Back to Home

Kling V2 New Image

Start from a prompt or translate one source image into a new visual treatment.

Create with Kling V2 New Image
Enter the translation desk
Translation desk

Keep the meaning. Translate the finish.

Treat the source image as a semantic anchor: name the content that must remain recognizable, then direct a new palette, material, medium or atmosphere without turning the task into a full redesign.

Restyling is not a full redraw: preserve the content boundary, then replace one visual layer.
SEMANTIC TRANSLATION / V2One source image, two layers handled separately

The workflow locks subject, pose and composition before translating palette, texture and light, then reviews the preserved and changed layers independently.

Matched portrait preserving the same face, pose and crop: natural photography on the left, with only colour, texture and medium translated into navy-and-orange screen print on the right
  1. KEEP
    Semantic anchorName what must remain recognizable

    Subject, pose, crop, object count and spatial relationships form the preservation boundary.

  2. TRANSLATE
    Visual translationChange only the named treatment

    Direct palette, material, medium, light or atmosphere as one clear treatment instead of redesigning everything.

  3. REVIEW
    Result reviewReview preservation and change separately

    Check that the subject remains recognizable, composition has not drifted and excluded elements did not appear.

Two input paths

Decide whether the brief needs a visual anchor

Upload a reference only when it owns the subject, composition or context. Start a new concept directly from structured text.

  • TEXT

    Generate a still from text

    Build a new visual direction from subject, action, environment, composition, light and exclusions.

    Generation mode includes a negative prompt
  • IMAGE

    Direct with one reference

    Let one image own the subject, composition or context, then state which visual layer may change.

    Current edit mode uses one reference image
Official 2.0 direction

Four strengths behind the 2.0 image direction

The official launch describes the 2.0 image generation model through semantic response, cinematic presentation, artistic style and practical controllable editing.

PROMPT

Stronger prompt adherence

Organize subject, action, composition, colour and exclusions as one clear brief so the generated still has a reviewable target.

CINEMA

Cinematic visual quality

Direct light, depth, framing and atmosphere toward film-poster, editorial and narrative-key-art treatments.

STYLE

Artistic style expression

Translate an existing image into another medium or visual language while asking its recognizable semantic content to remain.

EDIT

Controllable image editing

The official 2.0 launch covers redrawing, expansion, addition, modification and repair; the current Flux Art integration focuses on one-reference generation and restyling.

Capability boundary

Separate the official launch scope from the current workspace

Kuaishou's launch covers the broader 2.0 image generation and controllable-editing direction. Flux Art currently exposes a narrower text and one-reference entry, so this page does not present redrawing or expansion controls as available here.

Current integration

Flux Art workspace

  • Text generation and one-reference editing
  • 1K output
  • Eight fixed aspect ratios
  • Negative prompt in generation mode
  • Prompt length up to 2,500 characters
  • JPG and PNG reference images
Official launch

2.0 image direction

  • Stronger prompt adherence
  • Cinematic visual quality
  • Artistic style expression
  • Partial redrawing and expansion
  • Image addition, modification and repair
  • Restyling that retains semantic content
Prompt structures

Kling V2 New Image prompt structures

Use compact prompts that make the preservation boundary visible. One brief should answer what stays, what changes and how the image is framed.

Station attendant closing a glass waiting pavilion on a rain-wet elevated urban platform at blue hour
FRAME · Narrative key art

Cinematic still from text

A solitary station attendant closing a glass pavilion at blue hour, medium-wide eye-level frame, rain-dark stone, one warm interior light, quiet cinematic depth, restrained navy and amber palette, realistic materials, no signage, no extra people, 16:9.

Matched shoulder-up portrait of the same man in a camel trench coat, photographed on the left and screen-printed in indigo and orange on the right
STYLE · Art direction

Controlled editorial restyle

Keep the person's face, pose, crop and coat unchanged. Translate only the colour, texture and lighting into a refined screen-print editorial treatment with deep blue shadows, warm orange highlights and visible paper grain. Do not add text or props.

Saffron-yellow instant camera at a three-quarter angle on a pale mineral plinth with clean space on the left
PRODUCT · Campaign concept

Product atmosphere translation

Keep the product silhouette, controls and three-quarter camera angle from the reference. Replace only the setting with a pale mineral exhibition plinth and soft side light. Preserve scale and contact shadow; leave clean negative space on the left, 4:3.

Flux Art

How to use Kling V2 New Image in Flux Art

Kuaishou introduced its 2.0 image generation model around stronger prompt adherence, cinematic quality, artistic expression and controllable image editing. Flux Art currently exposes a focused 1K generation and single-reference workflow.

Create with Kling V2 New Image
  1. 01

    Choose text or one reference

    Start from text for a new still, or upload one image when subject, composition, palette or overall context should guide the result.

  2. 02

    Separate invariants from treatment

    List the subject, pose, crop and layout that must stay, then name the new medium, colour, texture, light or atmosphere.

  3. 03

    Set the frame and review the translation

    Choose one of eight fixed ratios at 1K, then compare recognizable content, composition and unwanted additions against the brief.

Source notes

Official sources and page information

Model capabilities are summarized from public Kling AI and Kuaishou materials. Available modes, parameters and pricing follow the current Flux Art workspace.

Published

Compare related Kling generation and editing models

Kling V2 New Image FAQ

What is Kling V2 New Image?

Kling V2 New Image is the Flux Art name for Kling AI's 2.0 image-generation direction, also associated with Kolors 2.0 and 可图 2.0. The official launch emphasized prompt adherence, cinematic presentation, artistic style expression and controllable image editing.

What can Kling V2 New Image do in Flux Art?

Flux Art currently exposes two focused workflows: generate a new still from a text prompt, or use one reference image to guide a new visual treatment. The options visible in the workspace are the current source of truth for this integration.

Can Kling V2 New Image generate an image from text?

Yes. Start in generation mode, describe the subject, action, composition, lighting, palette and finish, then choose a supported ratio. Add exclusions through the negative prompt when unwanted objects or visual traits need to be constrained.

How many reference images can I use?

The current Kling V2 New Image editing workflow in Flux Art accepts one reference image. Use that image as the anchor for subject, crop, layout or overall context, and write the requested visual changes separately.

What is the difference between generation and editing mode?

Generation mode creates a new image from text and supports a negative prompt. Editing mode starts from one uploaded JPG or PNG and uses the prompt to direct the new treatment while the reference supplies recognizable content and composition.

What resolution and aspect ratios are available?

Flux Art currently offers 1K output with eight fixed ratios: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3 and 21:9. Use the controls shown in the workspace as the current source of truth.

Does Kling V2 New Image support negative prompts?

Yes, in the current text-generation workflow. Use the negative prompt for unwanted subjects, text, logos, duplicated objects or visual defects. In reference-image editing, put the preservation and change boundaries in the main prompt.

How should I write a controlled restyle prompt?

Separate the brief into three parts: what must stay recognizable, what visual treatment should change and how the final frame should be composed. For example, preserve the face, pose and crop while changing only palette, texture and lighting.

How is Kling V2 New Image different from Kling V2.1 Image?

Kling V2 New Image is currently a focused 1K text-generation and single-reference workflow in Flux Art. Kling V2.1 Image adds 1K or 2K output and supports small multi-reference compositions with two to four images.

Can Kling V2 New Image output be used commercially?

Commercial use depends on the current Flux Art and Kling AI terms and on the rights attached to prompts and reference images. Before publishing client or advertising work, review the latest terms and confirm permission for people, brands, copyrighted characters and source assets.