# Calibration System

> **GOOD/BAD examples from actual projects to guide generation.**

---

## Purpose

Calibration documents provide paired GOOD/BAD examples from real projects. Claude already knows screenwriting theory — calibration shows how it applies to THIS system's aesthetic.

**Calibration is not instruction. It's demonstration.**

---

## How to Use

1. **Before generating character DNA:** Read calibration examples for behavioral specificity
2. **Before writing dialogue:** Read calibration examples for voice distinctiveness
3. **During dramatic QC:** Compare against calibration examples
4. **When something feels off:** Find the relevant category, check examples

---

## Calibration Document Location

```
/calibration/dramatic_quality_calibration.md
```

---

## Structure of Calibration Examples

Each category contains paired examples:

```markdown
## [Category Name]

### GOOD: [Brief Label]
[The example]
- Why it works: [Reasoning]
- Project source: [Which project]

### BAD: [Brief Label]
[The example]
- Why it fails: [Reasoning]
- Project source: [Which project]
```

---

## Categories

### 1. Signature Lines
Dialogue only this character would say. Passes swap test.

**Test:** Remove the character name. Can you still identify the speaker?

### 2. Behavioral DNA
On-screen behavior vs backstory. Things we SEE, not histories we're told.

**Test:** Is this something that constrains action in a scene?

### 3. Stress Behavior
Specific vs generic responses to pressure.

**Test:** Is this surprising? Would it stand out on screen?

### 4. Earned vs Declared
Built relationships vs announced ones. Actions that accumulate vs statements that skip.

**Test:** What have they DONE that earns this statement?

### 5. Voice Distinctiveness
Can identify speaker without name. Speech patterns that permeate.

**Test:** Take 5 lines from this character. Can you identify them blind?

### 6. Orthogonal Traits
Character depth vs theme-service. Traits that exist because they're REAL, not because they illustrate the thesis.

**Test:** Does this trait serve the theme? If yes for ALL traits, character feels engineered.

### 7. Emotional Register
Texture vs monotone. Range of tones within a series.

**Test:** Is there levity alongside darkness? Does the register move?

### 8. Theme Handling
Discovered vs designed. Emerged from character collision vs imposed on characters.

**Test:** Do characters discuss the theme explicitly? (Bad sign.)

---

## Adding New Examples

When a project produces notably good or bad examples:

1. Identify the category
2. Extract the specific example (keep it short)
3. Write reasoning for why it works/fails
4. Add to calibration document with project source

**Calibration grows over time. The document is never "complete."**

---

## The Calibration Mindset

When reviewing work, ask:

> "Is this closer to the GOOD examples or the BAD examples?"

If it's closer to BAD, identify WHY and fix.

---

## Cross-Reference

| Category | Also Check |
|----------|------------|
| Signature Lines | `/skills/format_v12/SKILL.md` (dialogue limits) |
| Earned vs Declared | `/skills/relationship_earning/SKILL.md` (timing) |
| Behavioral DNA | `/SCRIPTING_REQUIREMENTS.md` (checklist items 1.1-1.6) |

---

*This skill points to calibration examples. The examples themselves live in:*
*`/calibration/dramatic_quality_calibration.md`*