AI Image Generation Cheat Sheet

The Image Prompt Formula

Great image prompts follow a consistent structure. Not every element is required for every image, but this formula gives you maximum control over the output.

Formula: [SUBJECT] + [ACTION/POSE] + [SETTING/ENVIRONMENT] + [STYLE/MEDIUM] + [LIGHTING] + [CAMERA/COMPOSITION] + [MOOD/ATMOSPHERE] + [TECHNICAL PARAMETERS]

Element	What It Controls	Example
Subject	The main focus of the image	"A woman in her 30s wearing a navy blazer"
Action / Pose	What the subject is doing	"presenting to a small group, gesturing confidently"
Setting	Where the scene takes place	"modern glass-walled meeting room, city skyline visible"
Style / Medium	The artistic style	"editorial photography", "watercolour illustration"
Lighting	Light quality and direction	"soft natural window light", "dramatic studio lighting"
Camera	Perspective and framing	"shot on 85mm lens", "wide angle", "close-up"
Mood	Emotional tone of the image	"warm and inviting", "mysterious and moody"

Style Keywords Reference

Category	Keywords
Photography Styles	editorial, documentary, portrait, street photography, product photography, fashion photography, macro, landscape, architectural
Art Styles	oil painting, watercolour, pencil sketch, digital art, concept art, anime, art nouveau, art deco, impressionist, minimalist, pop art
3D & Rendering	3D render, isometric, clay render, unreal engine, octane render, low poly, voxel art, cinema 4D
Lighting	golden hour, blue hour, backlit, rim lighting, volumetric lighting, neon, candlelight, overcast, harsh midday sun, studio softbox
Mood & Atmosphere	ethereal, gritty, serene, dramatic, whimsical, dystopian, nostalgic, futuristic, cozy, cinematic
Composition	rule of thirds, symmetrical, bird's eye view, worm's eye view, tilt shift, bokeh, depth of field, panoramic, negative space

Tool-Specific Parameters

Midjourney

Parameter	What It Does	Example
`--ar`	Aspect ratio	`--ar 16:9` (widescreen), `--ar 9:16` (vertical/mobile), `--ar 1:1` (square)
`--v`	Model version	`--v 6.1` (latest)
`--s`	Stylisation (0–1000)	`--s 750` (high stylisation), `--s 50` (more literal)
`--c`	Chaos / variation (0–100)	`--c 30` (more variation between outputs)
`--no`	Negative prompt (exclude elements)	`--no text, watermark, blur`
`--q`	Quality (render time)	`--q 2` (higher quality, more GPU time)

ChatGPT Images 2.0

Conversational editing: Describe changes in plain language — "make the background darker" or "remove the person on the left"
Size options: 1024×1024 (square), 1792×1024 (landscape), 1024×1792 (portrait)
Style control: Add "natural" for photorealistic or "vivid" for more saturated, artistic outputs
Text in images: ChatGPT Images 2.0 handles text better than most — specify exact words in quotes
Iteration: Build on previous outputs: "Keep the same style but change the setting to a beach"
Reference images: Upload a reference and ask "generate an image in this style but showing [DESCRIPTION]"

Adobe Firefly

Commercial safety: All Firefly outputs are cleared for commercial use — ideal for brand assets
Style presets: Use built-in style filters: Photo, Art, Graphic, Digital Art
Content Type: Toggle between Photo and Art for fundamentally different outputs
Colour & Tone: Use the visual controls to adjust colour palette without re-prompting
Generative Fill: Select areas of existing images and replace with AI-generated content
Structure Reference: Upload a reference image to maintain composition while changing style

Example Prompts

Professional Headshot Style A confident [GENDER] professional in their [AGE RANGE], wearing [OUTFIT], photographed from chest up against a clean [COLOUR] background. Soft studio lighting, shallow depth of field, shot on 85mm lens. Professional corporate headshot style. Warm, approachable expression.

Blog Hero Image A flat-lay workspace composition featuring [ITEMS RELATED TO TOPIC] arranged on a [SURFACE — e.g. white marble desk, wooden table]. Top-down view. Soft natural light from the left. Clean, minimal editorial style. Muted colour palette with one accent colour ([COLOUR]). Negative space on [LEFT/RIGHT] for text overlay.

Social Media Graphic An isometric 3D illustration of [CONCEPT — e.g. a digital marketing workflow, an AI-powered office]. Clean, modern design with a [COLOUR SCHEME] palette. Subtle shadows, rounded corners on elements. Style: corporate illustration, friendly and approachable, similar to Notion or Stripe marketing visuals. No text in the image.

Iteration is everything. Your first prompt rarely produces the perfect image. Generate, evaluate, refine your prompt, and generate again. Most professionals go through 3–5 rounds before landing on the right output.