The Importance of Clear AI Prompts
In a coloring book on the topic of “Water Babies,” research revealed that the term originated and gained widespread recognition through Charles Kingsley's The Water-Babies, A Fairy Tale for a Land Baby, published in 1863. Research also revealed that it is related to folklore and mythology in certain Native American legends, where "water babies" are thought of as supernatural beings associated with water, often described as mischievous or ghostly infants. Nevertheless, the stories vary by tribe; some view them as protective spirits, while others see them as ominous entities. For example, in the Western United States, water babies are sometimes said to haunt lakes or springs. In modern usage, the term is often used in marketing for swimming lessons or water-related products for infants and toddlers. Each context gives "water babies" its unique flavor, ranging from literary imagination to cultural legend and everyday expression.
Based on expert knowledge and advice from AI's ChatGPT, Photoshop's “Generative Fill” and “Generate Image” features were used to create the book’s illustrations, making it essential to input clear and creative prompts that capture the different aspects of the theme of water babies. While AI's interpretation of the given input may not be 100% accurate, with several requests for continued generation assistance, one can arrive at an image that can eventually be edited for the desired outcome.
Below are a few prompts used to generate diverse and engaging illustrations on the topic of water babies using Photoshop’s Generative Fill and Generate Image:
Fantastical Underwater Scenes
Prompt: A group of playful water babies swimming together, surrounded by shimmering bubbles and soft sunlight filtering through the water.
Whimsical Creatures and Water Babies
Prompt: Black and white line illustrations for kids coloring, showing charming water babies with seaweed crowns swimming along with dolphins and seahorses in an ocean setting.
Native American Folklore-Inspired Imagery
Prompt: In the style of hand-drawn line art for coloring books, create a mystical lake surrounded by tall pine trees, with water babies in the water, observing the full moon and light filtering through the water.
Seasonal and Festive Scenes
Prompt: Water babies celebrating a midsummer festival underwater, with swimming fish, glowing lanterns, floating flowers, and gentle waves
Each context gives "water babies" its unique flavor, ranging from literary imagination to cultural legend and everyday expression.
Student Work
AI work in a Beginning Level Computer Applications Course
With most of the students pursuing a major in Digital Animation and their familiarity with gaming, the idea was to create an assignment that was both meaningful and creative. The tasks were to combine Illustrator for robot design and Photoshop’s Generative Fill for experimenting with cityscape backgrounds for their robot to inhabit. This approach turned out to be an excellent way to showcase the power of blending traditional design skills with AI in my online classroom. It provided students with a real sense of creative control — they could design their character from scratch and then instantly transport it into dynamic, AI-generated environments. Additionally, experimenting with different cityscapes added layers to their storytelling, making the robot's role as a "guardian" feel more immersive.
This exercise also demonstrates how AI can quickly bring various design scenarios to life, allowing students to explore different aesthetics and atmospheres without manually constructing every detail. It’s a practical, hands-on way to examine not only the visual enhancements AI can provide but also how it can enrich narrative through setting and mood.
The students gained a lot from this exercise, clarifying my goal of creating a balanced approach to incorporation AI into teaching design. Further experimentation involved students manually adjusting the robot’s posture or expression to align with different cityscapes, illustrating how AI-enhanced backgrounds can influence character design choices.
The Design Process
Adobe Illustrator was utilized to develop a sketch and then digitally develop the robot using the Gradient Mesh tool. Adobe Photoshop can also be employed for robot development, with its AI Generative Fill feature used for exploring the cityscape background. However, using Adobe Illustrator was better for scaling the robots without losing quality or having to be concerned with image resolution.
Students brainstormed, researched, and sketched several concepts for the cityscape background. Once they settled on an idea, they contemplated and explored color palettes, background textures, and layouts. Students then experimented with AI “prompting” to generate these color palettes and background textures, reinforcing the notion that AI assists rather than leads.
By exploring the critical skill of “prompting,” students were able to directly influence the quality and relevance of the output they received, resulting in several visually stunning AI-generated cityscape backgrounds. This proved to be a valuable learning exercise, as prompting must be very specific and parallels the design brief, where brainstorming, thumbnails, and iterations are essential to the process.
Some students even engaged in reverse-engineering by generating a few AI-created visuals and then attempting to sketch similar concepts by hand. This allowed them to explore which elements they could replicate manually, thereby reinforcing their understanding of composition and form. Here are two students explorations:
Student AI Project Highlight: David Chau
Introducing the ANGEL M-7—a robotic sentinel designed to watch over a futuristic cityscape. According to Chau, the robot’s job would be to make sure that people are where they are supposed to be, doing what they are supposed to be doing, and that no outsiders entered the city. The robot was duplicated and scaled several times and strategically placed within the cityscape for the desired visual effect. With Generative Fill enhancing the dynamic background, this project highlights the power of combining AI tools with hand-rendered designs to create immersive visual stories. The robot guard, Angel M-7, was created in Adobe Illustrator from a hand-sketched template and developed using the Gradient Mesh tool to achieve the 3-D look.
Student Project Highlight: Isabella Rachel
Introducing Botan X — “Green” robots designed to watch over a futuristic cityscape called Death’s Door. This design was crafted by a student in our Computer Applications course, where creativity meets cutting-edge technology. With Generative Fill enhancing the dynamic background, this project highlights the power of combining AI tools with hand-rendered designs to create immersive visual stories. The student chose to create a “green” robot whose tasks were to keep the city lively, and plant seeds and other vegetation around the city. The robots can be seen roaming the streets and sidewalks, planting seeds and various vegetation around the city. They can also collect trash throughout the city. The main robot would then come through to inspect the work that was done According to Rachel, the significance of this name is the fact that the city itself is completely the opposite of what the name entails. Instead of being a dark and scary city like you would think, it is full of life and green/earthy but still futuristic.
NOTE:
Photoshop’s “Generative Fill” and “Generate Image” is only a starting point in the classroom. Sometimes, students using different computer systems produce varying results with the same AI prompts. When this occurs, it often comes down to differences in several key factors:
AI Model Variations: Different systems may use distinct versions of an AI model (e.g., GPT, DALL·E, or others) with varying capabilities or training datasets. Even small differences in model versions can lead to variations in output quality, style, and relevance.
System-Specific Settings: AI platforms often allow customization, such as adjusting parameters like temperature (creativity level), response length, or iteration counts. These configurations directly influence how the AI interprets the prompt and generates output, leading to diverse results across systems.
Computing Power and Resources: More powerful systems can process complex prompts more efficiently, potentially accessing more advanced AI capabilities. In contrast, less powerful systems may produce simpler or more generic results due to resource constraints.
Prompt Interpretation and Context: Some AI tools adapt based on prior usage or context, which can vary from system to system. If a student's previous interactions influence the AI's interpretation, the same prompt might generate personalized results, even if the initial instructions are identical.
Updates and Access to External Databases: Some systems may have real-time updates or access to external databases that others do not, leading to disparities in the generated content. For instance, newer AI models may have broader knowledge and generate more nuanced responses.
Conclusion
While prompt writing provides a structured approach, the outcomes can still vary widely due to these technological and contextual differences. Encouraging students to fine-tune prompts based on their system’s strengths can help achieve more consistent and high-quality desired results.
