Before we embark on this journey into the realm of Midjourney prompts and the intricacies of consistency, it's crucial to acknowledge that everything I'm about to share is rooted in my personal experience and I'm eager to share with you. It's essential to recognize that my insights are not absolute truths but rather perspectives that may evolve as the AI landscape continues to unfold.
Understanding the consistency of a prompt involves a complex interplay of various factors, extending far beyond the assignment of a simple seed number. As I ventured deeper into the world of prompts, I came to a realization—consistency is a multifaceted concept, and to achieve it, one must first define the fundamental appearance of the desired output. I call it type of appearance or artistic expression.
As a starting point, I recommend pondering the question: "What type of appearance do I want to achieve?" In Midjourney, when I refer to appearance or artistic expression, I am essentially addressing the core essence of the prompt—how we want it to look. Do we envision an illustration, a drawing, a stick-figure, a cartoon, a sketch, photography, or a painting, among the most common types? This foundational decision is of paramount importance, for it serves as the compass guiding your creative journey. While certain appearances can be combined, introducing instructions that reference multiple appearances or artistic expressions can lead to chaos and erratic results in the 4-image grid.
Common mistakes include prompts that request a "cartoon" appearance while simultaneously instructing it to look like a detailed illustration. Such combinations can sow confusion, as AI may struggle to decipher whether you desire a cartoon-like or illustrative output. Caution must be exercised when blending appearances.
For example: the next grid correspond to a sketch (of course forced to color changes) But the entire prompt was focused on sketch so that it was consistent.
The next one is a BIG NO! The results are not consistent. Ai is confused with our request. Because in this case the prompt had 2 artistic expressions or appearances mixed: a stick figure with a style for cartoons
We want to make it clear that we're not trying to break down what art is or its history. Instead, we're looking at it from an AI perspective, especially in the world of Midjourney, where traditional ideas about art don't always apply. For example, traditionally, a cartoon might mean a type of drawing or 3D animation. But when we work with our AI bots, we need to be more flexible in how we think. It's like a unique conversation where understanding the AI's language matters more than following traditional art rules.
In my experience with Midjourney, appearances or artistic expressions appear to carry an implicit assumption of common materials. For instance, when requesting illustrations, Midjourney often returns results with a watercolor-like texture. Similarly, if we ask for a sketch, it's typically executed with a pencil-like material. While these assumptions can be altered, it requires additional instructions to deviate from the commonly associated materials.
What combinations of appearances have proven successful for me? The sketch of a cartoon or the drawing of a cartoon. Beyond these combinations, my experience has shown that venturing into mixing appearances can often lead to less-than-ideal outcomes.
If you delve into various magazines, forums, or AI resources, you'll find classifications for art types, appearances, and styles. DALL·E, for instance, has a book dedicated to illustration styles, which encapsulates the broader categories. For Midjourney, distinguishing between a cartoon, an illustration, a photograph, and a drawing is paramount. Each holds a unique place in the AI spectrum.
The key takeaway here is that when you sit down to compose a prompt that you wish to maintain consistency, you should ask yourself, "How do I want it to look? What feeling do I aim to convey?" and begin by defining your approach conceptually. This is my recommendation: strive to avoid haphazard mixing and aimless phrasing, such as "the illustration of a cartoon" or "a painting with an illustration."
Eg: Illustration appearance super consistent prompt
The next prompt is super consistent too, I created it imagining that it would be an illustration for midjourney, and AI understood me perfectly.
I hope this tip serves you well, and I encourage you to subscribe for more insights like this.
In conclusion, this article represents a personal viewpoint based on my experiences and not an absolute truth. In the realm of logical argumentation, everyone is entitled to their own opinions and perspectives.
Don't forget to hit that subscribe button for more!
I want to invite you to swing by my prompt store. You'll find a treasure trove of creativity waiting for you, and the best part? Our prices won't break the bank. Your support goes a long way in helping me keep this project alive and thriving. So, head on over, explore, and let's keep this creative journey going together!
1 comment
Great information. Thanks 🙏🏻 It is really useful for me , I am trying to make more consistent prompts