Overcoming Last Year's Disappointments and Starting Fresh
I still vividly remember my first encounter with Midjourney last year. The hassle of having to input prompts in English every time, the sluggish generation speed, and most frustratingly, the artificial-looking results that made me quickly lose interest. But now, a year later, Midjourney has evolved into a completely different tool. This is exactly why I decided to renew my subscription.
The pace of development in AI image generation tools is truly remarkable. Results that seemed awkward and unnatural just a year ago are now virtually indistinguishable from professional artwork. Midjourney, in particular, has made revolutionary improvements in both user interface and functionality.
Explore Feature: A Digital Art Museum with Ten Million Works
Midjourney's Explore feature truly deserves the nickname "ten million art museum." In this space where you can view countless works created by various users at a glance, the most important thing isn't just browsing—it's learning.
When most people discover an image they like, they simply check the prompt and move on. But here's the real pro tip: drag that image directly into the input field. This allows you to immediately use that image as an image prompt or style reference.
Understanding the difference between image prompts and style references is crucial. Image prompts reference specific subjects or compositions. For example, if you upload a specific product photo and input a different background in your prompt, you'll generate an image that maintains that product while placing it in the new background. This revolutionary feature allows even complete Photoshop beginners to easily create advertising posters.
Style references, on the other hand, capture the mood or atmosphere of an image. Upload a dreamy photo and input a different location, and you'll get a new location image that maintains that dreamy atmosphere.
The Magic of Style Codes: Infinite Creative Possibilities
Style reference codes provided by sites like Midjourney SR or Style Safari are absolute game-changers. Using these codes, you can effortlessly create images with consistent styles.
What's even more amazing is that you can control the style intensity. Simply add `--sw` after the style code followed by a number between 0 and 1000. The default is 100—lower numbers apply the style subtly, while higher numbers apply it more intensely.
You can even combine two codes. Add two codes after `--srf` to apply both styles simultaneously, and you can control the ratio between the two styles by adding double colons (`::`) with weight settings. For example, the difference between setting it to 1:2 versus 2:1 is quite dramatic.
To summarize: use image prompts for form and style references for mood. If you're a beginner, start with one code and gradually build your understanding of how different styles are expressed.
Understanding Advanced Settings: Complex at First Glance, Simple at Heart
Midjourney's advanced settings may look complicated at first, but once you understand the core concepts, they're quite straightforward.
Aspect Ratiois the basic option for setting image proportions.Stylizationcontrols artistic quality. You can set values between 0 and 1000—lower settings produce realistic photos, while higher settings create more artistic, creative works. Comparing the default value of 100 with 0 and 1000 clearly shows the effect.
Weirdnesscontrols the level of creativity. The default is 0, but raising it to 100 produces unique and unexpected results.Varietydetermines how diverse the four images generated at once will be. If you want to see varied images during the drafting phase, increase this value.
SelectingRaw Modeminimizes Midjourney's inherent aesthetic style and produces more realistic images that are more faithful to your input prompts.
Draft Mode: A Revolution in Speed and Cost
The recently added Draft Mode is truly a game-changer. You can quickly generate images even with Korean input, and the operating cost is much lower. While the quality is slightly reduced, it's perfect for quickly visualizing ideas. When you get a draft you like, you can click upscale to convert it to high quality.
Here's a hidden gem: the Korean recognition accuracy has improved so much that you can input micro-prompts. Simple sentences like "Draw Earth as seen from space" can yield excellent results.
InMode Options, you can select image creation speed. Turbo mode is 4x faster than Fast mode but consumes twice the time. The Fast time you can check in your subscription account represents the monthly allocation for fast creation based on your plan. Once this time is exhausted, only Relax mode is available. Unused time doesn't roll over to the next month, so it's best to use it all.
Editing and Variations: Professional Results Without Photoshop
Once you've generated an image, it's time to enhance it with editing and variations. Don't worry if you don't know Photoshop at all.
Click on a generated image to reveal new options.Variationadds changes to the existing image. Subtle provides minor changes, while Strong offers more dramatic alterations. Use this when you don't want to completely change the image with a new prompt.
Upscaleincreases resolution, but it's not just about making things bigger. Subtle maintains the original almost identically while increasing resolution, while Creative adds details during expansion. This is often used as the final step for finished products.
TheEditorfunction allows simple editing. Partially erase parts of a photo and click regen to modify just those sections. For more detailed editing, enter the editor, upload a photo, use the eraser to remove parts you want to change, then input a new prompt.
For example, erase the torso from a portrait and input "Iron Man suit" to transform it into someone wearing an Iron Man suit. You can also change short hair to long hair or add a hat.
However, editing functions have limitations. Midjourney is weak with text generation, so adding text doesn't work well. For logos or text, it's better to use other tools in combination.
Personalize: Creating Your Own Style
One of Midjourney's most powerful features is Personalize. Click create and various images appear, where you can select images with your preferred style to train your personal profile.
For example, if you create a profile by selecting only pastel-style images, any subsequent prompts will generate images with a pastel feel. You can create specific image profiles to quickly and easily enhance consistency in your artwork.
Even more impressive is theMood Boardfeature. Here, you can directly upload external images to train styles. For instance, upload several Pokémon images to create a style profile, which generates a specific code. Use this code in prompts to draw any subject in Pokémon style.
I recently had experience creating consistent visual styles for music work by training specific album covers and inputting music-related keywords. Used this way, it becomes a powerful tool for branding and content creation.
Those SRF codes and complex number codes we've seen in Midjourney prompts are exactly these style references and profile codes. Pretty simple once you try it, right?
Essential Practical Tips: Maximizing Efficiency
Now that we've covered all the features, let me share key tips for practical application.
Midjourney can create amazing images with just a single word. Beginners should focus on core keywords like "dog," "mountain," "ocean" rather than long sentences to more easily create desired images.
An important tip here is placing elements you want to emphasize at the beginning of your prompt. For example, "dog, beach, person" and "beach, person, dog" contain the same elements but focus on different subjects. The first emphasizes the dog, while the second emphasizes the beach.
Many people approach Midjourney by randomly generating multiple images and picking favorites. But there's a more efficient method. Before generating images, visualize in your mind what the prompt should produce, then compare it with the actual results.
Thinking about why they're different and how to modify the prompt dramatically improves your skills. This method develops expertise ten times faster than just creating images.
Conclusion: Midjourney as a Creative Tool in the AI Era
Compared to a year ago, Midjourney's development is truly remarkable. Beyond mere technical improvements, the entire user experience has been revolutionarily enhanced. Korean language support, increased speed, various editing functions, and most importantly, more natural results have now made Midjourney a professional creative tool.
Particularly, the Personalize feature and style reference system go beyond simple image generation to enable building consistent brand identity or artistic worldviews. This will be a powerful tool not only for individual creators but also for corporate marketing professionals.
Of course, limitations still exist. Weak text generation, sometimes unpredictable results, and the still-necessary understanding of English prompts are areas that need improvement.
However, considering the pace of AI image generation technology development, these limitations will likely be resolved soon. What's important is how creatively we utilize these tools. Technology is just a tool—ultimately, the user's imagination and creativity determine the quality of results.
It's important to view Midjourney not as a simple image generation tool, but as a creative partner to grow with. New features will continue to be added, and we must consider how to use them creatively. This is the mindset creators in the AI era should embrace.