Skip to content
Let’s plan the right software for your processes. Call us for a demo or a quote: +90 546 737 48 29

TR EN DE

Creating images with AI

AI image tools generate new images from a description you write (a prompt) or edit an existing image. Image generation in ChatGPT, image generation in Google Gemini, Midjourney, Adobe Firefly and Canva's AI tools are widely used options in this field. They give quick results for presentations, social media, websites and concept sketches.

In brief

  • Write the prompt as subject, style, setting and lighting, composition and aspect ratio.
  • Start with a short prompt; rather than rewriting it, say what should change.
  • Text inside images may contain errors; add the wording in a design program.
  • For commercial use, read the tool's terms of use.

Note

This field changes quickly: the features, usage rights and terms of the tools are updated often. This guide gives general information as of October 2026; for details, see the official page of the tool you use.

On this page

How to generate a good image

  1. Choose the type of tool that suits your work

    Image generation built into chat assistants (ChatGPT, Gemini) suits writing requests conversationally and refining the result step by step. Tools dedicated to image generation, such as Midjourney, offer a large number of style and setting options. Adobe Firefly and Canva's AI tools sit inside design programs and are used alongside presentation, poster and social media design.

    Diagram of tool types: inside a chat assistant, dedicated image tools, inside a design program : Enlarge

  2. Write the prompt in parts

    A good image prompt contains these parts: the subject (what should be in the image), the style (photograph, illustration, 3D, watercolour), the setting and lighting (daylight, studio, night), the composition (close-up, wide angle, bird's-eye view) and the aspect ratio (square 1:1, landscape 16:9, portrait 9:16). You can also say what you do not want: "no text", "no people in the background".

    Diagram of an image prompt: subject, style, setting and lighting, composition, aspect ratio : Enlarge

  3. Start with a short prompt and assess the result

    Rather than writing a very long prompt on the first attempt, start with the subject and style, look at the result and add what is missing. Generating several options and choosing the best one is often quicker than searching for a single perfect prompt.

    Chat illustration: the user asks for a product photo and the assistant shows placeholders for the generated images : Enlarge

  4. Edit and improve

    If the result is not what you wanted, say what needs to change instead of rewriting the prompt from scratch: "make the background plain white", "make the lighting warmer", "move the product to the left". Many tools let you change only a selected part of the image, remove the background or extend the image. Text inside images may contain lettering errors; adding the wording afterwards in a design program is more reliable.

    Diagram of the editing flow: first draft, correction instruction, new version : Enlarge

Example prompts

  • Product photo: "A matt black flask standing on a white marble surface, soft studio lighting, plain light grey background, product photography style, square format."
  • Social media: "A young woman working on a laptop in a café in autumn, warm natural light, documentary photography style, portrait 9:16, no text."
  • Illustration: "An engineer holding a tablet next to a robotic arm in a factory, flat-colour vector illustration, navy blue and orange tones, landscape 16:9."
  • Presentation background: "An abstract dark navy background made of flowing lines, with empty space for text on the right, 16:9."

What to watch for with copyright and commercial use

  • Read the terms of use: Commercial usage rights may vary with the tool and the plan you use.
  • Real people, brands and logos: Do not generate the likeness of real people without permission; do not use other brands' logos or copyrighted characters.
  • An artist's style: Copying the style of a living artist by naming them can raise ethical and legal problems.
  • Do not mislead: Do not present an AI-generated image as a real photograph; in a product image, do not show the product as different from what it is. Some platforms require AI content to be labelled.
  • Copyright protection may be uncertain: The copyright status of AI-generated content differs from country to country; take legal advice for important commercial work.

Frequently asked questions

Can I use an image I generated as a logo?

Be careful: AI output may resemble an existing logo, and trade mark or copyright protection may be uncertain. It is safer to use it as a concept sketch and have the final logo prepared with a designer.

Why does the text in images come out garbled?

Although the tools are improving here, lettering errors can still occur. Ask for a short piece of text in quotation marks; if it does not come out properly, add the wording afterwards in a design program.

Can I have my own photo edited?

Many tools can edit an image you upload. Do not upload photos that belong to others, or in which other people appear, without permission, and check the tool's privacy terms.

BYK Yazılım Support Team
This guide is written and regularly reviewed by the BYK Yazılım support team. Last updated: 4 October 2026.

Related guides

Want to use AI with your own data?

BYK Yazılım builds AI assistants that connect to your business software, so you can ask questions about your production and order data in plain language.

Contact us AI integration