AI image generation has changed how we think about visual creation.
A few years ago, creating a professional visual often required:
Camera
+
Lighting
+
Location
+
Subject
+
Lens
+
Editing
+
Design Software
Today, an AI image model can generate many of those visual decisions from natural language.
But there is a catch.
Better prompts don't simply contain more words. They contain better visual decisions.
That is why I started thinking about image prompting as a visual design system rather than a collection of random keywords.
From βMake an Imageβ to Visual Specification
Compare these two prompts.
Basic
Create a futuristic city.
The model has to invent almost everything.
Now consider:
A futuristic metropolitan city at blue hour,
wide-angle architectural photography,
dense vertical buildings,
reflective glass surfaces,
pedestrian-level perspective,
soft atmospheric haze,
controlled cyan and amber lighting,
deep depth of field,
cinematic composition,
high-detail environmental realism.
The second prompt defines:
Subject
+
Environment
+
Perspective
+
Lighting
+
Camera language
+
Atmosphere
+
Visual style
That's the fundamental idea behind these 100 lenses.
A Visual Prompt Is a Stack
I think a useful mental model is:
VISUAL IDEA
β
SUBJECT
β
ENVIRONMENT
β
COMPOSITION
β
CAMERA/LENS
β
LIGHTING
β
MATERIAL
β
ART DIRECTION
β
POST-PROCESSING
β
FINAL IMAGE
Not every prompt needs every layer.
But thinking in layers gives you much more control.
1. Photography Lenses
The first group focuses on photographic realism.
/photorealistic
/hyperrealistic
/cinematicphoto
/portrait
/headshot
/editorial
/fashion
/lifestyle
/streetphoto
/travelphoto
/wildlife
/macrophoto
These aren't simply βquality settings.β
They describe different visual goals.
For example:
Portrait
usually emphasizes:
- subject identity,
- facial lighting,
- expression,
- skin detail,
- composition.
Whereas:
Street photography
might emphasize:
- candid moments,
- environmental context,
- natural lighting,
- urban composition,
- documentary realism.
2. Perspective Is a Design Decision
The camera position can completely change an image.
Consider:
/topdown
You see the scene from above.
/lowangle
The subject can appear larger and more imposing.
/highangle
The subject may appear smaller or more vulnerable.
/wideangle
creates a stronger sense of spatial depth.
/telephoto
can compress the apparent distance between objects.
So instead of:
βShow a futuristic car.β
you could specify:
Low-angle automotive photograph,
camera positioned close to ground level,
wide environmental perspective,
dramatic city architecture in background.
Now the camera becomes part of the storytelling.
3. Depth of Field Controls Attention
Two useful lenses are:
/depthoffield
/bokeh
They are related but not identical.
A shallow depth of field can separate a subject from the background.
For example:
Professional portrait,
85mm portrait-lens look,
shallow depth of field,
soft background separation,
natural skin texture.
The visual hierarchy becomes:
SUBJECT
β
sharp
BACKGROUND
β
soft
This is useful because visual attention is partly controlled through focus.
4. Lighting Is One of the Strongest Prompt Variables
Lighting can completely change the emotional character of an image.
The toolkit includes:
/goldenhour
/bluehour
/sunset
/sunrise
/nightmode
/neon
/moody
/dramaticlighting
/volumetriclight
/rimlight
/silhouette
/backlit
For example:
Mountain landscape at golden hour,
warm directional sunlight,
long shadows,
soft atmospheric haze,
natural volumetric rays.
Compare that with:
Mountain landscape at blue hour,
cool ambient light,
deep shadows,
subtle mist,
cinematic atmosphere.
Same subject.
Completely different emotional result.
5. Lighting Should Describe Direction
Instead of simply saying:
βBeautiful lightingβ
try describing:
Key light
+
Fill light
+
Rim light
+
Light direction
+
Intensity
+
Color temperature
For example:
Portrait with soft key light from camera-left,
subtle fill light,
thin rim light separating the subject from the background,
natural skin tones,
controlled contrast.
That's much more actionable visually.
6. Film and Analog Aesthetics
The toolkit includes:
/blackandwhite
/filmgrain
/kodak
/fujifilm
/polaroid
/vintage
/retro
/80s
/90s
These are useful for specifying a visual era or photographic character.
But an important distinction:
A film aesthetic isn't only a color filter.
It can involve:
Color response
+
Contrast
+
Grain
+
Exposure
+
Highlight roll-off
+
Lens characteristics
+
Composition
For example, instead of:
Make it look vintage.
you could describe:
Analog-inspired editorial photograph,
subtle film grain,
soft highlight roll-off,
moderate contrast,
slightly muted colors,
natural imperfections,
archival photographic character.
7. Futuristic Worlds
The collection also contains environment-oriented lenses:
/cyberpunk
/steampunk
/dieselpunk
/solarpunk
/postapocalyptic
/fantasy
/scifi
These shouldn't be thought of as simple βstyle buttons.β
They describe world-building directions.
For example:
Solarpunk
Think:
Renewable energy
+
Green architecture
+
Human-centered cities
+
Natural materials
+
Clean technology
Cyberpunk
Think:
Dense urban environment
+
High technology
+
Strong artificial lighting
+
Weathered infrastructure
+
High visual contrast
The interesting part is understanding the visual ingredients behind the label.
8. Space and Natural Environments
The collection includes:
/space
/planet
/galaxy
/nebula
/underwater
/ocean
/mountains
/forest
/desert
/cityscape
These are primarily environment generators.
A useful environmental prompt can combine:
Subject
+
Scale
+
Atmosphere
+
Terrain
+
Lighting
+
Weather
+
Camera
Example:
Massive alien observatory on a rocky planet,
vast planetary horizon,
thin atmospheric haze,
distant nebula visible in the sky,
small human-scale structure emphasizing enormous environment,
cinematic wide-angle composition.
The scale relationship is what makes the scene believable.
9. Architecture and Interior Visualization
For architectural visualization:
/architecturephoto
/interior
/minimalinterior
/luxury
The prompt can specify:
Building geometry
+
Materials
+
Lighting
+
Camera height
+
Time of day
+
Interior styling
+
Environmental context
For example:
Minimal contemporary interior,
natural stone surfaces,
warm indirect lighting,
large floor-to-ceiling windows,
clean geometric furniture,
soft daylight entering from the right,
architectural photography,
balanced composition.
This is much closer to an architectural brief than a generic image prompt.
10. Product Visualization
Product-focused lenses include:
/productphoto
/packaging
/advertising
/billboard
/brandingmockup
A product image can be specified as:
Product
+
Material
+
Surface
+
Lighting
+
Camera
+
Background
+
Brand placement
+
Composition
For example:
Premium wireless headphones,
matte black finish,
minimal studio environment,
softbox lighting,
controlled reflections,
dark gradient background,
three-quarter product angle,
commercial product photography.
This gives the model a much clearer visual target.
11. Branding and Graphic Design
The collection also includes:
/logo
/iconset
/mascot
/characterdesign
These require a different type of thinking.
A logo prompt should focus more on:
Shape
+
Geometry
+
Symbolism
+
Typography
+
Scalability
+
Visual simplicity
A character prompt might focus on:
Silhouette
+
Proportions
+
Clothing
+
Expression
+
Color
+
Personality
The visual objective changes depending on the artifact.
12. Traditional Art
The toolkit includes:
/oilpainting
/watercolor
/acrylic
/charcoal
/pencilsketch
/inkdrawing
/crosshatching
These are useful when the desired output is not photographic.
For example:
Mountain village,
traditional watercolor illustration,
transparent layered washes,
visible paper texture,
soft edges,
limited color palette,
delicate atmospheric perspective.
The important part is to describe how the medium behaves, not just name the medium.
13. Illustration and Sequential Art
For illustrated storytelling:
/comicbook
/manga
/anime
the prompt can specify:
Character
+
Panel composition
+
Linework
+
Expression
+
Perspective
+
Lighting
+
Background
For a comic panel:
Dynamic comic-book panel,
hero standing on rooftop during heavy rain,
strong perspective,
expressive pose,
dramatic backlighting,
architectural city background,
clear foreground-midground-background separation.
The scene becomes much easier to control.
14. 3D and Game-Engine Visualization
The collection includes:
/lowpoly
/voxel
/pixelart
/clayrender
/3drender
/octanerender
/unrealengine
/blender
These represent different rendering and visual-development directions.
For example:
Low-poly
Simplified geometric forms
+
Flat surfaces
+
Controlled polygon count
+
Stylized lighting
Photorealistic 3D
Physically plausible materials
+
Realistic lighting
+
Surface detail
+
Depth
+
Accurate reflections
Clay render
Monochromatic material
+
Geometry emphasis
+
Soft studio lighting
+
Minimal textures
Again, the useful part is understanding the visual characteristics.
15. Prompt Engineering Itself
The final two lenses are particularly important:
/midjourneystyle
/imageprompt
The most useful one conceptually is:
/imageprompt
because it turns an idea into a structured visual specification.
A practical structure is:
SUBJECT
+
ACTION
+
ENVIRONMENT
+
COMPOSITION
+
CAMERA
+
LIGHTING
+
MATERIALS
+
COLOR
+
ATMOSPHERE
+
STYLE
+
QUALITY
For example:
SUBJECT:
Futuristic electric motorcycle
ENVIRONMENT:
Dense metropolitan city
COMPOSITION:
Three-quarter front view
CAMERA:
Low-angle wide-angle photography
LIGHTING:
Blue-hour ambient light with controlled neon reflections
MATERIAL:
Matte metal and carbon fiber
ATMOSPHERE:
Light rain and atmospheric haze
STYLE:
Premium cinematic automotive photography
DETAIL:
Highly detailed surfaces and realistic reflections
That is essentially a visual specification.
A Better Prompt Architecture
I would structure complex prompts like this:
ββββββββββββββββββββββββββββββββ
β SUBJECT β
ββββββββββββββββββββββββββββββββ€
β ACTION β
ββββββββββββββββββββββββββββββββ€
β ENVIRONMENT β
ββββββββββββββββββββββββββββββββ€
β COMPOSITION β
ββββββββββββββββββββββββββββββββ€
β CAMERA / LENS β
ββββββββββββββββββββββββββββββββ€
β LIGHTING β
ββββββββββββββββββββββββββββββββ€
β MATERIAL / TEXTURE β
ββββββββββββββββββββββββββββββββ€
β COLOR / PALETTE β
ββββββββββββββββββββββββββββββββ€
β ATMOSPHERE β
ββββββββββββββββββββββββββββββββ€
β ART DIRECTION β
ββββββββββββββββββββββββββββββββ
Not every image needs every layer.
But when an image is difficult to control, this gives you a systematic way to diagnose the problem.
Prompt Debugging
This is one of the most useful ways to think about AI image generation.
Suppose the result is wrong.
Don't immediately add 50 more adjectives.
Ask:
Wrong composition?
Change:
/composition
/camera
/perspective
Wrong mood?
Change:
/lighting
/moody
/goldenhour
/bluehour
Wrong realism?
Change:
/photorealistic
/material
/depthoffield
Wrong environment?
Change:
/architecturephoto
/cityscape
/forest
/desert
Wrong visual medium?
Change:
/oilpainting
/watercolor
/3drender
/pixelart
This is essentially debugging a visual specification.
The 100-Lens Visual System
The complete system can be represented as:
IDEA
β
βΌ
βββββββββββ
β SUBJECT β
ββββββ¬βββββ
β
βββββββββββββββΌββββββββββββββ
βΌ βΌ βΌ
ENVIRONMENT COMPOSITION CAMERA
β β β
βββββββββββββββΌββββββββββββββ
βΌ
LIGHTING
β
βΌ
MATERIALS
β
βΌ
COLOR
β
βΌ
ATMOSPHERE
β
βΌ
ART DIRECTION
β
βΌ
FINAL IMAGE
This is the core philosophy behind the 100 prompts.
The 100 AI Image Prompting Lenses
| # | Shortcut | Purpose |
|---|---|---|
| 201 | /photorealistic |
Ultra-realistic image style |
| 202 | /hyperrealistic |
Extreme realism |
| 203 | /cinematicphoto |
Movie-like photograph |
| 204 | /portrait |
Professional portrait |
| 205 | /headshot |
Corporate headshot |
| 206 | /editorial |
Magazine editorial look |
| 207 | /fashion |
Fashion photography |
| 208 | /lifestyle |
Lifestyle photography |
| 209 | /streetphoto |
Street photography |
| 210 | /travelphoto |
Travel photography |
| 211 | /wildlife |
Wildlife photography |
| 212 | /macrophoto |
Macro photography |
| 213 | /droneview |
Drone aerial view |
| 214 | /360drone |
360Β° aerial concept |
| 215 | /topdown |
Top-down composition |
| 216 | /lowangle |
Low-angle perspective |
| 217 | /highangle |
High-angle perspective |
| 218 | /wideangle |
Wide-angle lens look |
| 219 | /telephoto |
Telephoto compression |
| 220 | /fisheye |
Fisheye perspective |
| 221 | /depthoffield |
Shallow/deep focus control |
| 222 | /bokeh |
Soft background blur |
| 223 | /goldenhour |
Golden-hour lighting |
| 224 | /bluehour |
Blue-hour lighting |
| 225 | /sunset |
Sunset atmosphere |
| 226 | /sunrise |
Sunrise atmosphere |
| 227 | /nightmode |
Night photography |
| 228 | /neon |
Neon lighting |
| 229 | /moody |
Dark cinematic mood |
| 230 | /dramaticlighting |
Strong dramatic lighting |
| 231 | /volumetriclight |
Volumetric light |
| 232 | /rimlight |
Rim lighting |
| 233 | /silhouette |
Silhouette composition |
| 234 | /backlit |
Backlit subject |
| 235 | /blackandwhite |
Monochrome style |
| 236 | /filmgrain |
Film-grain aesthetic |
| 237 | /kodak |
Kodak-inspired film aesthetic |
| 238 | /fujifilm |
Fujifilm-inspired color aesthetic |
| 239 | /polaroid |
Instant-photo aesthetic |
| 240 | /vintage |
Vintage aesthetic |
| 241 | /retro |
Retro design |
| 242 | /80s |
1980s-inspired aesthetic |
| 243 | /90s |
1990s-inspired aesthetic |
| 244 | /cyberpunk |
Cyberpunk world-building |
| 245 | /steampunk |
Steampunk world-building |
| 246 | /dieselpunk |
Dieselpunk aesthetic |
| 247 | /solarpunk |
Sustainable-future aesthetic |
| 248 | /postapocalyptic |
Post-apocalyptic environment |
| 249 | /fantasy |
Fantasy artwork |
| 250 | /scifi |
Science-fiction artwork |
| 251 | /space |
Outer-space visual |
| 252 | /planet |
Planetary environment |
| 253 | /galaxy |
Galaxy visualization |
| 254 | /nebula |
Nebula artwork |
| 255 | /underwater |
Underwater scene |
| 256 | /ocean |
Ocean landscape |
| 257 | /mountains |
Mountain scenery |
| 258 | /forest |
Forest environment |
| 259 | /desert |
Desert environment |
| 260 | /cityscape |
Urban skyline |
| 261 | /architecturephoto |
Architectural photography |
| 262 | /interior |
Interior visualization |
| 263 | /minimalinterior |
Minimal interior concept |
| 264 | /luxury |
Luxury visual aesthetic |
| 265 | /productphoto |
Product photography |
| 266 | /packaging |
Packaging mockup |
| 267 | /advertising |
Advertising visual |
| 268 | /billboard |
Billboard mockup |
| 269 | /posterdesign |
Poster concept |
| 270 | /brandingmockup |
Brand identity preview |
| 271 | /logo |
Logo concept |
| 272 | /iconset |
Custom icon system |
| 273 | /mascot |
Mascot character |
| 274 | /characterdesign |
Original character design |
| 275 | /conceptart |
Concept art |
| 276 | /environmentart |
Environment concept art |
| 277 | /mattepainting |
Matte-painting-inspired environment |
| 278 | /digitalpainting |
Digital painting |
| 279 | /oilpainting |
Oil-painting medium |
| 280 | /watercolor |
Watercolor illustration |
| 281 | /acrylic |
Acrylic-painting medium |
| 282 | /charcoal |
Charcoal sketch |
| 283 | /pencilsketch |
Pencil drawing |
| 284 | /inkdrawing |
Ink illustration |
| 285 | /crosshatching |
Cross-hatching technique |
| 286 | /comicbook |
Comic-book visual language |
| 287 | /manga |
Manga-inspired visual language |
| 288 | /anime |
Anime-inspired visual language |
| 289 | /ghibli |
Hand-painted animated-film-inspired aesthetic |
| 290 | /pixar |
Polished 3D animated-film-inspired aesthetic |
| 291 | /lowpoly |
Low-poly 3D |
| 292 | /voxel |
Voxel art |
| 293 | /pixelart |
Pixel art |
| 294 | /clayrender |
Clay-render visualization |
| 295 | /3drender |
Photorealistic 3D render |
| 296 | /octanerender |
High-end physically based render aesthetic |
| 297 | /unrealengine |
Real-time-engine cinematic aesthetic |
| 298 | /blender |
Blender-style 3D visualization |
| 299 | /midjourneystyle |
Structured generative-art prompt direction |
| 300 | /imageprompt |
Convert an idea into a detailed image prompt |
How I Would Use the 100 Lenses
Don't use all 100 at once.
That's probably the fastest way to create an over-specified prompt.
Instead, choose one lens from each relevant category.
Example: Professional Portrait
/portrait
+
/depthoffield
+
/rimlight
+
/editorial
Example: Futuristic City
/cityscape
+
/wideangle
+
/bluehour
+
/scifi
+
/cinematicphoto
Example: Product Advertisement
/productphoto
+
/advertising
+
/dramaticlighting
+
/minimal
Example: Fantasy Environment
/environmentart
+
/fantasy
+
/volumetriclight
+
/wideangle
This creates a modular prompting system.
One Idea β Multiple Visual Directions
Suppose the base idea is:
βA futuristic electric motorcycle.β
You could generate completely different directions.
Cinematic
/cinematicphoto
/lowangle
/bluehour
Commercial
/productphoto
/advertising
/minimal
Sci-Fi
/scifi
/neon
/wideangle
Concept Art
/conceptart
/environmentart
/dramaticlighting
3D
/3drender
/blender
/rimlight
Same concept.
Different visual language.
The Real Skill Is Visual Decomposition
The most important skill isn't memorizing 100 commands.
It's learning to look at an image and ask:
What is the subject?
Where is the camera?
What lens perspective is being simulated?
Where is the light coming from?
What is sharp?
What is blurred?
What materials are visible?
What is the environment?
What creates the mood?
What visual medium is being used?
That is essentially visual reverse engineering.
And once you can decompose an image, you can reconstruct the idea as a prompt.
Final Framework
I think AI image prompting is moving from:
βWrite a creative sentence.β
toward:
βSpecify a visual system.β
A strong prompt can be thought of as:
IDEA
β
SUBJECT
β
ENVIRONMENT
β
COMPOSITION
β
CAMERA
β
LIGHT
β
MATERIAL
β
COLOR
β
ATMOSPHERE
β
STYLE
β
IMAGE
The 100 lenses are simply shortcuts for controlling different parts of that system.
The goal isn't to use more keywords.
The goal is to make better visual decisions before generating the image.
Question for the Community
If you could keep only 5 image-prompting lenses, which would you choose?
Would you prioritize:
/photorealistic
/cinematicphoto
/dramaticlighting
/wideangle
/conceptart
/3drender
/imageprompt
/productphoto
βor something else?
I'd love to hear how photographers, designers, developers, artists, and AI creators approach prompting differently.
Top comments (0)