<aside>
Prada spent $100k+ on every campaign, I do it with $0 in under 15 minutes with my system. -@Jaymonth.
</aside>
You find references. Dump them into AI. Output looks nothing like what you wanted.
It's not your prompts. It's your inputs.
Most people throw everything into one folder — outfits they like, angles they like, locations they like. AI sees that mess and averages it. You get mush.
The fix: stop treating references as one category.
Every campaign I build starts with three separate boards:
| Board | What It Controls |
|---|---|
| STYLE | Clothing, colors, vibe, textures |
| COMPOSITION | Angles, poses, framing, cropping |
| ENVIRONMENT | Location, lighting, backdrop, atmosphere |
When you separate these, AI can layer them properly instead of averaging everything into generic output.
Here's what no one else will tell you other than me:
You don't need different images for each board.
When you're cloning a specific brand aesthetic — like Prada — use the same 8-12 campaign images across all three boards.
Why this works:
| Traditional Method | Cheat Code Method |
|---|---|
| Style board: fashion images | Same Prada images |
| Composition board: photography angle images | Same Prada images |
| Environment board: location images | Same Prada images |
| 3 different reference sets | 1 locked reference set |
| References might contradict | References perfectly aligned |
| AI tries to blend different sources | AI extracts one unified DNA |
When your references already have the style, composition, AND environment you want (like actual brand campaigns do), separating them into different source images just creates confusion.