> For the complete documentation index, see [llms.txt](https://docs.icraft.design/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.icraft.design/advanced-tips/use-ai-to-create-image-precisely.md).
# Use AI to Create Image Precisely
This section explains how to efficiently generate images that closely match your design intent in **iCraft** using the **AI Render Export** feature.\
By combining **geometric placeholders + color differentiation + precise prompts**, you can quickly create highly relevant scene images—even without advanced 3D modeling experience.
***
## Core Concept
AI image generation in **iCraft** is not based on pure imagination. Instead, it relies on the following inputs:
* The **current scene composition and camera view**
* The **position and scale of geometric objects in 3D space**
* **Color tags** representing different objects
* The **prompt** provided by the user
The system synthesizes all of this information to generate the final rendering.
Therefore, achieving **controllable, accurate, and expected results** does not depend on writing random descriptions. Instead, the key is:
👉 **Use geometry to build spatial structure, colors to create semantic labels, and prompts to map objects precisely.**
***
## Workflow
{% stepper %}
{% step %}
#### Build the Spatial Structure Using Geometric Shapes
In the editor, do not rush to use complex models. Instead, begin with **basic geometric shapes** (such as cubes, cylinders, and spheres). Place them according to **realistic proportions** to simulate objects such as walls, equipment, furniture, cabinets, pipelines. Make sure the **spatial relationships** (distance, height, orientation) are logical.
Additionally, it is recommended to use **colors as semantic labels**. Assign different colors to different geometric shapes so they can be distinguished in prompts. These colors are **not the final visual appearance**, but rather **visual hints for the AI**.
At this stage, focus on:
✅ Reasonable layout and accurate proportions\
✅ Clear composition\
✅ One distinct color per object category\
✅ Avoid similar colors (e.g., dark blue vs. light blue)\
✅ Prefer highly distinguishable color blocks
{% endstep %}
{% step %}
#### Define the Camera View and Composition
Before using the AI rendering function, make sure to:
* Create a **fixed camera view** in the camera panel
* Adjust **focal length and camera height**
* Choose the **best perspective**
* Ensure the **main subject is clear and unobstructed**
* Maintain a **balanced composition**
Recommendations:
* Use a **slight top-down perspective** to present the overall layout
* Avoid **extreme wide-angle lenses**
* Ensure **key objects are located near the center of the frame**
Once the composition is finalized, the AI image generation will use this view as its base.
{% endstep %}
{% step %}
#### Use AI Render Export
Click **AI Render Export** in the export panel to enter the image generation interface. You will need to enter a **prompt** in this interface. You can extend the preset prompt with additional custom requirements.
Example:
Select the preset **“Realistic Style”**, which automatically fills in the prompt:
> Realistic style, ultra-realistic rendering, 8K ultra-HD detail, HDR dynamic range imaging, micro-surface textures and reflective gloss, realistic depth of field, natural lighting and ambient light, accurate color reproduction, physically based lighting simulation, sharp edges and details, professional photography composition, minimalist aesthetics.
Then add additional prompt instructions:
> I want to generate this scene as a coffee bar corner.\
> The yellow cylinder represents a coffee machine.\
> Several blue cylinders represent a coffee grinder, coffee cups, and coffee accessories.\
> The green cubes represent storage cabinets and wine cabinets beside the bar counter.\
> The purple cubes represent decorative plants near the bar counter.\
> The brown cone above represents a warm yellow hanging lamp.\
> The overall scene should feel soft, relaxing, and lifestyle-oriented, with warm lighting, a gentle atmosphere, Instagram-style aesthetics, high-definition photography, and a cozy home environment.
Combined into the final prompt:
> I want to generate this scene as a coffee bar corner. Realistic style, ultra-realistic rendering, 8K ultra-HD detail, HDR dynamic range imaging. In the scene, the large black rectangular block represents the coffee bar table in a natural wood style. The yellow cylinder represents a coffee machine. Several blue cylinders represent a coffee grinder, coffee cups, and coffee accessories. The green cubes represent storage cabinets and wine cabinets beside the bar counter. The purple cubes represent decorative plants beside the bar counter. The brown cone above represents a warm yellow hanging lamp. The overall scene should feel soft and comforting, a lifestyle-oriented setting with warm lighting, a gentle atmosphere, Instagram-style aesthetics, high-definition photography, and a cozy home environment.
System will generate the image based on:
* the **current scene screenshot**
* **scene depth and geometric relationships**
* the **user prompt**
If you are not satisfied with the generated result, you can **regenerate the image** or **perform localized edits directly on the generated result**.
#### AI Generated Image:
{% endstep %}
{% endstepper %}
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.icraft.design/advanced-tips/use-ai-to-create-image-precisely.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.