Kling 4.0 is an AI video generation model developed by Kuaishou. Direct an anime scene with Kling 4.0: build a shot from text, start and end frames, or visual references, with 3–30 seconds at 720p or 1080p.
MODEL OVERVIEW
Direct an anime scene with Kling 4.0: build a shot from text, start and end frames, or visual references, with 3–30 seconds at 720p or 1080p.
Create with Kling 4.0
Describe the hairstyle, clothing and expression of one original character. Compare the face and silhouette with your design before choosing a variation.
In frame mode, upload a clear first image and optionally an ending image. Prepare JPG or PNG files up to 10 MB and crop the first image to your intended composition. Reference mode combines up to 10 images with one video of 2–30 seconds. Refer to them as @image_1 or @video_1 in your description. Reference video guides a new generation; this page does not directly edit the original clip. For EvoLink, the reference video must be at least 3 seconds. Upload the clip to this site before submitting.
Choose an integer duration from 3 to 30 seconds and 720p or 1080p. Text and reference workflows offer 16:9, 9:16, 1:1 and 21:9. With first/last frames, the first image determines the output shape. A three-second clip suits a single movement; leave more time for a camera move or developing action. This workspace does not offer 4K output.
CORE CAPABILITIES
Direct an anime scene with Kling 4.0: build a shot from text, start and end frames, or visual references, with 3–30 seconds at 720p or 1080p.
Start with a subject, setting, action and camera movement. A slow orbit around a ceramic vase or a tracking shot along a coastal road gives you a clear result to assess. Text-to-video needs no uploaded image. Use short action sequences for storyboards, product studies or social-video ideas; exact lettering and complex choreography may need another iteration.
In frame mode, upload a clear first image and optionally an ending image. Prepare JPG or PNG files up to 10 MB and crop the first image to your intended composition. Reference mode combines up to 10 images with one video of 2–30 seconds. Refer to them as @image_1 or @video_1 in your description. Reference video guides a new generation; this page does not directly edit the original clip. For EvoLink, the reference video must be at least 3 seconds. Upload the clip to this site before submitting.
Choose an integer duration from 3 to 30 seconds and 720p or 1080p. Text and reference workflows offer 16:9, 9:16, 1:1 and 21:9. With first/last frames, the first image determines the output shape. A three-second clip suits a single movement; leave more time for a camera move or developing action. This workspace does not offer 4K output.
Sign in, select text, frames or references, and describe the motion and relevant sound. Confirm your settings and estimate, then submit once and follow the task status. The model can generate native audio; a separate sound switch, voice upload, intermediate keyframe editor and reusable character library are not available here. Add exact titles, logos and final cuts in your editing software.
LIVE SETTINGS
These values come from the currently configured Kling 4.0 route, so this table stays aligned with what the generator can actually submit.
HOW TO USE IT
Direct an anime scene with Kling 4.0: build a shot from text, start and end frames, or visual references, with 3–30 seconds at 720p or 1080p.
Start with a subject, setting, action and camera movement. A slow orbit around a ceramic vase or a tracking shot along a coastal road gives you a clear result to assess. Text-to-video needs no uploaded image. Use short action sequences for storyboards, product studies or social-video ideas; exact lettering and complex choreography may need another iteration.
Choose an integer duration from 3 to 30 seconds and 720p or 1080p. Text and reference workflows offer 16:9, 9:16, 1:1 and 21:9. With first/last frames, the first image determines the output shape. A three-second clip suits a single movement; leave more time for a camera move or developing action. This workspace does not offer 4K output.
Watch the entire result and listen to the audio. Compare faces, product shapes, reference fidelity, motion continuity and unwanted text across the full timeline. Use the result download control and open the saved file before placing it in an edit. Generation time and visual accuracy vary. Use references you have permission to process and check the applicable service and model terms before publication.
BEST USE CASES
Anime & Manga
Describe the hairstyle, clothing and expression of one original character. Compare the face and silhouette with your design before choosing a variation.
Build a rain-soaked street or a quiet studio around a readable subject. Keep the light direction and palette consistent with the adjacent panels.
Give each composition one narrative purpose. Review the framing and the position of the subject before assembling it with other panels or shots.
Plan wide or vertical framing for the intended screen. Add exact captions in an editor and review the downloaded file before sharing.
PROMPT GUIDE
Give each composition one narrative purpose. Review the framing and the position of the subject before assembling it with other panels or shots.
Create with Kling 4.0Describe the hairstyle, clothing and expression of one original character. Compare the face and silhouette with your design before choosing a variation.
Build a rain-soaked street or a quiet studio around a readable subject. Keep the light direction and palette consistent with the adjacent panels.
Give each composition one narrative purpose. Review the framing and the position of the subject before assembling it with other panels or shots.
PROMPT EXAMPLES
Anime & Manga
An original manga heroine in a quiet atelier, navy coat, soft window light, clear silhouette, delicate watercolor background.
A fox spirit under a cherry tree at dusk, restrained pink petals, cool shadows, a simple readable composition.
An anime city after rain, reflections on the pavement, one cyclist in the foreground, clean perspective and cinematic framing.
WHAT TO REVIEW
Plan wide or vertical framing for the intended screen. Add exact captions in an editor and review the downloaded file before sharing.
In frame mode, upload a clear first image and optionally an ending image. Prepare JPG or PNG files up to 10 MB and crop the first image to your intended composition. Reference mode combines up to 10 images with one video of 2–30 seconds. Refer to them as @image_1 or @video_1 in your description. Reference video guides a new generation; this page does not directly edit the original clip. For EvoLink, the reference video must be at least 3 seconds. Upload the clip to this site before submitting.
With KIE selected, a five-second clip without video references costs 171 credits at 720p or 255 at 1080p. A video reference adds billable input time: output duration plus the greater of input duration and output duration. A five-second output with a three-second reference costs 204 credits at 720p. Changing provider, resolution or duration changes the estimate; check it before submitting.
Watch the entire result and listen to the audio. Compare faces, product shapes, reference fidelity, motion continuity and unwanted text across the full timeline. Use the result download control and open the saved file before placing it in an edit. Generation time and visual accuracy vary. Use references you have permission to process and check the applicable service and model terms before publication.
