🪪 Reference 1 = identity
A Reference 1 is the heart of the system. It’s the single source of truth for identity: face structure, facial proportions, skin tone, hairstyle, and the character’s recognizable appearance. Everything that makes someone “that person” lives here.
🎯 What Ref 1 controls
- •Face structure and proportions
- •Skin tone and texture
- •Hairstyle and hair shape
- •Eye shape and signature expression
✓ What to DO
- ✓Cite "use Reference 1 as the character identity"
- ✓List face, skin, hair, and proportions explicitly
✗ What NOT to Do
- ✗Let the outfit from Ref 1 "bleed into" the result
- ✗Expect identity to come from Ref 2
Key concepts
the identity foundation
tone and texture
shape and hairstyle
the recognizable “look”
👗 Reference 2 = Outfit
A Reference 2 has a deliberately narrow role: to direct only the outfit logic. Silhouette, fabric type, accessories, color palette, and styling mood. No face, no identity.
💡 The recurring phrase
All 10 prompts contain some version of "Use Reference 2 only as inspiration for the outfit...". That “only” is what keeps Ref 2 from hijacking the identity.
Reference 2 guides
the shape of the clothing
material and drape
jewelry, bags, details
colors and styling mood
⚠️ Attention
In styles like sentai, Ref 2 inspires the fashion, but the final outfit is transformed (from blue to black, for example). Inspiration ≠ copying: Ref 2 provides vocabulary, not instructions.
🤸 Always a New Pose
Each prompt includes the explicit instruction: "Do not copy the pose from Reference 1 or Reference 2", followed by a new pose described in detail. That’s what turns “repainting” into “moving the character through worlds.”
Deny the pose in the references
The first half of the instruction is a negation: don't inherit the pose from any of the input images.
Describe a specific pose
"Sitting in a vintage armchair, one leg crossed," "dramatic contrapposto," "heroic three-quarter view." The more specific, the better.
Marry pose with the world
The pose fits the setting: leaning against the car at night, sitting on the industrial ledge, standing in the skin shop.
Key concepts
don’t copy the pose
described pose
pose + setting
new each time
✂️ The Clean Split
The secret is the deliberate separation: who the person is (Ref 1) versus how the world is styled (Ref 2 + genre). Mix the two and the results become random; keep them separate and everything becomes predictable.
✓ Clean Split
- ✓Identity from Ref 1 only
- ✓Outfit from Ref 2 only
- ✓Recognizable, repeatable result
✗ Chaotic Mix
- ✗Face influenced by Reference 2
- ✗Outfit determined by Reference 1
- ✗Unpredictable result with each generation
💡 Practical tip
Think of each prompt as a two-column spreadsheet: "identity" and "styling." If you can't tell which column an instruction came from, rewrite it.
🧭 How to use the system
In practice, there are three moves: lock the face, change the gender, generate a new pose. Repeat for each style and you’ll have a whole series featuring the same character.
using Reference 1 as identity (face, skin, hair, proportions).
Use Reference 2 only for outfit, silhouette, accessories, palette.
Do not copy the pose. Create a new pose: [SPECIFIC POSE].
Place the subject in [WORLD/SETTING].
Key concepts
Ref 1 locks identity
the world changes everything
life and variety
🧠 Why It Works
The system is consistent because the face comes from one source and the fashion from another — without ever locking the image into a single pose. Strong identity + strong visual genre = a result that is, at the same time, recognizable and varied.
📊 Why the separation matters
- 1 identity → reused across 10+ styles with no rework
- 10 genres → radically different outputs of the same person
- 0 ambiguity → each instruction has a clear source
🔑 The key insight
Understanding the “why” lets you invent new styles beyond the 10—any visual genre fits the same framework: locked identity, borrowed styling, new pose.
📝 Module Summary
Next Module:
1.2 — Library of 10 Prompts