Tips on working with the AI prompt for visuals

Working with image generation

When it comes to image generation – and the vast majority of businesses use AI for generation over enhancement, working with an AI prompt is different to text-based prompt language. While in our module on text-based content, we looked at roles that the AI could play, a focus more on the audience or the type of output is required.

When it comes to prompting the AI to deliver you the best visual output, there are some other parameters that come into play. These are some elements you could consider to ensure your prompt is working hard for you.

• Subject. You need to have a protagonist, a centrepoint or a main focus. This could include a person, a place, a product or a brand.
• Perspective. Consider indicating the perspective you want the image to have. Options include “wide shot”, “close-up”, “drone view” or “satellite view.”
• Composition. Give the AI direction on how the image is composed. “The man is looking right” or “a tree to the left of the building.”
• Lighting and texture. Set the tone and ambience with lighting conditions. Imagine you are briefing a lighting director on a movie set. Use terms like “bright”, “heavy shadows”, “spotlight” or “overhead lighting.” WIth texture, include words that describe the scene or the background. “Wall” becomes “rough wall” or “fresh paint” becomes “glossy fresh paint”.
• Style/Medium. Incorporate a specific artistic style of medium, such as illustrations, photos, paintings or other styles you are specifically looking for. And go deeper by qualifying your preferred style – “oil painting” or “portrait photo”.
• Colour. By all means be clear about a “red car” or a “yellow taxi” but also use descriptive langauge to guide general colour use – “drab” or “vibrant” for example.
• Mood. Guide the AI with terms like “mysterious” or “relaxing”.
• Time period. Keywords like “prehistoric” or “futuristic” will help the AI produce the time period you need.
• Genre. If you have a particular genre in mind, include “fantasy” or “thriller”.

There are also other considerations which stretch the creativity further.

• Abstract concepts. AI is improving at interpreting abstract concepts, so use them as well. For example, you can request an image that conveys “the safety of home” or “the closeness of family”.
• What to leave out. The last consideration is what you DON’T put in. If you don’t want it in, say so. Language like “no people” or “no flowers” can help with an urban scene or a nature scene. “No logos” is good for corporate scenes.

Working with image enhancers

When it comes to using AI for enhancing images, it’s less about the prompt and more about the tools you use. It’s about selecting your preferred platform (or software) and then improving your imagery. If you are a rookie to image enhancement, some of the changes you can make to your images include:

• Image enhancement. The process of improving the quality, appearance, or specific attributes of an image, such as brightness, contrast, sharpness, and color.
• Resolution. The number of pixels (dots) per inch in an image, which affects its quality and detail. Higher resolution images generally result in better enhancements.
• Noise. The unwanted variations in brightness or color in an image. AI enhancers can help reduce noise and improve image clarity.
• Sharpness. Image sharpness refers to how well-defined and clear the details are in the image.
• Contrast. The difference in brightness between different parts of an image. Enhancing contrast can make images more visually appealing.
• Colour correction. The adjustment of colours to make them more accurate or visually pleasing. AI tools can correct colour balance.
• Artifacts. Unwanted visual anomalies that can appear in enhanced images, often caused by over-processing.
• Presets and filters. Many AI image enhancers offer pre-defined settings or filters that can be applied to images with one click.

Share your thoughts

How do you plan to use AI for image generation or enhancement?