> For the complete documentation index, see [llms.txt](https://docs.d5render.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.d5render.com/d5-lite/lite-user-manual/ai-generation.md).

# AI Generation

<figure><img src="https://d5.feishu.cn/space/api/box/stream/download/asynccode/?code=MTE2YzkzNjZlN2Y5YWUzOTc4Y2FjYTRhNDQwYjllNDVfUUJsZjJPSklMNVhYY2trVEJzaXJwZ05XSGExWlVvRlZfVG9rZW46TkMyZGJuWEZMb0JTclR4TFljRGNnZ3g3bnRiXzE3ODMzMzEwOTM6MTc4MzMzNDY5M19WNA&#x26;add_watermark=true&#x26;scene_type=CCM" alt=""><figcaption></figcaption></figure>

Besides real-time rendering, D5 Lite can use AI to quickly generate stylized scheme images from your current model — useful for exploring ideas and trying styles early in design. Use the button in the upper-left corner to switch between Render mode and AI mode.

### Getting a Base Image

* Capture from the modeling software: Lite captures the current view when it opens; to change the view, click the capture button in the lower-left corner to recapture.
* From Render mode: when you switch from real-time rendering to AI mode, Lite uses the current render as the base image and continues generating from it.

### Controlling the Result

* Style Presets: built-in categories for architecture, landscape, and interior, each with finer project types and styles.

<figure><img src="https://d5.feishu.cn/space/api/box/stream/download/asynccode/?code=MDlmMTcxMGIyNmZkNTc0ZTc5MTQ4N2ExOTZlOTdhMzRfYUJ3OHBGeThqWkQ4VTFqQXdNTTZVOHBaYm5HMDdPaDBfVG9rZW46QnkySWJzcUx1b3hwaDh4QVdPOWN0R2FsbjBsXzE3ODMzMzEwOTM6MTc4MzMzNDY5M19WNA&#x26;add_watermark=true&#x26;scene_type=CCM" alt=""><figcaption></figcaption></figure>

* Structure Weight: controls how closely the result follows the structure of the original. Lean toward Creative for more freedom (good for early ideation), or toward Accurate to stay closer to the original's lines and forms.
* Prompt: text describing what you want, up to 300 characters.
* Reference Image: upload one reference image (up to 50MB) for the AI to learn its style, color, and mood.
* Generation Settings: image quality (Standard / High), number of images (up to 4 at once), and seed.

| <img src="/files/JyzZTpXYTnGwLOvEO9LN" alt="" data-size="original"> | <img src="/files/05On24QAx3jrfMXp8tIz" alt="" data-size="original"> |
| ------------------------------------------------------------------- | ------------------------------------------------------------------- |

### Generate, Regenerate, and Iterate

* **Generate**: create the chosen number of new images from the current settings.
* Regenerate: use when you want a different result or more variation; the input image stays the same and parameters can change.
* Iterate: use when the result is mostly right and only details need tweaking; keep adjusting on top of the generated image.

<figure><img src="https://d5.feishu.cn/space/api/box/stream/download/asynccode/?code=OTQ2NmQ2NzYyNzQ0ZmNkMzExZTBjYTA5ZTg2NTllZmJfNzliRnhSQUY0cTltc3hKN1Jyb0VhMnhPVldwbnFhVUJfVG9rZW46QUs1WWJYS1VJb3ZJZ1V4VzJxVWNOZDNGbk9kXzE3ODMzMzEwOTM6MTc4MzMzNDY5M19WNA&#x26;add_watermark=true&#x26;scene_type=CCM" alt=""><figcaption></figcaption></figure>

### History

Every image you generate is kept in History. There you can review and compare different batches, pick a result to keep using, or iterate on it.

> Privacy: we do not use your AI interactions to train our models.

<br>
