---
name: ai-image-prompts
description: Constructs and critiques AI image-generation prompts using an eight-slot framework covering reference, subject, wardrobe, environment, lighting, composition, mood and technical look. Use when the user asks for an image prompt, wants a photo brief or plain-language idea turned into a prompt for Midjourney, DALL-E, Imagen, Firefly or similar, wants an existing image prompt reviewed or strengthened, or asks why generated images look generic, staged or fake.
---

# Constructing AI image prompts

A one-line prompt gets a one-line image. The model fills every gap with the most average answer it knows, which is where the stock-photo sheen comes from. The fix is not a longer prompt, it is a *specified* one.

This framework specifies an image in eight slots. The load-bearing idea: **an unfilled slot is not a neutral omission, it is a decision handed to the model.** Leaving a slot empty on purpose is fine. Leaving one empty by accident is the most common reason an image comes back disappointing.

## The eight slots

Fill and write them in order.

| # | Slot | Decides |
|---|------|---------|
| 1 | Reference & inspiration | What the whole image aims at, before anything specific is described |
| 2 | Subject | Who or what is in frame, and what they are doing |
| 3 | Wardrobe & styling | What they wear, and the aesthetic it signals |
| 4 | Environment | Where this happens, and how much of it is visible |
| 5 | Lighting | Whether the image reads as real |
| 6 | Composition & camera | Where the camera is and how much it sees |
| 7 | Mood & style | The register of the finished image, as opposed to its contents |
| 8 | Technical look | The look of the medium itself. Optional, and usually skipped |

The values each slot can take are in **Slot vocabulary** at the foot of this file.

## Two rules that govern everything

**Lighting carries the most weight.** Of the eight slots, lighting has the largest effect on whether an image reads as real rather than rendered. When effort has to go somewhere, it goes here. Specify all four parts — source, direction, quality, and what the light bounces off. The bounce surface is the one people skip, and it is the difference between an image that looks lit and one that looks photographed.

**Change one variable at a time.** When iterating, hold seven slots still and vary one. If three things change and the image improves, nothing has been learned about which of the three did it. So every prompt handed to the user comes with the single next variable worth varying — lighting first, unless something else is obviously wrong.

---

# Mode 1: Construct

Turn a plain-language description into a filled prompt.

## Step 1 — Identify the image category

From the user's brief, identify the primary subject:

| If the brief is about... | Category |
|--------------------------|----------|
| A person or people | People / Lifestyle |
| A product, object, or thing | Product / Still life |
| Food or drink | Food & beverage |
| A place, scene, or space | Environment / Architecture |
| An abstract idea or concept | Conceptual |

## Step 2 — Ask discovery questions

Most users have something in mind but lack the vocabulary to describe it. Before filling any slots, ask discovery questions in plain language. Present them together, not one at a time.

**Which questions to ask depends on the category and what the brief already specifies.** If the user's brief already answers a question (e.g., they described the wardrobe), skip that question.

---

### For People / Lifestyle (8 questions)

1. **What's the light like?**
   - Warm and golden (late afternoon)
   - Soft and even (overcast)
   - Bright and airy (midday, soft shadows)
   - Harsh and contrasty (midday, hard shadows)
   - Moody and dim (dusk/evening)

2. **What's their energy?**
   - Confident, owning the frame
   - Relaxed, effortless
   - Excited, genuine delight
   - Mysterious, guarded
   - Caught mid-moment, unposed

3. **Looking at camera?**
   - Direct eye contact
   - Looking at something in the scene
   - Looking off-frame
   - Doesn't matter / face not the focus

4. **What's in frame?**
   - Full body (head to toe)
   - Three-quarter (down to knees/thighs)
   - Waist up
   - Close-up (face and shoulders)

5. **Background?**
   - Soft and blurry (subject pops, environment falls away)
   - Sharp and in focus (full context, everything legible)
   - In between (there but not distracting)

6. **What's the overall feel?**
   - Fashion editorial (polished, intentional)
   - Lifestyle ad (aspirational but relatable)
   - Documentary/street (raw, authentic)
   - Social content (casual, relatable)
   - Vintage/film (grain, retro color)

7. **What are they wearing?**
   - Loungewear/comfy
   - Casual everyday (jeans, tee)
   - Polished/styled
   - Athleisure
   - Workwear/uniform
   - Formal

8. **Any details about them?**
   - Hair (up, down, messy, styled)
   - Accessories
   - Any other styling notes
   - Or: "natural, doesn't matter"

---

### For Product / Still life (6 questions)

1. **What's the light like?**
   - Bright and clean (studio, even)
   - Warm and natural (window light)
   - Dramatic (hard shadows, contrast)
   - Soft and diffused (no harsh shadows)

2. **Hero or in-context?**
   - Isolated on a clean background
   - Shown being used
   - In a styled setting/scene

3. **What should it say about the product?**
   - Premium, luxurious
   - Approachable, friendly
   - Rugged, durable
   - Minimal, clean
   - Playful, fun

4. **What's in frame?**
   - Product only, tight crop
   - Product with breathing room
   - Product in full scene/context

5. **Background?**
   - Clean/solid color
   - Textured surface (wood, marble, fabric)
   - Environmental (in a real space)
   - Blurred/soft

6. **What's the overall feel?**
   - Commercial/advertising
   - Editorial/magazine
   - Catalog (clean, informational)
   - Lifestyle (aspirational)
   - Social content (casual)

---

### For Food & beverage (6 questions)

1. **What's the light like?**
   - Bright and airy (fresh, appetizing)
   - Warm and moody (cozy, indulgent)
   - Natural window light
   - Dramatic (dark, contrasty)

2. **Styled or natural?**
   - Perfect, art-directed, every detail placed
   - Mid-meal, someone actually eating
   - Ingredients in process (cooking)
   - Casual, just-plated

3. **What's the hero?**
   - The dish itself
   - The ingredients
   - The moment of eating/drinking
   - The whole table/spread

4. **What's in frame?**
   - Tight on the food
   - Full plate with context
   - Table scene with hands/people
   - Overhead spread

5. **Background?**
   - Clean and minimal
   - Textured surface (wood, linen, marble)
   - Kitchen environment
   - Restaurant/dining setting

6. **What's the overall feel?**
   - Food magazine editorial
   - Restaurant menu
   - Recipe blog (approachable)
   - Social content (casual, real)
   - Fine dining (elevated)

---

### For Environment / Architecture (5 questions)

1. **What's the light like?**
   - Golden hour (warm, long shadows)
   - Midday (bright, crisp)
   - Overcast (soft, even)
   - Blue hour/night (moody, artificial lights)
   - Interior lighting only

2. **With people or without?**
   - Empty and still
   - Signs of life (objects, movement)
   - People present and active

3. **What should the space feel like?**
   - Inviting, welcoming
   - Impressive, grand
   - Cozy, intimate
   - Stark, minimal
   - Lived-in, real

4. **What's in frame?**
   - Wide establishing shot
   - Detail/vignette
   - Specific architectural feature

5. **What's the overall feel?**
   - Architectural photography (clean, precise)
   - Real estate (inviting, aspirational)
   - Documentary (authentic)
   - Editorial (stylized)
   - Travel/lifestyle

---

### For Conceptual (4 questions)

1. **Literal or metaphorical?**
   - Show the thing directly
   - Represent it through imagery/symbol

2. **What's the tone?**
   - Serious, weighty
   - Hopeful, optimistic
   - Playful, light
   - Mysterious, ambiguous

3. **Any visual references?**
   - Something they've seen that's in the right direction
   - A style or photographer
   - Or: "open to interpretation"

4. **What's the overall feel?**
   - Fine art
   - Commercial/advertising
   - Editorial
   - Abstract

---

## Step 3 — Map answers to slots

Use the discovery answers to fill the eight slots. Here's how common answers translate:

| Answer | Fills |
|--------|-------|
| Warm and golden | Lighting (late afternoon sun, warm tones), Mood (warm color treatment) |
| Soft and even / Overcast | Lighting (diffused, no hard shadows), Mood (neutral or cool) |
| Harsh and contrasty | Lighting (hard source, defined shadows), Mood (dramatic) |
| Moody and dim | Lighting (low light, evening), Mood (darker, atmospheric) |
| Confident / owning the frame | Subject (direct posture, assured expression), Mood (editorial) |
| Relaxed / effortless | Subject (natural posture, easy expression), Mood (casual, documentary) |
| Caught mid-moment | Subject (action, eyes off-camera), Mood (documentary, candid) |
| Direct eye contact | Subject (looking at lens) |
| Looking at something | Subject (eyes on object, mid-action) |
| Full body | Composition (wide enough for head to toe) |
| Three-quarter | Composition (framed to knees/thighs) |
| Waist up | Composition (medium shot) |
| Close-up | Composition (tight on face/shoulders) |
| Soft/blurry background | Composition (shallow depth of field, wide aperture) |
| Sharp background | Composition (deep depth of field, stopped down) |
| Fashion editorial | Reference (polished), Mood (intentional, styled), Lighting (controlled) |
| Documentary | Reference (not stock, not staged), Mood (authentic, raw), Lighting (natural/available) |
| Lifestyle ad | Reference (aspirational but relatable), Mood (warm, inviting) |
| Vintage/film | Technical (film stock, grain), Mood (retro color treatment) |

## Step 4 — Fill remaining slots from context

With discovery answers mapped, fill any remaining open slots using these defaults:

| Open slot | Default |
|-----------|---------|
| Reference | Negative direction: "Not a stock photo, not centred, not evenly lit" |
| Subject | Mid-motion rather than held pose; attention off-frame unless "posed" was chosen |
| Wardrobe | Follow from environment and activity; consistency matters more than specifics |
| Environment | Name a specific place rather than a category — specifics carry their own props and light |
| Lighting | Derive from the "light" answer. Always specify source, direction, quality, and bounce |
| Composition | From framing + background answers. Add angle (default eye level) |
| Mood | Should be set from discovery; this is the most important one to have answered |
| Technical look | Omit unless vintage/film was chosen or user mentioned a specific camera/stock |

## Step 5 — Write it as prose

One paragraph, slots in order, no line breaks and no slot labels. The prompt goes into a text field, and labels are noise the model has to parse past. Slot order matters — the model weights early tokens more.

## Step 6 — Output

```
[the prompt as a single prose paragraph]

**Your choices:** [restate their discovery answers in plain language]
**I filled in:** [the technical details they didn't specify — lighting direction, bounce, angle, etc.]
**Try varying:** [one variable worth experimenting with, and 2-3 alternative values in plain language]
```

## Step 7 — Offer to explore

After delivering the prompt, offer variations:

> "Want to see what changes if we try [different lighting / a different angle / warmer tones]? I can rewrite the prompt with any of those."

---

## Worked example

**Request:** "woman in mid 20s is on the front step of her house bending down to pick up a meal-kit box that was just delivered on her porch"

**Discovery questions asked:**

> Before I build this prompt, a few questions:
>
> 1. **What's the light like?** Warm and golden / Soft and even / Bright and airy / Moody and dim?
> 2. **What's her energy?** Excited / Relaxed and casual / Caught mid-moment?
> 3. **Looking at camera?** Direct eye contact / Looking at the box / Looking elsewhere?
> 4. **What's in frame?** Full body / Three-quarter / Waist up / Close-up?
> 5. **Background?** Soft and blurry / Sharp and in focus / In between?
> 6. **What's the overall feel?** Lifestyle ad / Documentary / Social content / Editorial?
> 7. **What's she wearing?** Loungewear / Casual everyday / Polished / Athleisure?
> 8. **Any details about her?** Hair, accessories, styling notes?

**User answers:** "Warm and golden. Relaxed casual. Looking at the box. Full body. Sharp focus. Documentary. She's wearing jean shorts, bare feet, white tee. Hair in a messy bun."

**Output:**

> Not a stock photo, not staged, not a lifestyle ad. A woman in her mid 20s on the front step of her house, mid-bend, reaching down to pick up a meal-kit box on the porch — a real moment, relaxed, eyes on the box, maybe a slight smile. White t-shirt, jean shorts, bare feet on the worn wood, hair pulled up in a messy bun, a few strands loose. A real front porch with its own details — a doormat, a potted plant, house numbers by the door, the texture of the painted wood. Late afternoon sun, low and golden, raking warm across the porch from camera left, long soft shadows, catching the edge of the box and the loose strands of her hair, bouncing off the siding. Full body framing, head to toe, the whole porch in context. Deep depth of field — everything sharp, the door behind her legible, the yard soft but present. Documentary and authentic, warm, unhurried, like someone happened to catch the moment from the sidewalk.
>
> **Your choices:** Warm golden light, relaxed and casual, looking at the box, full body with porch context, sharp focus throughout, documentary feel, white tee and jean shorts, bare feet, messy bun.
>
> **I filled in:** Porch details to ground the scene, late afternoon sun from camera left, a few loose strands of hair catching the light, deep depth of field for documentary everything-in-focus look.
>
> **Try varying:**
> - *Hair down instead of up* — more relaxed, catches more light, different silhouette
> - *Shot from lower, porch level* — more context, she towers slightly, more grounded
> - *Morning light instead of afternoon* — cooler golden, different energy

---

## Quick mode

If the user explicitly asks for speed ("just give me a prompt", "don't ask questions", "quick"), skip discovery and fill slots using these defaults:

- Natural/available light, soft
- Candid over posed
- Medium shot, shallow depth
- Documentary tone
- No film stock (clean digital)

State the assumptions clearly in the output so they can override.

---

# Mode 2: Critique

Assess a prompt the user already wrote, then offer a stronger version.

## Step 1 — Classify each slot

Walk the eight in order. Each is one of:

- **Missing** — no value at all. Say what the model will default to.
- **Vague** — names a category but not a value. "Good lighting", "nice outfit", "cinematic", "high quality", "professional" are categories that function as nothing.
- **Contradictory** — fights another slot. See the contradiction checklist below.
- **Specified** — say so briefly and move on.

Every note names the slot, quotes the words at fault, and says what would replace them.

## Step 2 — Check for contradictions

Common ones:

- **Two dominant light sources** — pick one dominant, name any other as fill
- **Light the environment can't produce** — an interior needs a window for direct sun
- **Tone against pose** — candid + direct eye contact is a contradiction
- **Wide shot + shallow depth** — shallow throws away the environment the wide shot exists to show
- **Film stock + opposite colour treatment** — Portra is soft; don't ask for "vibrant" with it
- **Period aesthetic that leaks** — "1970s wardrobe" with "clean modern digital" will fight

## Step 3 — Output

```
**Working:** [slots that are genuinely specified]

**Problems:**
- **[Slot] — [missing/vague/contradictory]:** [quote] → [what to put instead, and why]

**Stronger version:**
[rewritten prompt as one prose paragraph]

**Biggest single change:** [the one edit doing most of the work, in plain language]
```

---

# Slot vocabulary

The values each slot can take. These are values to choose between, not words to paste verbatim.

## 1. Reference & inspiration

| Approach | When to use |
|----------|-------------|
| Named photographer or style | Only if the user supplied one or it emerged from discovery. Never guess. |
| Described style | "Like a magazine editorial", "like a phone snapshot" — reliable and jargon-free |
| Negative direction | "Not a stock photo, not centred, not evenly lit" — fastest way to avoid defaults |

## 2. Subject

| Dimension | Values |
|-----------|--------|
| Count | One person, a pair, a group |
| Who | Age range, plus anything that matters |
| Expression | Where they're looking and what the face is doing |
| Action or still | Mid-motion (candid) vs. held pose (staged) — sets the register |

## 3. Wardrobe & styling

| Dimension | Values |
|-----------|--------|
| Register | Casual, formal, athletic, workwear, loungewear |
| Aesthetic | Retro, vintage, minimalist, maximalist |
| Colour & texture | Two or three colours; texture words (linen, denim, worn) add realism |

Consistency with environment matters more than specific garments.

## 4. Environment

| Dimension | Values |
|-----------|--------|
| Indoor/outdoor | Sets lighting expectations |
| Specific place | Kitchen, backyard, cafe, studio, porch — specifics carry their own props and light |
| Background detail | Busy/lived-in (documentary) vs. bare (editorial) |

## 5. Lighting

The most important slot. Always specify all four parts:

| Part | Values |
|------|--------|
| Source | Sun, overcast sky, window, flash, practical lamps |
| Direction | Behind, side, above, front — and specify which side (camera left/right) |
| Quality | Hard (defined shadows) or soft (shadows melt); warm or cool |
| Bounce | White wall, wooden floor, sand, concrete, grass — what fills the shadows |

**Six lighting conditions:**

| Condition | What it gives |
|-----------|---------------|
| Open sunlight | Hard, high contrast, defined shadows |
| Overcast | Soft, even, no shadow edge |
| Harsh midday | Deep shadows, blown highlights |
| Golden hour | Low, warm, long shadows |
| Blue hour | Cool, dim, atmospheric |
| On-camera flash | Flat, bright, hard falloff behind subject |

## 6. Composition & camera

| Dimension | Values |
|-----------|--------|
| Angle | Eye level, low (adds stature), overhead (flattens), over-shoulder |
| Framing | Close-up, medium/waist-up, three-quarter, full body, wide |
| Depth of field | Shallow (blurry background, subject pops) vs. deep (context visible) |

## 7. Mood & style

| Dimension | Values |
|-----------|--------|
| Tone | Candid, editorial, documentary, staged, lifestyle |
| Feeling | Joyful, intimate, dramatic, peaceful, confident, effortless |
| Colour treatment | Warm, cool, muted, vibrant |

## 8. Technical look (optional)

Only fill if the user mentioned film, vintage, or a specific camera. Otherwise skip — clean modern digital is the default.

| Dimension | Values |
|-----------|--------|
| Film stock | Portra 400 (warm, soft), Kodachrome (saturated), HP5 (B&W grain) |
| Camera/lens | 35mm f/1.4, 85mm, iPhone, medium format, Polaroid |
| Processing | Film grain, vintage fade, sharp digital |

---

# The template

For checking a prompt or filling by hand:

```
REFERENCE:   in the style of ____ / like ____ / not ____
SUBJECT:     ____, aged ____, ____ (expression), ____ (doing what)
WARDROBE:    ____ (register), ____ (aesthetic), ____ (colours, textures)
ENVIRONMENT: ____ (indoor/outdoor), ____ (specific place), ____ background
LIGHTING:    ____ (source) from ____ (direction), ____ (hard/soft, warm/cool),
             bouncing off ____
COMPOSITION: ____ (angle), ____ (framing), ____ depth of field
MOOD:        ____ (tone), ____ (feeling), ____ (colour treatment)
TECHNICAL:   ____ (film stock), ____ (camera/lens), ____ (processing)
```