Kling V4 Video Generator

Overview

Kling V4 is a video generation model from KlingAI for text-to-video and image-to-video. You describe a scene, actions, camera movement, and sound, then the model renders a short MP4 clip. If you provide keyframe images, the model uses them as visual anchors and fills in motion between them. This helps you lock character look, scene layout, and key story beats.

What you can build

• Product clips with controlled camera moves, like push-ins, pans, and reveal shots
• Character reaction shots that follow a provided start frame and planned ending frame
• Short ad variations where you keep the same keyframes and swap copy in the prompt
• Storyboard-style beats using multiple keyframes to force specific moments on the timeline
• Social videos in common aspect ratios like 9:16, 1:1, and 16:9

Inputs

• A written prompt that describes the subject, environment, motion, camera direction, and sound cues. For pure text-to-video runs, the prompt drives the whole clip. For keyframe runs, the prompt explains what should change between frames.
• An optional first keyframe image to start from for image-to-video. If you upload a first frame, you can reference it inside the prompt using the platform’s image placeholder format.
• Optional middle keyframe images, provided in timeline order. These let you force specific moments partway through the clip. The run supports up to 10 keyframe images total across first, middle, and last.
• An optional last keyframe image to steer the ending. This requires a first frame.
• A target duration you choose between 5 and 30 seconds.
• A target output resolution of 720p or 1080p.
• An optional aspect ratio choice for text-to-video. For image-to-video, the aspect ratio follows the uploaded keyframes.
• An optional watermark toggle to add a visible watermark to the result.

Image guidance that affects results: - Use clear, sharp images. Avoid heavy compression artifacts.

• Prefer images at least 300×300 pixels.
• Keep each image file under 30 MB.

Outputs

The model returns a single video file: - MP4 video output.

• The clip length matches the duration you selected.
• The frame size matches 720p or 1080p.
• The video can include an audio track when your prompt includes dialogue, ambience, or sound directions.

Limitations

• Clip length is limited to 5–30 seconds per generation in this Wiro model run.
• Output resolution is limited to 720p or 1080p in this Wiro model run.
• Keyframe control is capped at 10 total images across the timeline.
• The model may drift on fine details across time, like small logos, exact text kerning, or jewelry.
• Busy scenes with many faces, hands, or fast action can cause distortions or identity drift.
• Low-quality inputs increase failure risk. Blurry images, heavy JPEG artifacts, screenshots with UI chrome, and scanned or upscaled images can reduce consistency.

Safety & compliance

• Don’t generate sexual content involving minors, or any sexualized depiction of a minor.
• Don’t generate pornography or explicit sexual acts.
• Don’t request or generate hate, harassment, or derogatory content toward protected groups.
• Don’t request or generate terrorism or violent extremist propaganda.
• Don’t generate content that promotes illegal wrongdoing or provides instructions for harm.
• Respect rights and consent. Don’t use real people’s likeness or voice without permission.
• Don’t upload confidential information or trade secrets in prompts or reference images.
• Avoid infringing content. Don’t copy or recreate copyrighted or trademarked works when you don’t have rights.