skillZs
LIVE SKILL TAGS
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
REAL INSTALL DATA
← back to all skills
annahickman-maker/web-design-os3 installs

ai-image-prompting

Expert-level prompt engineering for AI image generation, specifically for brand photography, moodboards, strategy deck imagery, website hero images, and persona portraits. Use when generating image prompts for Nano Banana (Gemini), Midjourney, or any image model. Covers cinematic language, photography terminology, composition, lighting, film references, consistency across batches, and avoiding AI artifacts.

How do I install this agent skill?

npx skills add https://github.com/annahickman-maker/web-design-os --skill ai-image-prompting
view source ↗

Is this agent skill safe to install?

  • Gen Agent Trust Hubpass

    This skill is a comprehensive educational resource for improving AI image generation prompts. It contains no executable scripts, network requests, or data access operations.

  • Socketpass

    No alerts

  • Snykpass

    Risk: LOW · No issues

What does this agent skill do?

AI Image Prompting

The difference between a moodboard that feels like stock clichés and one that feels like a premium editorial shoot is almost entirely in the prompts. Generic prompts produce generic images. Specific, cinematic, well-directed prompts produce images that look like real photography.

This skill teaches the exact language, structure, and patterns for prompting AI image models to generate premium brand imagery.


Core principle

Prompt like an art director briefing a photographer, not like a user describing what they want. Art directors specify: composition, lighting, lens choice, colour grading, mood, film reference, subject direction. Users specify: "a happy woman in an office."

The art director prompt produces editorial images. The user prompt produces stock photo clichés.


The prompt structure

Every strong image prompt has 5-7 of these components:

  1. Subject - What's in the frame
  2. Composition - How it's framed
  3. Camera / lens - Technical specification
  4. Lighting - Light direction, quality, colour
  5. Mood / atmosphere - Emotional tone
  6. Colour treatment - Palette, grading
  7. Style reference - Photographer, film, publication, era

Example of a bad prompt:

"A woman drinking coffee"

Example of a strong prompt:

"Candid medium shot of a woman in her 30s drinking espresso at a marble countertop in a minimal kitchen, shot on Kodak Portra 400, natural morning light from the left window creating soft shadows, warm earth tones, shallow depth of field, editorial lifestyle photography in the style of Kinfolk magazine"

Same subject. Completely different output.


Photography vocabulary

These terms dramatically improve AI image quality. Use them.

Shot types

  • Close-up - face fills frame, intimate
  • Medium close-up - head and shoulders
  • Medium shot - waist up
  • Medium long shot - knees up
  • Full shot - whole subject
  • Wide shot - subject in environment
  • Extreme wide shot - landscape with tiny subject
  • Over-the-shoulder - framing through someone's shoulder
  • Bird's eye - looking straight down
  • Worm's eye - looking straight up
  • Dutch angle - tilted, creates unease

Lens types

  • 85mm - classic portrait lens, flattering compression
  • 50mm - "nifty fifty", natural perspective, shallow DOF
  • 35mm - wider, environmental, documentary feel
  • 24mm - wide angle, dramatic perspective
  • Macro - extreme close-up, texture detail
  • Anamorphic - cinematic widescreen with oval bokeh

Depth of field

  • Shallow depth of field - blurred background, subject isolation
  • Bokeh - the blurred background quality (creamy, busy, oval)
  • Deep depth of field - everything in focus
  • Rack focus - focus shift (rarely used in still prompts)

Composition rules

  • Rule of thirds - subject on a third-line intersection
  • Centered composition - subject dead center (often more modern)
  • Leading lines - lines guiding the eye to subject
  • Negative space - lots of empty space around subject
  • Framing - using architecture or objects to frame subject
  • Symmetry - balanced composition
  • Asymmetry - intentional imbalance
  • Low horizon - sky takes most of frame
  • High horizon - ground/floor takes most of frame

Lighting vocabulary

Lighting is the single biggest factor in whether AI images look real or generic. Specify it.

Light direction

  • Natural light from the left/right/above
  • Window light - soft, directional
  • Backlit - light behind subject, creating rim light
  • Side-lit - half of face in shadow (dramatic)
  • Split lighting - half face bright, half dark
  • Rembrandt lighting - triangle of light on shadow side of face
  • Loop lighting - small nose shadow (natural portrait)
  • Butterfly lighting - shadow under nose (glamour)

Light quality

  • Soft light - diffused, gentle shadows
  • Hard light - direct, sharp shadows
  • Diffused - filtered through cloud cover or softbox
  • Dappled light - broken through leaves
  • Harsh midday light - direct sun, dramatic shadows

Time of day

  • Golden hour - warm, low-angle, magical (hour after sunrise / before sunset)
  • Blue hour - cool, soft (30 min before sunrise / after sunset)
  • Morning light - cool, fresh, energetic
  • Midday light - bright, direct, editorial
  • Afternoon light - warm, sideways
  • Dusk - moody, low light, rich colours

Light colour

  • Warm light - yellow/orange cast
  • Cool light - blue/cyan cast
  • Neutral light - white balance correct
  • Mixed light - multiple colour temperatures

Film / camera references (hugely improve quality)

Using specific film stock references tells the model you want authentic photography, not digital AI output.

Film stocks with distinctive looks

  • Kodak Portra 400 - warm skin tones, creamy, editorial default (most versatile)
  • Kodak Portra 800 - warmer, grainier, low-light magic
  • Kodak Ektar - saturated, vivid, landscape
  • Kodachrome - vintage, warm reds, documentary
  • Fujifilm Pro 400H - cool, blue-green, fashion
  • Fujifilm Provia - accurate colour, clinical
  • Cinestill 800T - tungsten balance, neon glow, night work
  • Kodak Gold 200 - warm, nostalgic, casual
  • Ilford HP5 - classic black and white, documentary
  • Kodak Tri-X - high contrast black and white

Camera references

  • Hasselblad medium format - fashion editorial default
  • Leica M rangefinder - documentary, photojournalism
  • Mamiya 7 - medium format, sharp, editorial
  • Contax T2 - point and shoot, candid
  • Polaroid / instant - square format, soft focus, nostalgic
  • Large format 4x5 - highly detailed, commercial

Modern digital aesthetics

  • Sony A7 look - clean, modern, high resolution
  • Fujifilm digital - warm, film-like
  • Canon 5D - classic DSLR look

Style references

Reference specific photographers, magazines, or aesthetic movements.

Magazine / publication references

  • Kinfolk - minimal, natural light, muted earth tones, hand-held
  • Cereal - architectural, travel, minimalist
  • The Gentlewoman - editorial fashion, natural, elegant
  • Apartamento - interior, lived-in, intimate
  • Toast Magazine - warm, tactile, considered
  • Magnum Photos - photojournalism, documentary
  • i-D Magazine - fashion, editorial, experimental
  • Vogue editorial - high fashion, dramatic, stylised

Photographer references (use for specific moods)

  • Wolfgang Tillmans - intimate, casual, documentary
  • Roe Ethridge - commercial-editorial crossover, saturated
  • Petra Collins - nostalgic, pink, soft
  • Tim Walker - fantasy, elaborate, fashion
  • Annie Leibovitz - portrait, cinematic, storytelling
  • Cass Bird - fashion, androgynous, strong
  • Jamie Hawkesworth - documentary, fashion, editorial
  • Alasdair McLellan - youthful, summery, British

Aesthetic movements

  • Scandinavian minimalism - muted, spacious, tactile
  • Japanese wabi-sabi - imperfect, natural, aged
  • Modernist architecture photography - clean lines, high contrast
  • Slow living aesthetic - warm, unhurried, natural
  • Cottagecore - pastoral, nostalgic, soft

Prompting for brand photography (not portraits)

Most brand photography is NOT traditional portraits. It's lifestyle, product, environmental, or abstract. Adapt the prompts accordingly.

For product photography

  • Studio with soft light - even, professional
  • Marble surface - luxury, editorial
  • Linen texture - organic, natural
  • Warm wood grain - tactile, warm
  • Minimalist white - clean, commercial
  • Dark moody backdrop - dramatic, luxury

Example:

"Ceramic ceramic espresso cup on a hand-thrown linen cloth, shot from slightly above, soft natural morning light, warm earth tones, shallow depth of field with the background slightly blurred, food editorial photography in the style of Cereal magazine, Kodak Portra 400 film look"

For environmental / lifestyle

  • Subject interacting with environment
  • Wide shots that show context
  • Hands, feet, details rather than faces (abstraction)
  • Natural light sources

Example:

"Wide shot of a woman's hands knitting wool on a wooden table in a sunlit workshop, shot on Hasselblad medium format, morning light from a window on the left, warm earth tones and soft shadows, shallow depth of field, editorial craft photography, Kinfolk magazine aesthetic"

For architectural / interior

  • Wide angle to capture space
  • Natural light whenever possible
  • Clean lines, strong composition
  • Specific architectural details

Example:

"Minimalist modern interior, morning light streaming through floor-to-ceiling windows, wide angle composition showing a cream linen sofa, raw wooden floors, a single ceramic vase with dried eucalyptus, warm earthy palette, shot on a 24mm lens, editorial interior photography, Cereal magazine aesthetic"

For abstract / texture

  • Macro details
  • Textures as subject
  • Patterns and rhythm
  • Close-up focus

Example:

"Macro detail shot of raw linen fabric, natural light raking across the surface at a low angle, warm neutral tones, shallow depth of field, fine texture detail, shot on Kodak Portra 400, editorial texture photography for a slow living brand"


Consistency across a batch

Generating 12 moodboard images that feel cohesive requires consistent prompt elements across all of them.

The consistency template

For a batch of images, lock in:

  • Colour palette (use same palette words)
  • Light quality (same time of day / quality across images)
  • Film / camera reference (same stock)
  • Aesthetic reference (same magazine or photographer)
  • Mood (same emotional tone)

Vary:

  • Subject / composition (different angles and subjects for variety)
  • Depth of field (some shallow, some wide for visual interest)
  • Scale (some close, some wide)

Example consistent batch

Brand: wellness retreat, warm earthy palette, Scandinavian minimalism

Shared elements across all prompts:

  • "warm earth tones"
  • "natural morning light"
  • "Kodak Portra 400 film look"
  • "Kinfolk magazine aesthetic"
  • "shallow depth of field"

Varied elements:

  • Image 1: "close-up of hands holding a ceramic mug of herbal tea"
  • Image 2: "wide shot of a woman in linen walking along a misty forest path"
  • Image 3: "overhead flat lay of woven baskets with dried herbs on wood"
  • Image 4: "medium shot of a wooden sauna interior with steam rising"
  • Image 5: "macro detail of rough linen fabric texture"
  • Image 6: "silhouette of a figure in a doorway, backlit by warm light"

All 6 feel cohesive despite being different subjects.


Avoiding AI artifacts

AI image models have specific failure modes. Avoid them.

Common AI tells to prevent

  1. Extra fingers / wrong hands - Prompt "hands slightly out of frame" or focus on subjects that minimize hand detail
  2. Weird eyes - Avoid extreme close-ups of eyes, use medium shots
  3. Unnatural teeth - Prompt "closed mouth" or "lips slightly parted" instead of "smiling"
  4. Background chaos - Prompt for "simple background" or "negative space"
  5. Generic stock faces - Specify ethnicity, age, specific features, or avoid faces entirely
  6. Overly saturated colours - Specify "muted tones" or "desaturated"
  7. Over-rendered / plastic skin - Specify "film grain", "natural skin texture"
  8. Text in images - Avoid asking for text, AI can't spell well

Positive reinforcement words

Use these to improve quality:

  • "candid" (not posed)
  • "documentary" (authentic)
  • "editorial" (considered)
  • "film grain" (texture, not digital)
  • "natural light" (avoids artificial glow)
  • "shallow depth of field" (focus attention)
  • "authentic" (avoids uncanny valley)
  • "intimate" (emotional connection)
  • "unretouched" (natural skin)

Negative prompts (if model supports)

For models that accept negative prompts:

  • "plastic skin, overly smooth, doll-like"
  • "extra fingers, deformed hands"
  • "logo, watermark, text"
  • "stock photo, generic, cliché"
  • "oversaturated, HDR, cartoonish"

Matching AI images to brand tone

Different brand tones need different photography approaches.

Premium / luxury brand

  • Slower, more considered compositions
  • Marble, linen, ceramics as props
  • Soft natural light, never harsh
  • Muted warm palette
  • Editorial magazine aesthetic (Kinfolk, Cereal)
  • Shallow depth of field

Bold / energetic brand

  • Dynamic compositions, motion
  • Bright or high-contrast lighting
  • Saturated colours
  • Close-up detail shots
  • Fashion editorial aesthetic (i-D, Vogue)
  • Dramatic angles

Natural / wellness brand

  • Outdoor settings, natural environments
  • Dappled light, golden hour
  • Earth tones (no neons or saturated brights)
  • Hands, plants, water, simple objects
  • Slow living aesthetic (Kinfolk)
  • Shot on film stocks (Kodak Portra)

Tech / modern brand

  • Clean, minimalist compositions
  • Hard directional light, architectural
  • Monochromatic or limited palette
  • Geometric details
  • Modernist photography aesthetic
  • Crisp focus throughout

Artistic / creative brand

  • Experimental compositions
  • Mixed lighting (natural + coloured)
  • Bold or unusual palettes
  • Conceptual subjects
  • Fine art photography reference
  • Grainy, raw, authentic feel

Prompt templates by use case

Strategy deck cover (full-bleed hero)

[Environmental wide shot of subject doing brand activity] in [location matching brand mood], [specific lighting], [palette words], [film reference], [composition note - usually "negative space on [left/right] for title overlay"], editorial [industry] photography in the style of [magazine reference]

Strategy deck section divider

[Atmospheric wide shot with strong mood] of [subject in environment], [lighting], [palette], [film reference], negative space for title overlay in lower left, [editorial aesthetic reference]

Persona portrait

[Shot type] portrait of [age/gender/vibe description], [clothing], [expression/gesture - not posed], [environment], [lighting], [palette], [film reference], [editorial reference], authentic candid feel

Moodboard hero image

[Close-up or medium shot detail] of [specific brand-relevant object/scene], [lighting], [palette], [film reference], shallow depth of field, [mood], [aesthetic reference]

Website hero image

[Wide environmental shot] of [subject in brand-appropriate setting], [directional light with room for text overlay], [palette], [film reference], [aesthetic], composition leaves [left/right/bottom] clear for headline and CTA

About page imagery

[Intimate medium shot] of [subject working or engaged authentically], [natural lighting], [palette], [film reference], documentary candid style, shows personality and craft

Quality checklist

Before finalising a prompt, verify it includes:

  • Specific subject (not generic)
  • Shot type / composition
  • Lighting direction and quality
  • Time of day (if relevant)
  • Film or camera reference
  • Colour palette or grading note
  • Style / magazine / photographer reference
  • Mood / atmosphere
  • Negative space note (if for text overlay)
  • Anti-cliché words (candid, authentic, editorial)

A prompt that hits 7+ of these will produce significantly better results than one that hits 3.


Remember

AI image models respond to specificity, not verbosity. A 40-word prompt with the right words beats a 200-word prompt with generic language.

The most impactful words are:

  1. Specific film/camera reference (Kodak Portra 400)
  2. Specific magazine/photographer reference (Kinfolk magazine)
  3. Lighting direction and quality (soft morning light from the left)
  4. Shallow depth of field
  5. Editorial / candid / documentary (avoids stock feel)

If a prompt isn't working, add more of these and remove fluff.

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/annahickman-maker/web-design-os/ai-image-prompting">View ai-image-prompting on skillZs</a>