top of page
Writer's picturedanceequations

AI Images and Prompt History of Treedom - My Exploratory Process of a Children’s Picture Book

The capabilities of AI are not beyond our imagination. I grew up watching Star Trek and other science-fiction shows where “computer” was just as much of a main character as any other intergalactic species. Yet, however familiar the idea, I am both exhilarated and apprehensive about the future it heralds. Innovation, with its every leap and bound, carries with it a price tag. This eternal equilibrium has endured for centuries - from the rudimentary stone tool to the revolutionary wheel and the advent of steel, we continually build upon the knowledge bequeathed by preceding generations. Just when we presume that we have stretched the limits of our capabilities, we persistently break through uncharted boundaries, metamorphosing in the process. For Treedom my children's book, AI became an invaluable tool, nonetheless, the illustrations transformed through seemingly endless revisions and alterations through a very human endeavor of crafting and refining. Throughout my creative process, it became overwhelmingly clear that human skills were required to convey a cohesive idea and achieve a consistent artistic aesthetic. 


Midjourney 3, the AI tool I employed, struggled to comprehend the concept of imbuing emotion into a tree. Therefore, my creative process spanned thousands of generated images, each informed by varying styles and prompts. I scoured the web for suggestions, dedicated countless hours to watching instructional YouTube videos, and even engaged in somewhat whimsical exchanges with Chat GPT. I would inquire with Midjourney what it "perceived" within my images and attempt to reverse-engineer my creative process.


The image below signifies the first instance when AI unequivocally affirmed its potential, a watershed moment amidst a myriad of generated images, countless prompts, cut-and-paste endeavors, and scans of my drawings which inevitably led me to this juncture—when the protagonist of Treedom encounters the Council of Trees and their Dogwood delegates.


I could discern that the program was evolving, and with each stride I made, the AI bot also advanced. One fateful morning, perhaps on the three-thousand-and-first attempt, the breakthrough arrived. Gaia, in her authentic attire, stood alongside the correct tree, and communicated to Gaia, "Can you aid us?" It was all the reassurance I needed; I could proceed. With AI as my steadfast companion, I was poised to bring my book to fruition. Here, you'll find the initial raw image generated by AI, contrasted with the final polished rendition skillfully crafted by the talented Rachel Lang - rendering the image useful in my book.





//image: <https://s.mj.run/f7HvB0MvvYw> <https://s.mj.run/elKaZYT4oy0> A 12-year-old girl in urban clothing blue jeans and white t-shirt and with a pink pixie-cut hairstyle, standing in front of tree-ent happy mystical creature, viewpoint from behind, many tree-ents in the mystical forest, in children's book illustration style, watercolor style –gaia –ar 16:8


My journey into crafting tree-person illustrations invariably demanded a measure of courage. More often than not, I would end up with a creepy-looking human face imposed on a tree, rather than the mystical creature I was hoping for. Additionally, it might be my imagination, but I noticed that Midjourney exhibits varying “moods”. Indeed, generating images during specific times was essential if I was to achieve the outcome I wanted. I found that if I generated images during peak hours, the styles would change to whatever seemed popular on the group servers. 


Creating a single image represents one challenge, but generating a series of images for coherence within a children's book presents an entirely different endeavor. Even if I found one suitable rendition, the slightest deviation from the required style would render it unsuitable. Each piece must seamlessly integrate into the larger puzzle, and Gaia was continually being generated with slightly different clothing, sometimes wearing a backpack and other times a jacket. Sometimes the illustrations would have black outlines, sometimes would linger too close to photorealism, etc. There were days I’d be so melancholy over the process, then suddenly Midjourney would begin to deliver and I would fall into a hypnotic rhythm racing the clock knowing that a pause would soon come - my home responsibilities were knocking. With four young kids and a full-time job, my Midjourney moments were difficult to find.



My somewhat delicate AI bot necessitates a substantial warm-up. Once I sense that we're operating in harmony, I can employ my prompts, preferences, and images. My process included scanning in ideas I had personally drawn, meticulously constructed in Photoshop, or swiftly assembled in Canva. I would see what would appear, ask Midjourney what it saw, and then begin to use my preference, and “seed” images.


Here, I've successfully rendered my character in the precise style and colors needed. This achievement earned it a well-deserved spot in the "top choices" folder—a significant epiphany in my ongoing quest to bring animated nature to life. I need to note: that I never input anyone’s art but my own. I did not generate any of my images from outside of Midjourney and I arrived at my desired image through my own layering and prompts.





//imagine: 2d game art, children's book illustration, no outline, 12-year-old, pink pixie-cut hairstyle, wearing green urban clothing, planting sunflowers, night city in the distance, white watercolor edge


The inclusion of a single, three-letter word, "ent," played a pivotal role in my creative process. For this revelation, I owe a debt of gratitude to Tolkien. The introduction of this word allowed me to establish an "ent" preference, setting me on an inspired creative streak. Therefore, this image, or a version thereof, may have very well found its place within the pages of my book. Once again, you’ll see the raw AI image along with the final edited version.




//imagine: <https://s.mj.run/f3VwD8wKV_g> <https://s.mj.run/U0g5jRMWkYY> 2d game art, children's book illustration, no outline, ent holding its branch-like arms high in the air, tree-man with bark face and glowing eyes, 2 children trying to reach the branches, arms up, smoggy skies –ents –ar 16:8


This blog would become excessively lengthy if I were to delve into every difficulty and obstacle I encountered. To highlight just a few, I must mention Midjourney's struggle to generate a whale in my desired style; they often appeared as a hybrid between a shark and a whale, adorned with extra fins.





For those acquainted with Midjourney 3, anomalies such as additional fingers or distorted features were commonplace. Nevertheless, amidst these challenges, there was one image that remained unaltered and found its place in Treedom. Here we used the Midjourney image without alternations but roughly 6 hours within my creative process.





//imagine: 2d illustration, watercolor style, tree-ent mystical creature, in desert scene landscape, a desert landscape at sunset, African fever tree, African influence, tree-ent with face in bark, red highlights, sand dunes in the background, sandy ground


While some images required minimal editing from Rachel others quite a bit, one was entirely of her creation. This particular image stands as a testament to Rachel's exceptional talent and marks a pivotal moment in the Treedom narrative—Gaia's poignant encounter within the peak of the trees' strike as she marches alongside them. Despite numerous attempts with various prompts and drawings, no AI-rendered image could capture the essence I sought. This image stands as a testament to Rachel Lang's creativity and skill, embodying 100% of her artistic vision and her ability to match the style I was seeking. 



As I reflect on my journey with Midjourney 3, I'm amazed by how much it's evolved and become even more user-friendly. Currently immersed in the writing process of Treedom ll (The Awakening), I eagerly anticipate the new features and possibilities that await me when I return to the illustration phase with Midjourney. My AI journey is far from over—I'm committed to enhancing my work by incorporating additional electronic drawing tools, expanding beyond the limitations of paper and pencil. I look forward to connecting with fellow artists and individuals embarking on their AI journey. May this blog shed light on the intricate work behind the beautiful images in my book and provide valuable insights for those undertaking a similar process.





Thank you for supporting my projects and stay tuned for Treedom ll. You can find Treedom The Uprooting on Amazon.

Comments


bottom of page