an immersive music experience
I love making up stories, scripts, scenes, and cool shots in my head. But making them real was the tough part. Most of what I imagined was too hard or too big for me to actually film or create right now. That wall always drove me crazy.
Then AI showed up.
At first, I was like, "Nope, AI is bad! It's taking jobs!" I felt like I should boycott it. But then I had a simple thought: Every few years, something new comes along. New tools, new ways to create. Digital art did this to traditional painting, and AI is just the next big thing.
Instead of running away, I decided to jump in and use it. I realized that AI wasn't a threat, but a new medium. It was a new way to finally unlock the superpowers of my imagination.
I went and got the subscriptions. What started as just being curious quickly turned into a fun conversation and a long discussion with the machine.
an immersive music experience
Please keep in mind that the techniques, examples, and workflows discussed here are based on my personal experience using Midjourney as the generative AI platform
an immersive music experience
an immersive music experience
UNDERSTANDING THE MACHINE
Generative AI is honestly a lot to take in at first. Imagine having the power to create anything you can dream up. Endless possibilities are right there, just one prompt and one click away.
But then reality hits you.
What do you even make? And even if you know what you want, it rarely turns out exactly how you pictured it.
You quickly realize that it takes a lot of time and practice to actually figure out how prompting in Midjourney really work. It's a whole new skill!
an immersive music experience
an immersive music experience
Macro photograph of a primordial, amorphous swirl of energy, captured as iridescent liquid light. The light should exhibit a metallic, oil-slick color shift (vibrant blues, greens, golds). Soft, organic curves dominate the composition. Dynamic, swirling flow, yet overall ethereal and gentle. Light emanates from the liquid itself (diffused ambient glow), minimizing sharp shadows. Extreme shallow depth of field, sharp focus on the central fluid motion, high contrast texture, cinematic macro, studio photography.
My First Generation
Hmm, not quite what I had in mind.
So, what went wrong?
Prompting will never be perfectly accurate. It's an art of iteration.
To succeed, you must vividly describe your vision while removing redundant or weak phrases that unnecessarily dilute the central idea.
Here is the breakdown of the prompt.
(Hover to interact)
Macro photograph of a primordial,
amorphous swirl of energy, captured as iridescent liquid light.
The light should exhibit a
metallic, oil-slick color shift (vibrant blues, greens, golds).
Soft, organic curves dominate the composition.
Dynamic, swirling flow, yet overall ethereal and gentle
Light emanates from the liquid itself (diffused ambient glow), minimizing sharp shadows.
Extreme shallow depth of field, sharp focus on the central fluid motion,
high contrast texture, cinematic macro, studio photography.
Revised prompt
Ultra-macro photograph of a thick, swirling, iridescent ferrofluid. The liquid exhibits a metallic, oil-slick color shift: vibrant blues, emerald greens, and reflective golds. Internal glow, photoluminescent, soft and diffused backlighting. Elegant, slow-motion vortex, smooth organic curves. Extreme shallow depth of field, razor-sharp focus on the central fluid motion. Emphasize textural contrast from reflections, not shadows. Cinematic macro, studio lighting.
THE PROMPT PROCESS
One thing that I realised is that prompts are super sensitive. Changing just one small word can completely change the picture's whole rhythm.
For example, asking for "a cat watching a storm" gives you one vibe. But asking for "a lone tabby kitten observing a tempest, cinematic lighting" gives you something totally different. Every adjective, every style word is a tiny magic spell.
How I write my prompts
Instead of describing the action, I focus on the viewfinder telling the AI what's in the shot, not just the story.
I always start with one powerful, specific word that locks in the feeling I want—a word like "Melancholy," "Savage," or "Monumental." This word acts as the anchor, forcing the AI’s imagination to stay within that emotional lane, even if the rest of the prompt is loose.
Over time, I built a personal list of words that often ruin my images or make them too generic. I call these my "noise words," and I actively leave them out. This makes the AI work harder and find a unique solution.
Building over a prompt
Your first prompt is almost never the final one. You always have to tweak and build on it to get the result you want.
1
3
7
Prompt 1
a black car floating in the air, wrapped inside an oval glass capsule with a pure dark background, 3d rendering, high-definition details, and a sense of technology. the overall composition is symmetrical, with soft lighting creating shadows on its body. high-resolution photography, high quality.
…a little advance process
My absolute favorite parts of Midjourney are StyleReference (--sref) and the Edit Mode. They make me feel like I'm in "god mode".
With S-Ref, I can instantly grab any vibe or mood I want just by giving the AI a reference picture. And the Edit Mode gives me total freedom of manually adding things, erasing parts or changing how intense certain elements look in the generated image.
Once I figured out how to use S-Ref, Omni Ref, manual editing, and strong prompts, I took my game up a notch.
I started combining different AI tools. For example, I use Nana Banana to quickly generate several different camera angles of the same scene. Then, I take those images and stich them together in Veo to create a short animation.
S-Ref
using Nano Banana to generate different angles
using generated angles in Veo
FUTURE PETALS
A series of flowers from the future
HOTEL CRIMSON VEIL
A series of architectural marvel of crimson fever
COLLEGE ERA
A series of mixed media collage compositions
