AI creation overview
Use AI to generate images, videos, vectors, and 3D content directly in the editor.
What AI creation is
Lovision has a built-in AI generation system that lets you create and iterate on visual content without leaving the editor. You describe what you want, optionally provide references, and the AI places the result directly into your project as a layer you can continue editing.
What you can create
| Output type | Description |
|---|---|
| Image | Raster images from text prompts. Supports style references and canvas-element references. |
| Vector | Scalable vector graphics generated from a description. |
| Video | Short video clips from a text prompt or an image reference. |
| 3D | 3D object previews generated from a prompt or supported reference. |
Chart, slide, and website generation modes may appear in product planning or disabled tool entries, but they are not currently selectable creation modes in the editor.
How generation works
All AI creation tasks follow the same general flow:
- Open the prompt input — click the AI creation button in the toolbar or press the designated shortcut.
- Select the output type — choose Image, Video, Vector, or Create 3D from the available tool tabs.
- Write your prompt — describe the content you want. Be specific about style, composition, colors, and subject.
- Add references (optional) — attach image files or select existing canvas layers to guide the model's output.
- Choose a model — pick from the available models for the output type. Different models have different strengths and credit costs.
- Configure parameters — set size, aspect ratio, style strength, or model-specific options.
- Generate — click the Generate button. When the task completes, the result is placed on the canvas.
Editing generated output
Generated results are regular layers. After generation:
- Move, resize, and style the output like any other layer.
- Use Inpaint, Upscale, or Image adjustments to refine generated images.
- Delete and regenerate if the result doesn't meet expectations.
Credits
Each generation costs credits based on the model and output type. Video and 3D generation typically cost more than image generation. See Credits and concurrency limits for current pricing.
Tips
- A specific, detailed prompt produces better results than a vague one. Include subject, style, lighting, and composition.
- Use canvas-element references to match the style or content of existing design layers.
- Run multiple generations with the same prompt and pick the best — variation between runs is expected.
- Higher-quality models cost more credits but produce more consistent, detailed results.