Home AI工具 AI指南 AI模型 AI创作者 🛒 购买 立即开始
🎬 KLING AI ⏱ 3 min read 🧩 元素

Elements — Technical Guide

训练 AI 永久识别您的角色、物体或服装——只需上传一次参考照片,然后在任何 Kling 视频或图像生成中按名称引用它们

🧩

Elements

klingai utility /app/elements →
训练 AI 永久识别您的角色、物体或服装——只需上传一次参考照片,然后在任何 Kling 视频或图像生成中按名称引用它们
元素是您的角色和物体库。您只需训练 AI 一次——上传几张人物、物体、服装或场景的照片——AI 就会学习它的外观。从那时起,您只需在提示中按名称引用该元素,AI 就会准确知道生成什么。

提供两种训练模式。图像参考允许您上传 1 张正面照片以及最多 3 个额外角度——AI 学习视觉外观并可以在任何生成中重现它。视频参考允许您上传一段短视频片段(3 到 8 秒)——AI 学习外观和运动风格,如果片段中的人说话,甚至可以自动提取声音。

一旦创建了元素,它可以在整个 Kling 生态系统中使用。在 Omni Image 中,在您的模板提示中将其引用为 <<<element_name>>>。在 Kling Video 中以相同方式引用。该元素在所有生成中保持一致的身份——相同的面孔、相同的比例、相同的独特特征——无需您再次上传照片。

标签让您按类型组织库:角色、动物、物品、服装、场景或效果。创建元素是免费的——仅在您使用它们进行图像或视频生成时才会产生费用。
✦ Best Results Tips
👤 清晰的正面照片至关重要
第一张图像必须是干净、光线良好的正面照片——面部直视相机,没有任何遮挡。这是 AI 用于学习身份的主要参考。其他所有内容都基于此图像。
📐 添加多个角度
上传最多 3 张来自不同角度的额外照片——侧面、三分之四视图、稍微高于或低于。更多角度让 AI 从各个方向更全面地理解面部和身体。
🎬 视频参考用于运动和声音
如果您希望 AI 学习角色的运动和声音,请使用视频参考模式。上传一段 3 到 8 秒的短片,展示该人说话或移动——AI 将外观、运动风格和声音一起提取。
🏷️ 使用描述性名称
给您的元素一个清晰、易记的名称——Maeva、红色裙子、海滩场景——而不是 element_001。您将在其他工具的提示中输入此名称,因此它应该易于记忆和输入。
♻️ 一次创建,处处使用
元素可以在 Omni Image 和 Kling Video 中使用。将您的角色训练为元素,您无需重新上传参考照片——只需在任何提示中输入 <<<element_name>>> 即可包含它们。
🆓 免费创建
创建和管理元素不需要任何费用。只有在您生成使用该元素的图像或视频时才会支付积分。在不花费任何费用的情况下构建您的整个角色库。

Elements — Available Models

Custom Elements
Default
custom-elements
Creates reusable character/object reference assets.
📥 You Give
📝Text Prompt 👤Face Reference Image 👕Clothing Reference (optional) 🎨Style Reference (optional) 🎬Video Reference (optional) 🔊Voice / Sound (optional)
AI Magic
klingai
🛠️ You Get
🛠️ Utility
Element types
Character Animal Item Costume Scene Effect
📌
Image reference (1 frontal + 3 angles)
Reference mode
📌
Video reference (3-8s, 1080p)
Reference mode
Features
Voice auto-extraction from video Persistent library Cross-tool references

💰 Elements — Pricing

Estimated cost
Failed jobs are automatically refunded
Elements
🛠️ Utility
Take full control of your video's composition. This guide shows how to use the "Elements" feature with 1-4 images for character consistency and multi-subject interactions.

The Elements feature is now available with the Kling AI 1.6 model for Image to Video generation! Upload 1–4 images, select the subjects (people, animals, objects, or scenes) in the images as elements, and describe their actions and interactions. A video will be created based on the elements and the prompt.

Compared with Text to Video generation, the Elements feature allows more creative control so that you can decide the elements to keep consistent in the video. Compared with the Frames feature, the Elements feature is more flexible with more references than the content in the Start / End Frame. With Elements, you can explore more possibilities by putting a subject in different scenes, or having multiple subjects interact with each other!

Using Elements for character consistency

Upload one or multiple images as elements of a subject (person, character, animal, object, etc.), and a video will be created based on the image references with a consistent style. This can be particularly useful to achieve consistent looks on a character across different shots.

In addition, you can also set scenes or clothing as elements, and specify the actions in the prompt, to make the subject look and move in a certain way, in a certain scene. This allows for more creative control in your work.

Elements

Result

Prompt: On the stage @element3, a girl wearing fashionable clothes @element1 and a crystal crown @element2 calmly gazes at the camera.

Elements

Result

Promt: A standing cat @element1 character wearing a jacket @element2 and sunglasses @elemen3 strikes a pose towards the camera on the stage.

Elements

Results

Prompt: A white Bichon Frisé @element1 wearing a red floral Chinese-style winter @element2 coat licks its paw.

Elements

Result

Prompt: In a café @element1, a cartoon-style elderly man @element2 lifts a cup @element3 to drink coffee.

Using Elements for interactions between characters

Upload images of multiple subjects (people, animals, objects, etc), and describe their interactions.

Elements

Result

Prompt: Two girls @element1 @element2 hug each other.

Elements

Result

Prompt: A boy @element# rides a Pegasus @element2 , soaring through the air, in a magical style.

Elements

Result

Prompt: A cartoon character @element1 wearing a white hat, and a cartoon-style bear @element2 , sitting side by side, wave and nudge at each other.

🧩 元素

试用 元素