maintaining the atmospheric depth. The focus is on the left figure's face and uniform details, with the right figure and background softened, guiding the viewer's gaze. The overall atmosphere is solemn and tense, with a soft film grain and natural light transitions, evoking both classical painting and documentary photography realism.

Example by Kling AI · source ↗

One-line edits that change only what you ask

Kling calls it feature retention: "Change the blue cup to a red one" and the rest of the picture stays untouched. Kling's own template is "Keep everything else unchanged, modify [target] to [desired effect]".

Transform to deep winter, heavy snow covering all surfaces, bare frozen trees, overcast sky, footprints in fresh powder, cold blue color grade

Example by fal.ai · source ↗ · This is an edit: add your own scene photo as a reference picture first.

Native 2K and 4K

Image 3.0 Omni "supports direct 2K and 4K ultra high definition output", so a poster or print does not need upscaling.

Up to 10 reference pictures

Omni and O1 combine up to 10 pictures: a person, a car, a jacket, shoes, even a dog, into one scene, and keep a face the same from any angle.

Cinematic post-apocalyptic landscape. Wide shot of the ruins of Times Square, New York, completely overgrown with massive tropical jungle plants and vines. A waterfall is cascading down a crumbled skyscraper. Sunset lighting, volumetric fog, birds flying in the distance, matte painting style, 8k, hyper-detailed.

Example by WaveSpeedAI · source ↗

Which Kling Image to pick

VersionPick it forSizesReferences
Image 3.0 OmniCinematic stills, series and storyboards, faces from any angle1K · 2K · 4KUp to 10
Image O1"What you envision is what you get": precise edits, many references1K · 2KUp to 10
Image 3.0Detail edits, style control, readable signs and captions1K · 2K1
Image 2.1The budget choice for simple text-to-picture1K · 2K1
Image 2.0 NewThe older model, for simple pictures1K1

Use Omni or O1 as soon as you add more than one reference picture. Image 3.0, 2.1 and 2.0 New take one.

How to prompt Kling Image

From Kling's official user guides.

  1. For edits: "Keep everything else unchanged, modify [target] to [desired effect]."
  2. For people: "Keep [name]'s facial features", then describe clothes, action, background, light and mood.
  3. Use clear position words: left, right, foreground, behind.
  4. Give each reference a job: this one is the subject, this one the place, this one the style.
  5. On Omni, speak film: composition, shot size, camera angle, lighting, tone. You can even ask for a focal length ("85mm").

Warm, detailed scenes

Extremely realistic image of a hobbit's study filled with maps and curiosities, roaring fireplace, rain against round windows, stacks of handwritten journals, warm tungsten lighting, inviting atmosphere. Hobbit is sitting on the chair.

Example by fal.ai · source ↗

FAQ

Is Kling Image the same as Kling video?

Same company, different models. This guide covers the picture models. Kling 3.0 video has its own guide.

Which Kling Image makes 4K?

Image 3.0 Omni. It makes 2K and 4K directly.

How many reference pictures can I use?

Up to 10 on 3.0 Omni and O1; one on 3.0, 2.1 and 2.0 New.

Do I need a Kling subscription?

No. Every version runs in the Lab and you pay per picture.