Table of Contents
Artificial Intelligence Translates the Language of Animals
Kim Kyung-jin, Attorney at Law
The Story of AI Learning to Listen to Whales, Dolphins, Birds, and Bees
Twelve chapters on how AI listens to dolphins, sperm whales, humpback whales, birds, and bees to find rules in their sounds, and what this technology means for its risks and for animal rights. Written in simple sentences a child can read, with verified sources in every section.
Table of Contents
AI Deciphers Ancient Scripts
Kim Kyung-jin, Attorney at Law
Ancient Records Revived by AI
In twelve chapters, this book explains how AI revives records once unreadable, from burned scrolls and wooden slips buried in mud to broken clay tablets. It covers virtual unrolling at Herculaneum, virtual collation of oracle-bone texts, reading Silla wooden tablets, computational analysis of undeciphered scripts, and multispectral archives, with verified references for each chapter.
Table of Contents
Artificial Intelligence for New Materials Design and Rocket Propulsion Engineering
Kim Kyung-jin, Attorney at Law
AI Potentials, Self-Driving Laboratories, and Physics-Informed Machine Learning (PIML)
Ten chapters on how artificial intelligence is changing new materials and rocket propulsion: atomic simulation, generative models, self-driving labs, high-temperature alloys, metal 3D printing, combustion, cooling design, and engine diagnosis and control. Written without equations, with verified sources in every chapter.
AI Library
The Age of Autonomous Scientific Discovery
Kim Kyung-jin, Attorney at Law
AI Scientists and Self-Driving Labs
This book follows how AI scientists and self-driving labs are changing the way science generates and verifies claims. It covers literature-based discovery, natural-language protocols translated into robot commands, multi-agent research systems, closed-loop laboratories, materials search, the verification gap, chains of evidence, research harnesses, journal ethics, and legal responsibility.
AI Library
A New Era of Life Sciences Opened by Artificial Intelligence
Structural Proteomics, Genomic Foundation Models, Autonomous Laboratories, and Global Governance
Kim Kyung-jin, Attorney at Law
This book is a research volume compiled with artificial intelligence. A human selected the materials and structured the work, while AI models drafted the sentences and cross-checked the facts.
AI Library
The Double Structure of Digital Sovereignty
Europe’s Departure from Palantir and the Chains of American Big Tech
Kim Kyung-jin, Attorney at Law
This is a record of 2026, when European intelligence agencies and defense ministries began removing analytics tools from America’s Palantir. It covers the replacement decisions made by France’s General Directorate for Internal Security (DGSI), Germany’s Federal Office for the Protection of the Constitution (BfV), and the Netherlands Ministry of Defense; the incident in which US export controls severed an ally’s ac…
New English Edition
Artificial Intelligence in Horticulture
Kim Kyung-jin, Attorney at Law
Across five chapters and ten sections, this book examines computer vision for crop diagnosis, harvesting robots and autonomous field systems, smart greenhouses and digital twins, precision irrigation and supply-chain quality control, high-throughput phenotyping, and predictive breeding.
New English Edition
Artificial Intelligence in Food Crop Agriculture
Kim Kyung-jin, Attorney at Law
Across six chapters and eighteen sections, the book examines digital agricultural infrastructure, remote sensing, crop diagnosis, yield forecasting, precision irrigation, genomics, molecular breeding, agricultural robotics, climate-smart agriculture, and global food security.
New English Edition
The Future of Forestry and Agroforestry
Kim Kyung-jin, Attorney at Law
Driven by Artificial Intelligence and Digital Innovation
Across five chapters and fifteen sections, the book follows satellites, drones, LiDAR, digital twins, forest-specific language models, wildfire and pest forecasting, forestry robotics, agroforestry, timber traceability, and forest carbon markets.
New English Edition
Smart Livestock Farming: AI Enters the Barn
Kim Kyung-jin, Attorney at Law
Sensors listen, cameras watch, and artificial intelligence helps farmers decide.
Across five chapters and fifteen sections, the book follows precision livestock farming from animal health and reproduction to robotic milking, virtual fencing, digital twins, methane reduction, welfare, and data ownership.
Table of Contents
Han Dong-hoon, Busan Buk-gu Gap: A Record of the 100 Days Before and After the Election (Mar. 26-Jul. 3, 2026)
Kim Kyung-jin
Table of Contents and 13 sections
From March 26 to July 3, 2026, this record follows the spring after expulsion, the Busan Buk-gu Gap by-election, victory as an independent, and the first bill submitted in the National Assembly.
Table of Contents
Artificial Intelligence and Medicine
Kim Kyung-jin, Attorney at Law
AI in clinical care, hospitals, education, and research
AI in medical imaging, risk prediction, treatment planning, hospital operations, education, and research, with patient safety, privacy, and accountability.
[AI Library] Chapter 15. Mastering Parameters
Nano Banana Pro Practical Prompt Book
Part 5: Advanced Techniques and Automation
Chapter 15. Mastering Parameters
Kim Kyung-jin
When you ask an AI to draw a picture, saying just "draw a cat" versus "draw the same cat as before, in high resolution, on a wide screen" leads to completely different results. These specific settings passed to the AI are called 'Parameters.' It is like adjusting the heat, the amount of salt, and the cooking time when cooking. In this chapter, you will learn how to master Nano Banana Pro's parameters. By mastering four key parameters,aspect ratio, resolution, seed value, and temperature settings,you can become a true director of AI.
Platform Optimization Strategies by Aspect Ratio
What is an Aspect Ratio? You might have wondered whether to hold your smartphone vertically or horizontally when taking a photo. In AI image generation, this choice determines the first impression of the result. Aspect Ratio refers to the proportional relationship between an image's width and height. For example, 1:1 is a square, 16:9 is a wide rectangle, and 9:16 is a tall rectangle.
Nano Banana Pro generates square (1:1) images by default. However, if you upload a great image you created as a YouTube thumbnail, the sides will be cropped; if you upload it to TikTok, black bars will appear at the top and bottom. This is because each platform has a preferred ratio. Creating your image with the correct ratio from the start prevents it from being damaged or having important parts cut off during editing.
Exploring Recommended Ratios by Platform. 1:1 Ratio (Square): The width and height are equal. Suitable for Instagram feed posts, KakaoTalk profile pictures, and brand logos. Because the viewer's gaze is focused on the center, it provides a stable feeling when emphasizing a person's face or a single product.
4:5 Ratio (Portrait Rectangle): This ratio takes up the most screen space in an Instagram feed. Since it is slightly taller than 1:1, it is advantageous for showing a person's upper body or combining a product with descriptive text. It is often used for shopping mall card news or advertisement images.
9:16 Ratio (Tall Vertical Rectangle): This ratio fills the screen when holding a smartphone vertically. It is used for TikTok, YouTube Shorts, Instagram Reels, and Stories. This ratio is also used for creating smartphone wallpapers. Being vertically long, it provides an immersive feeling when depicting full-body model photos, tall buildings, or the depth of a forest.
16:9 Ratio (Landscape Rectangle): This ratio is used for YouTube thumbnails, computer wallpapers, and PPT slides. The TV screens and movie screens we watch are also close to this ratio. It is suitable for capturing wide landscapes or scenes with multiple people. When creating a thumbnail, it is good to leave empty space on one side for text.
21:9 Ratio (Cinematic Wide): A very wide ratio like a movie theater screen. Used for creating dramatic landscapes or a movie poster feel. It is also utilized for game backgrounds or website top banners.
How Aspect Ratio Affects Composition. Even with the same prompt, the AI sets the composition differently depending on the ratio. For example, if you request "a standing woman," the background spreads out wide to the left and right of the woman in a 16:9 ratio. In contrast, in a 9:16 ratio, the full body from head to toe fills the screen. The AI automatically adjusts the person's position and the background's proportion according to the set ratio. Therefore, choosing a ratio is not just about size; it's a way to convey your creative intent.
If you force-crop a square image into a vertical one later, you will run into problems like the person's head being cut off or their feet disappearing. It is best to decide where the image will be used first and get into the habit of setting the appropriate ratio from the beginning.
[Practical Prompt Examples]
Case A: Travel Vlog Image for YouTube Thumbnail (16:9 Ratio)
Korean Prompt: Draw a landscape where the emerald sea of Jeju Island spreads out wide. On the right side of the photo, a woman wearing a straw hat and a white dress stands with both arms open, looking at the sea. The sky is clear with fluffy clouds. Leave empty space on the left side of the photo so text can be added later. Express it with bright and vivid colors that capture the excitement of travel. Create it in a 16:9 landscape aspect ratio.
English Prompt:
A wide view of Jeju Island's emerald sea. On the right side of the image, a woman wearing a straw hat and white dress stands with arms spread wide, looking at the ocean. The sky is clear with fluffy clouds. Leave empty space on the left side for text overlay. Bright and vivid colors that capture the excitement of travel. 16:9 landscape aspect ratio.
Key Points to Follow:
Request space for the thumbnail title in the prompt in advance
Placing the person on one side naturally creates space for text
16:9 is the recommended ratio for YouTube thumbnails
Case B: Dance Challenge Background for TikTok/Shorts (9:16 Ratio)
Korean Prompt: Draw a back alley of a futuristic city filled with cyberpunk-style neon signs. The ground is wet from rain and reflecting the neon lights. Purple and pink lights are glowing brilliantly. Capture a full-body view of a male dancer in baggy hip-hop style clothes performing a dynamic dance. The camera is at a low angle, looking up. Create it in a 9:16 vertical ratio to fill a smartphone screen.
English Prompt:
A cyberpunk style back alley in a futuristic city filled with neon signs. The ground is its wet from rain, reflecting neon lights. Purple and pink lighting glows brightly. A male dancer in loose hip-hop style clothing is dancing dynamically, full body visible. Camera angle is low, looking up. Fill the smartphone screen completely with 9:16 vertical aspect ratio.
Key Points to Follow:
The 9:16 ratio fills the entire smartphone screen
Requesting a full-body view allows you to take advantage of the vertical ratio
It can be used immediately as a background for Shorts or Reels
Case C: Routine for Converting One Scene into Multiple Ratios
Step-by-step Prompts:
Step 1 - Generate Basic Image (16:9):
Create a modern Korean cafe interior with large windows and natural light. A young woman sits at a wooden table with a latte and laptop. Plants decorate the space. Warm and cozy atmosphere. Photorealistic style. 16:9 aspect ratio.
Step 2 - Convert the Same Scene to 4:5:
Reframe the same cafe scene as 4:5 vertical format. Keep the woman, table setting, and lighting identical. Crop to focus on the upper body and table. Maintain consistent style and mood.
Step 3 - Convert the Same Scene to 9:16:
Reframe the same cafe scene as 9:16 vertical format. Extend the background naturally to show more of the ceiling and floor. Keep the woman's identity, clothing, and all props unchanged. Maintain lighting consistency.
Usage Tips:
Create images for multiple platforms with a single generation.
Change only the aspect ratio while maintaining consistency in characters and props.
A practical technique to save time and cost.
2. 4K High-Resolution Output and Creating Images for Printing
Why resolution matters: You might have experienced an image looking sharp on a smartphone screen but becoming blurry or pixelated when printed large. This is due to "Resolution." Resolution refers to the number of small dots (pixels) that make up an image. The more dots there are, the sharper the image; the fewer dots, the blurrier it becomes.
Think of it like a mosaic. The more small tiles you have, the more detailed the picture; if tiles are large and few, the image looks coarse. While fewer pieces might suffice for a screen, printing a poster or framed photo requires many more. Nano Banana Pro Resolution Options: Nano Banana Pro (Gemini 3 Pro Image) supports three resolutions. 1K (approx. 1024x1024 pixels): This is the default setting. It's enough for websites or social media and looks great on smartphone screens. It is ideal for quickly checking ideas or testing multiple versions.
2K (approx. 2048x2048 pixels): Contains four times as many pixels as 1K. Suitable for blog hero images, high-definition web banners, and small prints. Details remain sharp even when zoomed in.
4K (approx. 4096x4096 pixels): The highest resolution. Used for large posters, banners, magazine covers, and wall art. Every strand of hair, the texture of fabric, and even small distant objects are rendered clearly. This setting is necessary if your goal is printing.
Understanding DPI for Printing: The sharpness of a print is measured in DPI (Dots Per Inch). It indicates how many dots fit into one inch (approx. 2.54 cm). While 72 DPI is enough for screen images, printing requires at least 300 DPI.
For example, to print on A4 paper (210×297 mm) at 300 DPI, you need about 2480×3508 pixels. Since Nano Banana Pro's 4K setting (4096×4096) meets this requirement, you can print clearly up to A4 size. To create larger posters, you must further enlarge the generated image using an upscaling tool (a technology that increases size while improving quality).
High-Resolution Image Strategy: Since high-resolution images take more time and cost to generate, there is an efficient workflow. First, explore ideas at 1K. Quickly try different compositions and styles. Repeat this step until you find a direction you like. Second, once you get a result you're happy with, regenerate it at 2K or 4K. Use the same prompt but simply increase the resolution setting. Third, check details before printing. Zoom in to ensure eyes, fingers, and text are not distorted.
Adding Quality-Related Keywords to Prompts: Adding terms that encourage high quality helps the AI draw with more precision. "highly detailed," "sharp focus," "8K resolution," and "photorealistic" are helpful.
[Practical Prompt Examples]
Case A: Fantasy Illustration for Large Poster Printing
Korean Prompt: Draw a fairy house deep in a mystical forest. It's a house carved out of a giant mushroom, with golden light glowing from the windows. Fireflies are flying around. Describe it with enough precision to show the veins of every leaf and the texture of the moss. Use 4K ultra-high resolution to make even the smallest details sharp. I'll use it for printing large posters.
English Prompt:
A fairy house deep in a mystical forest. The house is carved out of a giant mushroom, with golden light glowing from the windows. Fireflies float around the scene. Render with extreme precision - visible leaf veins, moss texture, intricate bark details. 4K ultra high resolution, sharp focus throughout. Every small detail must be crisp and clear. Intended for large format poster printing. Photorealistic fantasy style. Key Takeaways:
Explicitly request "4K" and "high resolution."
Specifically mention parts that need detailed rendering.
If you mention the purpose is for printing, the AI pays more attention to detail.
Case B: Vector Style Character for T-shirt Printing
Korean Prompt: Draw a cute Welsh Corgi wearing sunglasses and riding a skateboard in a vector graphic style. Keep the background clean white. Use bold, clear outlines and flat, bright colors. Do not include shading or gradients. I plan to print it large on a T-shirt. The edges should be clean to make it suitable for screen printing. Create it in high resolution.
English Prompt:
A cute Welsh Corgi wearing sunglasses and riding a skateboard. Vector graphic style illustration. Clean white background. Bold black outlines, flat vibrant colors (no gradients, no shading). Designed for screen printing on t-shirts. Crisp clean edges required. High resolution output for large format printing.
Key Points:
Vector style does not lose quality when printed
Request simple color planes with "no gradients, no shading"
A white background makes it easy to composite onto any color t-shirt
Case C: Portrait for a Magazine Cover
Korean Prompt: Create a close-up portrait of a woman in her 60s with wise eyes. Skin texture, pores, and wrinkles showing the passage of time should look realistic. Studio lighting with rim light (edge lighting) highlighting her silver hair. Front-facing camera composition. Make it extremely detailed at an 8K resolution level, with quality suitable for a magazine cover. Photorealistic style.
English Prompt:
A close-up portrait of a 60-year-old woman with wise, thoughtful eyes. Visible realistic skin texture, pores, and natural wrinkles showing graceful aging. Studio lighting with rim light highlighting her silver hair. Camera facing straight on. 8K resolution level, extremely detailed. Magazine cover quality. Photorealistic style.
Key Points:
Details are essential for portraits
Specifying the lighting method adds a sense of depth
"magazine cover quality" provides a quality standard for the AI
Case D: Routine for Separating Text from Print Images
When including text in print materials, AI-generated text may have typos or appear distorted. There is a safe way to prevent this.
Step 1 - Generate Only the Background Image Without Text:
Create a clean poster background for a coffee shop promotion. 4K resolution. Warm brown tones with coffee beans and steam. No text. Leave large empty area in the center for typography to be added later. Print-ready quality.
Step 2 - Add Text in Design Tools: Directly enter Korean text in tools like Photoshop, Canva, or MiriCanvas. This allows you to include accurate text without typos.
Practical Tips:
It is safer to directly enter important information like brand logos, prices, and contact details
Have the AI image focus on setting the background and atmosphere
Professional designers also frequently use this method
3. A/B Testing Through Seed Value Fixing
What is a Seed Value? If you ask Nano Banana Pro to draw a "cat wearing a blue hat," a different cat appears every time. Some cats are chubby, some are thin. The shape of the hat also varies. This is because the AI uses randomness when creating images. The number that serves as the starting point for this randomness is called a 'Seed'.
A seed value is like the image's 'fingerprint' or 'DNA.' Using the same seed number produces almost the same image under the same conditions. If you get a cat image you like, record its seed number. Later, by fixing this number and modifying only part of the prompt, you can change specific parts while keeping the cat's appearance and posture.
The Principle of A/B Testing In marketing or design, A/B testing is an experiment comparing two versions. For example, to find out whether a "red button or a blue button gets more clicks," all conditions except for the button color must remain the same. Only then can you accurately compare the effect of the color.
The same applies to AI images. When creating a product detail page for an online mall, you might want to compare whether a model wearing red clothes or blue clothes attracts more customer interest. If you don't fix the seed value, the model's face, posture, and background will all change, making an accurate comparison difficult. However, by fixing the seed value, you can obtain two images where only the clothing color differs and everything else remains the same.
Practical Effects of Using Seed Values Using seed values enables the following tasks. Comparing clothing colors: Create images of the same model wearing different colored clothes to see which color suits them better. Testing background atmosphere: Place the same product in a forest background and a city background to see which atmosphere is more effective. Experimenting with lighting effects: Apply daytime and nighttime lighting to the same scene to compare the difference in mood. Maintaining character consistency: Keep the face and body type consistent when the same character appears in multiple scenes in a webtoon or children's book.
Points to Note When Using Seed Values A seed value does not guarantee a perfect duplicate. Slight differences may occur depending on model updates or server status. However, excluding the modified parts of the prompt, the overall composition and style are maintained. When conducting A/B testing, it is best to change only one element at a time while keeping the rest of the prompt exactly the same.
[Practical Prompt Examples]
Case A: Checking Seed Number After Initial Image Generation
Step 1 - Generate Basic Image:
Draw a cyborg girl standing on the rooftop of a futuristic city. She has silver bobbed hair and a mechanical arm. She is staring straight into the camera. Neon-lit skyscrapers are visible in the background. Use a realistic 3D rendering style. Once the image is generated, check the seed number in the image info. For example, assume the seed number is "123456789".
Case B: A/B Testing Clothing Color Comparisons
Version A - Black Leather Jacket: A cyborg girl standing on the rooftop of a futuristic city. Silver bobbed hair and a mechanical arm. Wearing a black leather jacket. Neon-lit skyscraper background. Realistic 3D rendering. Seed: 123456789
Version B - White Lab Coat: A cyborg girl standing on the rooftop of a futuristic city. Silver bobbed hair and a mechanical arm. Wearing a white lab coat. Neon-lit skyscraper background. Realistic 3D rendering. Seed: 123456789
English Prompt (Version A):
A cyborg girl standing on a rooftop in a futuristic city. Silver bobbed hair with a mechanical arm. Wearing a black leather jacket. Background shows neon-lit skyscrapers. Realistic 3D rendering style. Seed: 123456789
English Prompt (Version B):
A cyborg girl standing on a rooftop in a futuristic city. Silver bobbed hair with a mechanical arm. Wearing a white lab coat. Background shows neon-lit skyscrapers. Realistic 3D rendering style. Seed: 123456789
Result: The girl's face, pose, and background composition stay the same, and only the clothes change.
Case C: Product Background Comparison Test
Basic Prompt (for obtaining seed number): A minimal perfume bottle on a white marble table. Soft natural shadow. Luxury mood. Studio lighting. High resolution. Seed: 55555
Version A - Gold Background: A minimal perfume bottle on a white marble table. Soft natural shadow. Luxury mood. Golden warm lighting. Background has subtle gold texture. High resolution. Seed: 55555
Version B - Silver Background: A minimal perfume bottle on a white marble table. Soft natural shadow. Luxury mood. Cool silver lighting. Background has subtle silver texture. High resolution. Seed: 55555
Practical Tips:
The shape and position of the perfume bottle remain identical
Compare which atmosphere suits the product better by only changing the background tone
You can offer clients a choice between "Option A or Option B"
Case D: Character Expression Change Test
Basic Character Setting: A friendly Korean cafe owner in their 30s. Wearing a brown apron over a white shirt. Standing behind the counter. Warm smile. Cozy cafe interior background. Illustration style. Seed: 77777
Expression Change 1 - Serious Expression: A friendly Korean cafe owner in their 30s. Wearing a brown apron over a white shirt. Standing behind the counter. Serious focused expression while preparing coffee. Cozy cafe interior background. Illustration style. Seed: 77777
Expression Change 2 - Smiling and Welcoming Customers: A friendly Korean cafe owner in their 30s. Wearing a brown apron over a white shirt. Standing behind the counter. Welcoming gesture with bright cheerful smile. Cozy cafe interior background. Illustration style. Seed: 77777
Practical Tips:
The same character can naturally appear in various situations
Various expressions are possible while maintaining brand character consistency
Used for storytelling in webtoons or card news
4. Adjusting Temperature to Balance Creativity vs. Accuracy
What is Temperature? When we give AI tasks, we sometimes expect two different results. On one hand, we want it to "do exactly what I say," and on the other, we expect it to "show me great ideas I hadn't thought of." The parameter that balances these two tendencies is 'Temperature'.
Temperature is the AI's "creativity thermometer." When the temperature is low, the AI behaves calmly and predictably. It faithfully follows the prompts you enter and produces the most common and safe results. Conversely, when the temperature is high, the AI becomes active and unpredictable. It adds elements not in the prompt on its own or attempts unique interpretations.
Low Temperature Characteristics (0.1-0.3): At low temperatures, the AI becomes a 'model student.' It interprets the prompt literally and draws exactly as instructed. It doesn't use wild imagination. The results are stable and consistent even when repeated.
Suitable Tasks:
Logo Design: Requires precise shapes and colors
Product Photography: Must accurately represent the actual appearance of a product
Images with Text: Letters must not be distorted
Diagrams for Technical Documents: Requires accurate structural representation
For example, if you request a "red apple," the AI honestly draws a plain red apple as we know it at a low temperature.
High Temperature Characteristics (0.7-1.0): At high temperatures, the AI becomes a 'free-spirited artist.' It refers to the prompt but adds its own imagination. It sometimes adds elements not in the prompt or tries original interpretations. The results are diverse, and unexpected works can emerge. Suitable Tasks:
Fantasy Art: Representing imaginary worlds
Concept Art: Exploring new ideas
Abstract Art: Expression free from specific forms
Brainstorming: Testing various directions
Even if you request a "red apple," at a high temperature, the AI might draw an apple made of jewels, an apple floating in space, or an apple-shaped house.
Using Medium Temperature (0.4-0.6): Medium temperature is suitable for most daily tasks. It follows the prompt's instructions while adding a natural touch. Start with a medium value; increase the temperature if the result is too predictable, or decrease it if the result is too wild.
Midjourney's Stylize and Chaos Parameters: In Midjourney, functions similar to Temperature are controlled by 'Stylize (--s)' and 'Chaos (--c).' A higher Stylize value makes the AI try to decorate more artistically, while a higher Chaos value increases the variation in results. In Nano Banana Pro, you can also control this creativity through the way you word your prompts.
[Practical Prompt Examples]
Case A: Product Logo Requiring Accuracy (Low Temperature Setting)
Korean Prompt: Design a logo for a coffee shop. There is a simple coffee cup icon in the center, with steam rising above the cup. Use only dark brown and gold. The background should be white and in a minimalist style. Do not include decorative elements; draw only as instructed. Set a low Temperature to make it accurate.
English Prompt:
Design a coffee shop logo. Center: simple coffee cup icon with steam rising above it. Colors: only dark brown and gold. White background. Minimalist style. No decorative elements. Follow instructions precisely. Low temperature setting for accuracy.
Expected Result: A clean and organized logo containing only the instructed elements
Case B: Fantasy Art Requiring Creativity (High Temperature Setting)
Korean Prompt: Draw a dream world. It's a surreal landscape with strange buildings that defy the laws of physics, whales flying in the sky, rainbow-colored glowing clouds, and floating islands. Give it a dreamlike and mysterious atmosphere, and feel free to add original elements I haven't imagined. Set a high Temperature to make it creative.
English Prompt:
Create a dreamworld scene. Buildings defying physics, whales flying in the sky, rainbow-colored glowing clouds, floating islands. Surrealistic landscape. Dreamlike and mystical atmosphere. Add creative elements I haven't imagined. Feel free to experiment. High temperature setting for maximum creativity.
Expected Result: A fantastic image containing unexpected and original elements
Case C: Balanced General Illustration (Medium Temperature Setting)
Korean Prompt: Draw a cat looking out the window on a rainy day. There are raindrops on the window, and the cat is showing its back. The room has cozy yellow lighting. Express it emotionally in a watercolor style. Use moderate creativity to draw it naturally.
English Prompt:
A cat looking out the window on a rainy day. Raindrops on the window. Cat shown from behind. Room has warm yellow lighting. Watercolor style, emotional atmosphere. Apply moderate creativity for natural result.
Expected outcome: An image that follows the instructions while maintaining a natural atmosphere.
Case D: Temperature Test Routine
Compare three temperature settings with the same prompt.
Base prompt: A tree in a field.
Low temperature (focused on accuracy): A tree in a field. Realistic style. Follow description literally. No artistic interpretation. Low temperature. → A realistic image of a single, ordinary tree standing in a field.
Medium temperature (balanced): A tree in a field. Natural lighting. Peaceful mood. Medium temperature for balanced result. → A landscape image with natural lighting and a peaceful mood.
High temperature (focused on creativity): A tree in a field. Artistic interpretation welcome. Surprising elements encouraged. High temperature for creativity. → An image with glowing trees, unusual colors, or added fantasy elements.
Usage tips:
Comparing the three results lets you experience the effect of Temperature firsthand.
Develop an intuition for choosing the right temperature based on the goal of your work.
Keeping records helps you refer back to them for similar tasks later.
Summary of this chapter
1. Aspect Ratio
Set the ratio from the start based on the platform where the image will be used.
1:1 (Instagram feed), 9:16 (Shorts/Reels), 16:9 (YouTube thumbnails)
Cropping later can ruin the composition.
2. Resolution
Select 1K for web, 2K for high-definition web, and 4K for print.
Explore ideas quickly at 1K, and regenerate the final result at high resolution.
Printed materials require 300 DPI.
3. Seed Value
Using the same seed number allows you to reproduce similar images.
Use this when comparing results by changing only one element in A/B tests.
This is an essential feature for maintaining character consistency.
4. Temperature
Low temperature: Accurate and predictable results (logos, product photos).
High temperature: Creative and diverse results (fantasy, concept art).
Adjust according to your goal, starting from a middle value.
Once you can freely handle these four parameters, you will grow from a simple AI user to a director who controls AI. Getting the results you intend without relying on chance is the true meaning of being a parameter master.
Kim Kyung-jin
Lawyer · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.