Gordon Napier
Methods for generating images
Three reliable ways to make a poster, infographic, or illustration with GenAI, focusing on tools within the Adobe suite (available RMIT-wide)
How to structure a text-to-image prompt in Adobe Firefly to generate specific and high-quality images
Generative AI can create images from text, but controlling the output to match your specific vision requires a structured approach. Prompt fluency—the skill of blending clarity, emotion, and intent in your instructions—is crucial for reducing friction and accelerating creative execution. By using a layered, descriptive formula, you can guide the AI to produce visuals that more accurately reflect your unique style and conceptual goals, making it a powerful tool for creating custom educational materials or visual aids.
00:53Start with a basic three-part prompt. For a simple image, structure your prompt to include the [subject], a [description of how it should look], and the [context or style]. For example:a medieval castlesurrounded by a moat of blue waterbeside a mountain range of trees.01:22For more control, use a detailed prompt formula. Build a more descriptive and layered prompt by including several of the following components:Image type: e.g.,photo,illustration,panoramicSubject: The main focus, e.g.,a striking male modelAction: What the subject is doing, e.g.,sits on a chair and looks sidewaysAngle: The framing of the shot, e.g.,medium close-up shot,three-quarterLighting: e.g.,soft studio lighting,golden hourBackground: The setting, e.g.,neutral gray background with subtle abstract floral patternsColor palette: e.g.,monochromatic palette with accents of white and blackStyle: The overall aesthetic, e.g.,modern, hyper-realistic photographyOptional details: e.g.,camera lens,depth of field,texture,mood02:27Iterate on your prompt by adding layers of detail. Start with a simple prompt (e.g.,portrait of a man) and progressively add more descriptive elements from the formula (e.g.,high-fashion editorial portrait of a male model in an avant-garde blazer) until the output matches your vision.03:22Refine the output using Firefly's controls. Beyond the text prompt, use the controls in the Firefly interface to apply specific styles, effects, or reference images to further guide the generation process.03:38Optionally, enable the Prompt Enhancement feature. Turn on the 'Prompt enhancement' toggle to allow Firefly to automatically refine your prompt in the background, making it more detailed and descriptive to generate richer images.
Tool | This activity uses Adobe Firefly. |
|---|---|
Access | Adobe Firefly requires an Adobe account; usage may be subject to credits or subscription limits. |
Prompt Formula | A structured prompt can include: Image type, Subject, Action, Angle, Lighting, Background, Color palette, Style, and Optional details. |
Example Prompt | The presenter used the detailed prompt: "A high-fashion black-and-white editorial portrait. A striking male model with a sculptural hairstyle made entirely of colorful yarn, in an avant-garde blazer split into black and white halves, and a high black turtleneck. Sits on a chair and looks sideways. Medium close-up shot, three-quarter. Soft studio lighting. Neutral gray background with subtle abstract floral patterns. Monochromatic palette with accents of white and black. Modern, hyper-realistic photography." |
Caveat | Prompting is an iterative creative process; it is normal to refine your prompt several times to achieve the desired result. |
Related skills
- Refine and iterate prompts — This activity is centered on the skill of refining prompts, moving from a simple idea to a highly detailed set of instructions to control the AI's output.
- Use simple prompts — The tutorial begins by demonstrating how to construct a basic, effective prompt before introducing more complex structures.
- Create multimodal outputs — This activity involves translating a text-based vision (the prompt) into a visual medium (the image), which is a core multimodal skill.
- Apply human judgement to validate AI outputs — The process of iterating and refining prompts requires the user to continually apply their own creative and critical judgment to evaluate how well the AI's output matches their intent.