Crafting image prompts is like crafting a signature dish
November 22, 2023
In recent months, I've been exploring Chat GPT-4 and Leonard.AI to learn how to create image prompts. My goal is to grasp their capabilities and limitations, and apply this knowledge to personal and professional projects. For guidance, I frequently reference these resources:
Creating image prompts is akin to perfecting a signature dish. Just like in cooking, you start with a basic recipe and enhance it with additional ingredients and spices. Initially, you rely on intuition and improvise, but with practice, your understanding deepens. You learn to choose the right tools, ingredients, and techniques, gradually improving your skills. Eventually, you recognize patterns, adapt your methods, and understand how ingredients react under different conditions. This growing confidence leads to developing your unique style, culminating in a signature dish.
When I first started, my prompts were brief and vague. Though the results were decent, they didn't meet my expectations.
Create an image of a surfer with wavy hair, tattoos, catching a large wave
Refining the Prompt Recipe: From Basic to Detailed
As I progressed, my approach evolved from basic one-sentence prompts to more elaborate, multi-sentence descriptions. This transition mirrors the process of adding ingredients in cooking. I began to modify the initial prompt with more details such as adding lighting, textures, context, and a visual style to enhance the image.
Asian surfer with wavy hair, tattoos, catching a large wave. The scene is set in a natural, beach-tone environment and rendered in a photo-realistic style. It showcases a fluid, shiny, soft-glow appearance, capturing the essence of a tropical Hawaiian beach. The image is highly detailed, emulating a high shutter speed action shot like those taken with a Nikon D6, and highlights the vibrant, energetic motion against the bright, blue ocean backdrop
After gaining more experience through practice and experimentation, I switched from using Chat GPT-4 to Leonard.AI to understand the differences in their image output capabilities. Leonard.AI offers extensive fine-tuning options, including the use of negative prompts, which specify elements to exclude from the generated image. Additionally, it provides optimization tools for enhancing and upscaling images, enabling more precise refinement of details such as faces and hands.
Surfer riding down huge wave, vibrant, colorful, soft-glow, ocean landscape, dramatic, Influenced by Surfer Magazine, Camera: POV, Render Style: 4K resolution, surfer wavy long hair
Negative prompts: nudity, naked, eyes closed
Be patient and practice
Remember, practice is key. Master chefs didn't achieve their culinary skills overnight. To excel in any skill worth mastering, consistent and extensive practice is essential. Keep persevering, and you'll gradually notice steady progress.
AI-generated images I created after many months of practice and fine-tuning... and still learning.
"I had the pleasure of working with Ivan on a recent project, and I was blown away by their creativity and attention to detail. I highly recommend Daniel for anyone looking for a talented and professional designer."
"I had the pleasure of working with Ivan on a recent project, and I was blown away by their creativity and attention to detail. I highly recommend Daniel for anyone looking for a talented and professional designer."
