In this article I’m going to show you the two main ways to replicate styles β the smarty-pants AI analysis route and the direct-hit image reference method. PLUS, we’ll dive into some next-level bonus techniques for blending, controlling, and even inventing totally new styles.

Core Method A: The AI Analyst Approach (Deep Dive Style Prompt)
In this approach, you’re basically going to use another AI β like ChatGPT-4 that can handle images β to deconstruct the style you like. It’s about digging into the stylistic DNA. Here is an example prompt you can use:
Carefully examine the provided image and provide a comprehensive analysis of the art style. In your description, include the following aspects, emphasizing those most relevant to this particular artwork:
- Colour Palette and Chromatic Relationships:
- Analyse the range and combination of colours, including hue, saturation, value, and tonal relationships. Are they vibrant or subdued? Identify predominant colons, unique colour harmonies, and any distinctive use of contrast, blending, or colour theory application.
- Note how colour choices contribute to the overall mood, era, or cultural significance of the work.
- Medium and Technique:
- Analyse the methods used in the artwork’s creation, whether traditional (brushwork, sculpting, printmaking) or digital techniques.
- For traditional media: Describe the texture, mark-making approach, layering methods, and material interaction.
- For digital/photographic media: Identify rendering techniques, filter usage, manipulation approaches, and technical processes.
- Consider how the technical execution reflects mastery, experimentation, or innovation within the chosen medium.
- Composition and Visual Organization:
- Examine the arrangement of elements, use of space, and overall visual structure. Is the composition balanced, asymmetrical, dynamic, or static?
- Identify focal points, visual pathways, hierarchies, and how the artist guides the viewer’s eye. Note any use of established compositional theories (rule of thirds, golden ratio, radial composition).
- Analyse perspective choices, spatial relationships, and how depth is created or deliberately flattened.
- Thematic Expression and Symbolism:
- Identify how stylistic choices convey themes, emotions, or narratives without describing specific content.
- Note any symbolic visual language, recurring motifs, or distinctive visual metaphors.
- Consider how the formal elements (colour, line, shape, texture) serve the conceptual aspects of the work.
- Distinguishing Stylistic Features:
- Highlight unique or defining characteristics that set this work apart.
- Analyse treatment of light and shadow, representation vs. abstraction, naturalism vs. stylization.
- Identify distinctive visual signatures that might identify an artist, period, or movement.
- Historical and Contemporary Context:
- Position the work within relevant art historical movements, traditions, or schools.
- Consider both historical influences and how the style relates to contemporary artistic discourse.
- Note any fusion of traditional approaches with innovative techniques or cross-cultural influences.
- If identifiable, analyse how this work relates to the artist’s broader portfolio (typical style or departure).
- Viewer Experience and Impact:
- Evaluate how stylistic choices shape the intellectual and emotional experience of viewing.
- Consider accessibility, intended audience response, and how the style invites or challenges engagement.
- Analyse whether the style prioritizes aesthetic beauty, conceptual depth, emotional resonance, or technical virtuosity.
Provide a nuanced, technically-informed analysis that captures the essence of the art style. Focus exclusively on stylistic elements rather than descriptive content. Adapt your analysis to emphasize the aspects most relevant to this particular artwork’s distinctive qualities.

Core Method B: The Direct Approach (Reference Image Power)
Get the URL for that image β just upload it to Discord right-click, copy link, or add it directly into Midjourney. Easy. Then, in your /imagine prompt, you literally just paste that image URL right at the beginning, hit space, and then type your subject. Like: a cybernetic sloth riding a unicycle. Midjourney looks at the image, looks at your text, and does its best to mash ’em together, heavily favouring the style of the image you provided. It’s often ridiculously effective and super simple. Why? Because you’re letting Midjourney interpret the style holistically, using its own internal voodoo. The downside? Less explicit control. It might pick up on stuff in the reference image you didn’t intend, and you’re kinda trusting its interpretation. But for speed and raw impact? Often no bueno to ignore this one.
Comparing Methods & Gaining Insights with /describe
So, Method A (Analyst) vs. Method B (Reference). Which one’s “best”? β it depends! If you need that super-specific, dissected control, maybe ’cause you’re defining a complex brand style from lots of examples, Method A is your jam. If you want fast, powerful results and have a strong single reference? Method B often crushes it. It’s justβ¦ easier. Less friction. But hey, wanna know a neat little trick to help with both? Midjourney has this built-in command: /describe. You feed it an image, and it tells you what it sees, spitting out four different prompt ideas trying to capture that image’s essence. This is GOLD. Why? Because it gives you a peek into Midjourney’s “brain,” shows you the keywords it associates with that style. You can use those keywords to beef up your Method A analysis prompts OR get ideas for text to combine with your Method B image references. It’s like getting insider trading tipsβ¦ but for AI styles. Use it.
Advanced Techniques: Blending, Controlling, and Transferring Styles
Okay, got the basics? Cool. Now let’s start cookin’ with gas β time for the advanced stuff. Replication is neat, but manipulation is where the real fun begins. First: Style Blending. Don’t just copy one style; MASH ‘EM UP! You can do this a few ways: use multiple image URLs at the start of your prompt ( subject), fuse text descriptors (“Impressionist style mixed with comic book art”), or even use one image URL and add text for a second style. Experiment! Next up: Controlling Intensity. You want more or less of that style? You got knobs to turn! The Image Weight parameter (–iw) β usually 0 to 2 β tells Midjourney how hard to lean on your reference image versus your text. Crank it up (like –iw 1.8) for MORE image style. Then there’s Stylize (–s) β 0 to 1000 β balancing your prompt against Midjourney’s own artistic flair. Lower values often let your specific style instructions shine more. And don’t forget prompt weighting (keyword::2) to emphasize certain text elements. Finally: Cross-Medium Transfer. Make a photo look like a sculpture, or a 3D render feel like a sketch. Just describe it! “Photograph of a dragon in the style of a medieval tapestry.” It works surprisingly often. This is where you start getting disproportionate results from simple inputs.

Ethical Considerations and Conclusion
Look, mimicking styles is cool for learning or personal projects. But if you’re out there trying to perfectly copy a living artist’s style, especially for commercial stuffβ¦ tread carefully. It can grate on your belief system, and theirs. Don’t be a jerk. Respect artists. You can use the process we’ve gone over to get core styles, you can then blend or enhance them with your own inventions.
Try creating Novel Styles β just jam weird, unrelated keywords together. “Bioluminescent Goth,” “Minimalist Heavy Metal,” “Art Deco Glitchwave.” See what Midjourney coughs up β sometimes it’s garbage, sometimes it’s genius. You can also try Conceptual Styling β use words about feelings or abstract ideas alongside visual terms, like “a style evoking optimistic dread.” Total crapshoot, but potentially unique! You might even use Negative Styling (–no) to keep out stuff you hate (like –no photorealistic), though it’s fiddly for broad styles. And think about Sequential Styling β using one output as the image prompt for the next to make a style evolve. Now, underpinning ALL this creative chaos are some Best Practices. Crucial stuff: Use good, clear reference images β garbage in, garbage out. Nail your Aspect Ratio (–ar) for consistency. Iterate! Don’t expect perfection first try. Tweak, reroll, experiment.