Summary
This project began with a simple question: “Can we make prompts easier to create?”
Effectively using prompts requires the ability to express commands, imagery, and visual design concepts through language — a challenge for many creators.
However, with just a bit of guidance, these barriers can be easily lowered.
We defined commonly used prompts, camera techniques, and visual expression terms, and incorporated them into an intuitive UI. By following a simple manual, anyone can now create prompts with ease using this design tool.
Through this process, we gained insight into the evolving role and importance of communication within the creative field in the age of AI.
▲ Completed Prompt Prototype Version
Background
“Describing images through words is both subjective and objective. There exists a shared language for expressing colors, general adjectives to describe their moods, and even commonly recognized artistic techniques.
The real challenge, however, lies in translating these inherently subjective perceptions into objective verbal expressions.
Another challenge was the technical terminology itself.
For instance, as shown in the image below, we needed to explain complex commands such as –ar, –sref, and -personalize, along with various visual expressions, through language. To achieve this, a strong sense of creative communication—the ability to maintain visual consistency and clearly articulate the desired image—was essential.”
Another challenge was the technical terminology itself.
For instance, as shown in the image below, we needed to explain complex commands such as –ar, –sref, and -personalize, along with various visual expressions, through language. To achieve this, a strong sense of creative communication—the ability to maintain visual consistency and clearly articulate the desired image—was essential.”
Process 1. Creative Communication
What is creative communication?
For instance, designers often use descriptive language—such as references to color, texture, and form—to articulate their ideas when evaluating products or designs.
However, for those without a design background, such expressions may not come easily. We believed that the ability to describe images precisely is a key factor in effectively leveraging generative AI.
To explore this further, we began collecting key prompts used by designers and prompt specialists. These prompts were categorized and standardized into three main types: camera techniques, image descriptions, and style prompts that quantify visual style.
To explore this further, we began collecting key prompts used by designers and prompt specialists. These prompts were categorized and standardized into three main types: camera techniques, image descriptions, and style prompts that quantify visual style.
▲ Prompt Reference Table: Translating and explaining various style prompts used to convert images into text
Process 2. Establishing Prompt Benchmarks
“Although we had established clear descriptions and definitions for various prompts, it was necessary to understand how the output changes when these prompts are quantified.
For instance, when generating four images, the –Chaos value — which controls the level of variation and consistency — required an understanding of benchmark points such as minimum, maximum, and median values.
Through testing and fine-tuning these values, we identified the general range most commonly used. In particular, changes in the –Chaos value had a significant impact on the results. As shown in the example below, even small adjustments to prompt values can lead to substantial differences in image style and quality.”
Through testing and fine-tuning these values, we identified the general range most commonly used. In particular, changes in the –Chaos value had a significant impact on the results. As shown in the example below, even small adjustments to prompt values can lead to substantial differences in image style and quality.”
Setting the –Chaos value to 0 maintains consistency across the four generated images. Conversely, setting the –Chaos value to 100 removes that consistency, resulting in a wider variety of outputs.
However, all outputs still remain conceptually centered around the “New York City” prompt.
By analyzing these variable values across minimum (min), mid, and maximum (max) levels, we found that this method can significantly reduce the learning curve. We then conducted in-depth research and testing to further study its effects.
By analyzing these variable values across minimum (min), mid, and maximum (max) levels, we found that this method can significantly reduce the learning curve. We then conducted in-depth research and testing to further study its effects.
Process 3. Service Flow Design & UI Prototype Development
Building on the research and studies conducted so far, we designed a service flow that allows even beginners to easily create prompts.
Each section is structured into sequential steps, where users can select options through dropdown menus.
When an option is selected, the right panel displays a clear description and example, helping users easily understand complex terminology.
The core design principle was “as simple as possible.” To support intuitive understanding, we incorporated familiar examples—such as images from the Marvel movie Thor and short, concise summary phrases—to make the interface approachable and engaging.
The core design principle was “as simple as possible.” To support intuitive understanding, we incorporated familiar examples—such as images from the Marvel movie Thor and short, concise summary phrases—to make the interface approachable and engaging.
Additionally, the previously studied prompt databases were either transformed into selectable UI formats or visualized through designed UI screens to complete the process.
Additionally, the previously studied prompt databases were transformed into selectable UI formats, and corresponding UI screens were designed to complete the prototype.
Prototype Demonstration Video
Review
As we gathered, organized, and refined prompts scattered across the world, we came to realize that prompts might not be as difficult as they once seemed.
We made complex prompts easier for beginners to use and discovered that there is deep value in simplifying what once felt complicated.
Now, try creating your own prompts—more easily than ever before.
- Amorepacific Creatives
- 기획 및 디자인
- 김승진
