Generating Images with AI

You can ask the AI to create images directly in chat — for example, illustrations for presentations, symbolic images for internal communications, or drafts for event posters. Simply describe what you would like in your message.

Availability

Image generation is only available if your administrator has enabled an image generation model for your organization. If it’s not available, please contact your administrator.

How to Generate an Image

1. Start a new chat or open an existing one
2. Describe in your message what image you would like
3. Send the message — the AI will create the image and display it in the chat
4. Download the image by right-clicking on it and selecting “Save image as…”

Example Requests

“Create an illustration of a city park with a playground and walking paths, in the style of a friendly watercolor drawing.”

“Generate a symbolic image for an information brochure about waste separation, factual and in our administration’s colors (blue and green).”

“Create a cover image for an invitation to a town hall meeting, showing a city hall with several people in front of it, in cartoon style.”

Limited Number of Image Generations

So that all staff in your organization can use image generation equally, the number of images you can create within a certain period is limited. You therefore cannot generate an unlimited number of images one after another.

When you reach the limit, you will receive a message in the chat. It tells you when you can generate images again — usually after a few minutes to hours. Once the time has passed, you can simply continue without doing anything further.

Tip

Even while you are waiting for the limit to reset, you can continue to use the chat normally for other tasks — for example, writing texts, summarizing documents, or asking questions.

Tips for Good Results

The quality of the generated images depends heavily on your description. With these tips, you’ll achieve significantly better results:

• Be specific – Describe not only the content but also the style. “A photorealistic shot of a desk with file folders” yields better results than “A desk”.
• Name the style and type – Indicate whether it should be a photograph, illustration, sketch, watercolor, cartoon, or symbolic image.
• State mood and colors – “Friendly and inviting in warm earth tones” or “factual in shades of blue and grey” helps the AI capture the character of the image.
• Describe the framing – Explain what should be in the foreground, background, and at the edges.
• Mention the purpose – If you mention what you need the image for (e.g., “for a presentation”, “as a cover image for a flyer”), the AI can better adapt the format and composition.
• State what should not be in the image – You can also tell the AI what to leave out, e.g., “without text”, “without people in the foreground”.
• Plan multiple attempts – If the first result doesn’t fit, ask the AI for an adjustment (“Make the image brighter”, “Show the people from the side”) or rephrase your request.

Important: Limits of Image Generation

• Text in images is unreliable – Lettering, labels, or logos are often rendered incorrectly. Use a graphics program for these instead.
• No real people – Do not generate images that are intended to show real, identifiable people — this is legally problematic and generally not permitted in public administration.
• Check copyright – Before publication, clarify whether the image may be used for your purpose (e.g., website, press release). The legal regulations regarding AI-generated images are still evolving.
• Review the content – AI images occasionally contain errors (e.g., too many fingers, distorted objects). Review the image before using it.

Next Steps

• Documents, Images, and Knowledge Collections – Upload images into the chat and have them analyzed
• Prompting: Basics & Best Practices – How to formulate better requests