Before the invention of printing, people organized their knowledge of the world around them in hand-made books: codices, manuscripts, and bestiaries.
Many of these fascinating artifacts continue to captivate and inspire study in the modern age. Some served as legitimate documents of their times, others as compendiums of real and mythical creatures, plants, or phenomena. On rare occasions, an unknown author would leave behind a true mystery—an indecipherable collection of texts, symbols, and drawings of the most bizarre nature.
In this study, we will explore the most famous manuscripts, codices, and bestiaries known to Midjourney. We'll learn how to reproduce ancient books that Midjourney isn't familiar with, using Text and Image Prompts. Finally, we'll speculate about the manuscripts of the future and discuss how to create them.
Manuscripts/codices are handwritten documents, often in the form of illustrated books. They can include religious and scientific texts, historical records, or mythological narratives. And some of them contained ciphered texts, were written in false languages, and accompanied by mysterious pictures. Many are studied and are not decoded by scientists around the world to this day. Let’s start with those!
Two of the most famous and impressive examples of such artifacts are the Voynich Manuscript and the Codex Seraphinianus.
Created centuries apart—the Voynich in the 15th century and the Seraphinianus in the 20th—these books are a testament to human creativity and enduring mystery. While the Voynich features detailed, almost scientific sketches and has ties to medieval Europe, the Seraphinianus is a surreal art piece with no known cultural context. Both leave us with more questions than answers…
The Voynich Manuscript's exact date of creation has long been a subject of scholarly debate. Named after Wilfrid Voynich, the book dealer who acquired it in 1912, the manuscript is generally believed to date back to the 15th century, based on its writing style, illustrations, and textual references.
Composed in an as-yet-undeciphered script, the manuscript features elaborate illustrations spanning subjects like botany, astronomy, and human anatomy. Despite decades of extensive research, its content, purpose, and origins remain shrouded in mystery. Linguists, historians, and cryptographers alike have been stumped, making the Voynich Manuscript one of history's most enigmatic and enduring puzzles.
Created in the late 1970s by Italian artist Luigi Serafini, the Codex Seraphinianus is a tour de force of imagination. Its 360+ pages feature an invented, asemic script alongside intricate illustrations of surreal landscapes, bizarre creatures, and perplexing machines. Far from a coded message, its text is intentionally indecipherable to challenge a viewer's understanding of language and art.
Here is what Midjourney is capable of given Serafnin's iconic Codex as a style reference:
Beyond the world's most unsolvable enigmas, Midjourney is also acquainted with less mysterious works that serve as authentic documents of their epochs and civilizations. Let's take a brief look at some of these, starting with feeding just their titles to MJ, and then delving into more complex prompts and some prompting strategies for best results.
...and virtually for any Midjourney style
There are a fest of best practices for generating impressive results with codices and manuscripts styles (and any refence style, in fact) in Midjourney.
Base Prompts vs. Optimal Prompts
While simple Base Prompts like "...in the style of Codex Seraphinianus" or "...in Voynich manuscript style" can work, Optimal Prompts—where you specify the context and place your reference first—usually yield better results.
Simply put: more details often equals better results. Be reasonable, though: super long prompts might bring cool results, but they are way less controlable. Put the most important details closer to the prompt’s beginning.
Adjusting the --stylize value
Lowering the --stylize value slightly often produces results more faithful to the original works. For example, the default value of 100 tends to generate "more beautiful" images according to Midjourney's criteria, featuring enhanced 3D volume, color gradients, and abundant complexity. However, lower values (in my experience, the “sweet spot” is between 35 and 75) can create more authentic, simplistic, 2D, and naive images.
For titles that Midjourney isn't familiar with—and there are many—you can always employ Text and Image Prompts to replicate the desired original styles. The strategy here is straightforward: blend generic descriptions and contexts with specific details. Let's check out some examples:
Another approach is to feed an image of the original manuscript page to Midjourney's /describe command and use its analysis to generate the desired result. Here's a great example!
Codex Gigas, also known as the "Devil's Bible," is one of the largest medieval manuscripts in the world. It includes the entire Latin Vulgate Bible, along with other texts like works of history and medicine, all written in Latin. It's particularly famous for a full-page illustration of the Devil, one of the most famous and enigmatic illustrations of Satan in medieval literature.
Legend has it that the book was written in a single night by a monk who sold his soul to the Devil to complete it!
Although famous, Midjourney is unaware of this hypnotizing illustration. However, given the original, it can /describe it. Here is one of the four prompts MJ produced after analyzing the scan:
medieval demon and dragon from a manuscript by thomas de la caixa, in the style of undefined anatomy, hanging scroll, green and crimson, illusory images, unprimed canvas, pre-columbian art, dissected books --v 5.2
Let's take a look at what this prompt produces, and what can be done to improve the outcome.
As often happens, the prompt has a weird structure, yet it returns a somewhat relevant result!
But can we get closer to the original? Yes—by using this weird text alongside the original picture via Image Prompt!
To go even further, increase the image weight (--iw 2). And edit the /describe's prompt by removing unnecessary parts and correcting crucial details—like changing the color from green to earthy.
The question is: How can we reimagine these age-old works to find inspiration, craft modern art projects, develop new visual languages, or even engage in moments of visual meditation—using Midjourney? Or, to put it another way, what might a current-day Codex Seraphinianus, a future Voynich Manuscript, or a space-exploration-age Aberdeen Bestiary look like?
For this experiment, I asked ChatGPT (of course, we are talking about future!) to fantasize with me, and imagine and describe such artifacts. I then shortened and edited the resulting prompts to be more along my prompting strategy, and here is what happened:
Human history in general and the history of art specifically are full of fascinating artifacts—like manuscripts, codices, bestiaries, and many others—that have captivated our minds for centuries.
These works not only serve as windows into the past but also as springboards for our imagination and creativity today. In the digital age, AIs like Midjourney allow us to engage with these ancient treasures in unprecedented ways, expanding our artistic horizons and enabling us to produce new forms of art. Want to experiment futher? Try combining codices, manuscripts, and bestiaries prompts with other artist styles…
Interestingly, it's not always necessary for Midjourney to be familiar with specific source material to generate results that resonate with those sources. Through experimentation, daring approaches, and an open mindset, we can adapt and evolve these historical styles into something entirely new.
If you like our guides and studies, please, consider supporting us. It's thanks to our Patrons that we are able to maintain and develop Midlibrary, create better educational content, and keep it free for all!
All samples are produced by Midlibrary team using Midjourney AI. Naturally, they are not representative of real artists' works/real-world prototypes.
We'll be grateful for shares and backlinks!
Midlibrary Catalog grows largely through the contributions of our Community.
Thank you for taking time to share your suggestion!
We do our best to keep this website running as smoothly as possible.
However, stuff happens. Thank you for letting us know about it!
Every week we publish a new Midjourney study and a new Editor's Pick.
Receive our newsletter to never miss an important Midlibrary update!