In brief
With the right prompt, a good tool, and a few smart habits, you can create usable AI images for social media, campaigns, or internal presentations. The most well-known tools are ChatGPT (DALL-E), Gemini, and Midjourney. But remember, AI is a starting point; a human remains the final editor.
What can AI do with visuals?
Let’s start at the beginning. In this article, we focus mainly on the most well-known and widely-used application of visuals: creating and editing new images. With generative AI, you can create new images based on a text description. Think of mood images for social media, posters, logos, stage design concept sketches, or an initial visualization of an idea. You can use this directly, but it can also serve as a great starting point for a discussion or briefing with an illustrator or photographer.
Using the same method, you can also modify existing images or photos, remove backgrounds, enhance quality, add objects, or extend images to a different format.
But these are certainly not the only applications: AI can also help manage collections and make visuals more accessible, but we’ll delve into that further at the end of this article.
Which AI visual tools are available?
The next logical question is: where should I go? Which tools can I use for this? There are quite a few now. Essentially, they do the same thing, but they differ in quality and functionalities.
Here, we discuss the most well-known and widely-used tools because they deliver high-quality results. At the time of writing (spring 2026), there are no major European AI visual tools. However, it is possible to create visuals using local open-source models. This is more complex and requires significant computing power.
The most well-known AI visual tools:
ChatGPT (OpenAI): generates images via DALL-E. The more detailed the prompt, the better the result. Since April 2026, a new AI model is used that requires less complex prompting for good results. You can adjust the entire image or make a selection for more precise results using prompts.
Gemini (Google): generates and edits images via Nano Banana. This creates high-quality images. The difference with DALL-E: you don’t need to manually select anything; you describe what needs to change, and Nano Banana adjusts only that specific part.
Copilot (Microsoft) and Claude (Anthropic): generate simple images or infographics and offer basic edits via text instructions, such as background removal or color adjustments. However, the quality is less good than the other two tools.
The above options are all good tools for gaining initial experience. Want higher quality or more control over style? Then there are specialized visual tools:
Midjourney: delivers consistent, high-quality visuals in various styles, from photorealistic to artistic. This tool is developed for and with creative creators. Midjourney also has its own prompt syntax with parameters, which requires more practice than the other tools.
Adobe Firefly: integrated into Photoshop and Illustrator. For targeted and creative edits with high quality. This AI model is trained on licensed images, making it safe for commercial use.
Canva AI: combines image generation with (your own) design. Useful if you want to generate an image and immediately use it in a poster, social post, or presentation. Generated elements remain editable as separate layers.
Stable Diffusion by Stability.ai: for advanced users, the AI visual models from stability.ai are an interesting option. These language models can be used locally and customized to your preference.
Tips for working with AI and visuals
Now that you know this, we can share some tips. These 6 handy tricks will help you achieve better results with your images.
Start with one application
Choose one specific task for which you want to use AI, such as generating a mood image for social media, an infographic for employees, or a sketch of theater decor. This way, you get to know the tool without feeling overwhelmed. Experiment and then expand.
Let AI write your prompt
A good image prompt is half the battle, but describing the image in your head precisely can be difficult. Explain in plain language what you want to achieve. Then ask for a ready-made image prompt. Or ask the tool to let you ask questions to achieve a better result. This yields an even better outcome. Also, ask for multiple variations, such as one more photographic and one more illustrative. Already have an initial image but want to refine it? Let AI adjust the prompt based on what you like and dislike.
Send an image as a style reference
Most chat tools accept not only text but also an image as input. This way, you use existing visual material, such as a previous campaign photo, your style guide, or an inspiration image, as a starting point.
Upload a photo or design and ask, for example:
„Generate a new image in the same color scheme as this image.”
„Create a campaign image in the mood and style of this photo, but for an outdoor concert.”
„This is our logo. Generate a background image that fits this style.”
The result will align much better with your organization’s visual identity without you having to capture everything in words.
Be mindful of copyrighted work. Do you want to upload this into an AI model or not? Legally, it’s a gray area; make a conscious choice.
Describe mood and purpose, not just the subject
The more specific your prompt, the better the result. Describe not only what you want to see but also the mood, style, and purpose. “A warm, inviting photo of a theater hall for social media” yields better results than simply “theater hall.” Also, mention the format if it’s relevant: landscape, portrait, or square. For photorealistic images, it helps to include aperture, shutter speed, and ISO values.
Use AI images as a starting point
A generated image is usually a first draft. Use it to explore an idea, clarify something internally, or brief a creator. An illustrator or photographer adds context, quality, and intent that an AI model does not have.
Note: Using AI comes at the expense of creators’ work. Make thoughtful decisions as an organization on how you use AI.
Save your best prompts
Have a prompt that creates the desired style or image? Save it centrally and share it with colleagues. This way, you create visuals in the same style and with consistent visual language. Your colleagues won’t have to reinvent the wheel, and you reduce your environmental impact.
Responsible and wise use
Using AI for visuals comes with several important concerns and considerations. You’ve already encountered some above. Other considerations for visuals include:
Environmental impact: image generation is significantly more energy-intensive than text generation. Use AI visual tools consciously: not for every little thing, but where it truly adds value.
Transparency: Make agreements about what you disclose when posting or using AI-generated images. For deepfakes, it’s mandatory to disclose they were created by AI (European AI Act) (opens in new tab).
Bias: check whether your images are representative and inclusive. AI models reproduce existing patterns and stereotypes from training data. Ask yourself three questions: who is central in the image, what background or context is shown, and which groups are missing? These three questions quickly reveal blind spots.
Make agreements about this within your organization in an AI policy. Our AI policy step-by-step guide or the workshop on AI policy and ethics can help you get started!
In conclusion
AI makes creating and editing visuals highly accessible. With the right tools, clear agreements, and a critical eye, you can use AI responsibly for visuals. At DEN, we don’t see it as a replacement for creativity but as a tool. You remain the editor: you decide what works, what fits your organization, and what resonates with your audience.
Can I use AI-generated images commercially?Show explanationHide explanation
That depends on the tool. Adobe Firefly is trained on licensed images and explicitly safe for commercial use. Tools like Midjourney and DALL-E have additional usage terms — check them per tool before using images commercially.
Are AI-generated images copyright protected?Show explanationHide explanation
In most European countries, fully AI-generated images without human creative input are not copyright protected. If you substantially modify the image or give it a clear artistic direction, copyright may apply. Consult a lawyer if in doubt.
How do I prevent AI images from reinforcing stereotypes?Show explanationHide explanation
Be explicit in your prompt about diversity and representation. Describe not only the subject but also who is depicted. Always check the result before publishing — AI models reproduce existing patterns, requiring a critical eye.
What if a generated image doesn’t look the way I want?Show explanationHide explanation
Adjust the prompt: add more context about mood, style, and purpose, or send a reference image. You can also ask for multiple variations at once or ask the tool what’s missing in your description.
More articles per topic
New(s) on DEN
Foto's en filmpjes van mensen online delen: wat mag wel, wat mag niet?
Vacature: Interim adviseur IT-infrastructuur en Verandermanagement
Digitization, digitalization, and digital transformation: what do they mean and what are the differences?
Digital transformation: what is it and what can I achieve with it?
