Fills the frame with a small detail, like texture, droplets or insect eyes, with a blurred background.
When to use it
For textures, insects, jewelry, food and product details.
Where it works
Not an official command: the model simply reads '/macro' as the words 'extreme close-up macro shot'. Typing '/macro' gives the same result as writing 'extreme close-up macro shot' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Asking for a whole scene in a macro shot.
Expecting everything to be sharp; macro has very little in focus.
Example
Dew drops on a green leaf, macro photography, extreme close-up, fine texture detail, very shallow depth of field
/flatlay
Also written: flat lay, top-down, overhead shot
A photo taken straight down from above, with items arranged on a surface.
What the AI does
Places objects neatly on a flat surface viewed from directly above.
When to use it
For food, products, desk setups and 'what's in my bag' images.
Where it works
Not an official command: the model simply reads '/flatlay' as the words 'flat lay, top-down view'. Typing '/flatlay' gives the same result as writing 'flat lay, top-down view' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Asking for people's faces, which don't work well from above.
Listing too many items, so it looks cluttered.
Example
Breakfast with coffee, croissant and fresh berries, flat lay, top-down overhead view, neatly arranged items on a clean surface
/closeup
Also written: close-up, closeup, close up
A shot that shows the subject large and near, like a face filling the frame.
What the AI does
Frames the subject tightly so details and expression stand out.
When to use it
For portraits, emotions and product details.
Where it works
Not an official command: the model simply reads '/closeup' as the words 'close-up shot'. Typing '/closeup' gives the same result as writing 'close-up shot' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Also asking to show the full body or the setting.
Not saying what should be in focus.
Example
An elderly fisherman smiling, close-up shot, subject fills the frame, sharp detail, softly blurred background
/wideshot
Also written: wide shot, long shot, full shot
A shot that shows the whole subject and much of its surroundings.
What the AI does
Pulls the camera back to show the setting and the subject's place in it.
When to use it
For landscapes, travel, establishing a scene and storytelling.
Where it works
Not an official command: the model simply reads '/wideshot' as the words 'wide shot'. Typing '/wideshot' gives the same result as writing 'wide shot' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Expecting fine detail on a small subject.
Also asking for a close-up.
Example
A hiker standing on a mountain ridge, wide shot, full scene visible, subject small in a large environment
/birdseye
Also written: bird's-eye view, aerial view, from above
A view from high above, as if from a bird or drone.
What the AI does
Places the camera high overhead, showing layout and patterns below.
When to use it
For cities, maps, crowds, landscapes and patterns.
Where it works
Not an official command: the model simply reads '/birdseye' as the words 'bird's-eye view'. Typing '/birdseye' gives the same result as writing 'bird's-eye view' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Confusing it with flat lay (close and straight down at objects).
Expecting faces to be visible.
Example
A colorful farmers' market from above, bird's-eye view, high aerial angle looking down, patterns and layout visible
/lowangle
Also written: low angle, worm's-eye view, from below
A shot taken from below, looking up at the subject.
What the AI does
Makes the subject look tall, powerful or impressive.
When to use it
For heroes, buildings, monuments and dramatic portraits.
Where it works
Not an official command: the model simply reads '/lowangle' as the words 'low-angle shot'. Typing '/lowangle' gives the same result as writing 'low-angle shot' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Using it when the subject should look small or vulnerable.
Also asking for a top-down view.
Example
A skyscraper against a clear sky, low-angle shot, camera looking up, subject appears tall and powerful
/eyelevel
Also written: eye level, eye-level, straight on
A shot taken at the subject's eye height.
What the AI does
Gives a natural, neutral, honest view of the subject.
When to use it
For portraits, interviews and everyday scenes.
Where it works
Not an official command: the model simply reads '/eyelevel' as the words 'eye-level shot'. Typing '/eyelevel' gives the same result as writing 'eye-level shot' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Expecting drama; eye level is neutral.
Mixing it with 'low angle' or 'high angle'.
Example
A barista handing over a cup of coffee, eye-level shot, camera at the subject's eye height, natural neutral perspective
/dutchangle
Also written: dutch angle, tilted angle, canted angle
A tilted camera, so the horizon is on a slant.
What the AI does
Tilts the frame to create tension, unease or energy.
When to use it
For thrillers, action and unusual, dynamic shots.
Where it works
Not an official command: the model simply reads '/dutchangle' as the words 'Dutch angle'. Typing '/dutchangle' gives the same result as writing 'Dutch angle' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Using it for calm or formal images.
Tilting so much the image looks like a mistake.
Example
A man running down a narrow alley, Dutch angle, tilted camera, slanted horizon, tense dynamic framing
/bokeh
Also written: bokeh, blurred background lights, background blur
Soft, round blurred lights in an out-of-focus background.
What the AI does
Blurs the background into soft circles of light so the subject stands out.
When to use it
For portraits, night scenes, festive and product shots.
Where it works
Not an official command: the model simply reads '/bokeh' as the words 'bokeh'. Typing '/bokeh' gives the same result as writing 'bokeh' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Expecting a sharp background too.
Using it with flat lay or wide landscapes.
Example
A glass of champagne at an evening party, beautiful bokeh, soft round blurred background lights, sharp subject
/shallowdof
Also written: shallow depth of field, depth of field, dof, blurred background
Only a thin slice of the image is in focus; the rest is blurred.
What the AI does
Keeps the subject sharp and blurs the foreground and background.
When to use it
For portraits and products where the subject should stand out.
Where it works
Not an official command: the model simply reads '/shallowdof' as the words 'shallow depth of field'. Typing '/shallowdof' gives the same result as writing 'shallow depth of field' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Asking for 'everything in focus' at the same time.
Not saying what should be sharp.
Example
A hand holding a cup of tea in a café, shallow depth of field, sharp subject, softly blurred background
/35mm
Also written: 35mm, 35mm lens, 35mm film
A classic camera lens/film look that shows the subject with some surroundings.
What the AI does
Gives a natural documentary or street-photo feel, sometimes with a film look.
When to use it
For street photos, travel and everyday storytelling.
Where it works
Not an official command: the model simply reads '/35mm' as the words '35mm lens photo'. Typing '/35mm' gives the same result as writing '35mm lens photo' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Expecting exact real-lens optics; it is only a style hint.
Mixing it with '85mm' or 'fisheye'.
Example
A street vendor selling flowers on a busy corner, shot on 35mm, natural perspective, documentary street photography feel
/85mm
Also written: 85mm, portrait lens, 85mm lens
A classic portrait lens look with a flattering face and blurred background.
What the AI does
Frames the subject tighter with soft background blur and natural proportions.
When to use it
For headshots and portraits.
Where it works
Not an official command: the model simply reads '/85mm' as the words '85mm portrait lens'. Typing '/85mm' gives the same result as writing '85mm portrait lens' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Using it for wide landscapes.
Expecting exact real-lens results; it is a style hint.
Example
A young chef smiling in a kitchen, 85mm portrait lens look, flattering perspective, creamy background blur
/fisheye
Also written: fisheye, fish-eye, ultra wide
An ultra-wide lens that curves straight lines into a bubble look.
What the AI does
Bends the edges of the image and fits a very wide view into the frame.
When to use it
For skate, music, playful and unusual perspectives.
Where it works
Not an official command: the model simply reads '/fisheye' as the words 'fisheye lens'. Typing '/fisheye' gives the same result as writing 'fisheye lens' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Using it for formal portraits or products.
Expecting straight lines to stay straight.
Example
A skateboarder doing a trick at a skate park, fisheye lens, ultra wide curved perspective, bulging distorted edges
/symmetry
Also written: symmetry, symmetrical, centered composition
A layout where both halves of the image mirror each other.
What the AI does
Centers the subject and balances the left and right sides.
When to use it
For architecture, interiors, portraits and calm, formal designs.
Where it works
Not an official command: the model simply reads '/symmetry' as the words 'symmetrical composition'. Typing '/symmetry' gives the same result as writing 'symmetrical composition' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Adding off-center elements that break the balance.
Expecting perfect mirror symmetry every time.
Example
A long hallway with arches, perfectly symmetrical composition, centered subject, balanced left and right
/ruleofthirds
Also written: rule of thirds, off-center composition
A layout that places the subject a third of the way across the image, not in the middle.
What the AI does
Puts the subject off-center to leave space and feel more natural.
When to use it
For landscapes, portraits and images with room for text.
Where it works
Not an official command: the model simply reads '/ruleofthirds' as the words 'rule of thirds composition'. Typing '/ruleofthirds' gives the same result as writing 'rule of thirds composition' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Also asking for 'centered'.
Expecting precise placement; generators follow it loosely.
Example
A lone tree in a wide field, rule of thirds composition, subject placed off-center, open space on one side
/negativespace
Also written: negative space, empty space, copy space
Large empty areas around the subject.
What the AI does
Keeps a lot of clean, empty background around a small subject.
When to use it
For minimal designs, ads and images that need space for text.
Where it works
Not an official command: the model simply reads '/negativespace' as the words 'negative space'. Typing '/negativespace' gives the same result as writing 'negative space' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Not saying where the empty space should be (top, left).
Adding busy details that fill the space.
Example
A single red umbrella on a white beach, minimal composition, lots of negative space, small subject, clean empty background
/longexposure
Also written: long exposure, light trails, silky water
A photo with a slow shutter that blurs movement over time.
What the AI does
Turns moving water silky, clouds streaky and car lights into trails.
When to use it
For waterfalls, seascapes, night traffic and star trails.
Where it works
Not an official command: the model simply reads '/longexposure' as the words 'long exposure photo'. Typing '/longexposure' gives the same result as writing 'long exposure photo' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Expecting sharp moving people.
Using it for simple product photos.
Example
A waterfall flowing over mossy rocks, long exposure photography, silky smooth water, streaking clouds, light trails
/motionblur
Also written: motion blur, speed blur, movement blur
Blur that shows fast movement.
What the AI does
Blurs moving parts so the image feels fast and energetic.
When to use it
For sports, cars, dancing and busy streets.
Where it works
Not an official command: the model simply reads '/motionblur' as the words 'motion blur'. Typing '/motionblur' gives the same result as writing 'motion blur' in plain words. Works in GPT image, Gemini/Imagen, Midjourney, Stable Diffusion and AI Compass's generator.
Common mistakes
Blurring the main subject's face when it should be sharp.
Using it for still life.
Example
A cyclist racing through a city street, dynamic motion blur, sense of speed, blurred moving background
Every guide and dictionary card is also in the AI Compass app, with saved items, search and “Try it” buttons that open chat, image or video with the example filled in. Reading is free; creating uses credits, and the price is shown before you start.