All prompts

Image generation prompts

Prompts for image models, written the way image models actually read.

Image prompts are not sentences; they are stacked constraints — lens and focal length, lighting, motion, surface, what must not change. Several here are edit prompts built to preserve a subject's face while everything around it moves, which is the hardest thing to hold steady. They are plain text rather than XML, because that is what these tools take.

4 prompts published

image

Neon Night Crossing — Light-Trail Portrait Edit

An image-edit prompt that plants your reference subject at a busy night crossing, still and sharp while the crowd and the traffic streak past. Locks the long-exposure headlight and taillight trails, wet asphalt reflecting them, deep blacks cut with cold neon blue and magenta, and an edge-to-edge frame.

30 0 0by Andrei Badulescu
Preview
image

Crowd Motion-Blur Portrait — 200mm Telephoto Edit

An image-edit prompt that drops your reference subject into a packed city street, tack-sharp while the crowd smears into long-exposure blur. Locks the 200mm compression, a frame filled edge to edge with no sky, a dark high-collar puffer and a cold desaturated grade. Closes with explicit do-nots so the face survives the edit.

28 0 0by Andrei Badulescu
Preview
image

Editorial Studio Portrait — Softbox Beauty Lighting

Turns a nine-word idea into a full studio brief. Locks the whole shot: voluminous dark waves, natural soft makeup, medium-close-up framing, diffuse softbox light with catchlights in the eyes, and a seamless neutral grey backdrop with soft falloff. Use it for model tests, beauty and fashion headshots, or clean editorial portraits — swap hair, wardrobe or backdrop colour and the lighting recipe holds. Works on Midjourney, Flux, Imagen and any image model that takes plain-text prompts.

50 0 0by Andrei Badulescu
Preview
image

Double-Exposure Portrait — Stormy Seascape, Face Untouched (Gemini)

Turn any portrait into a cinematic double exposure without touching the face. Hair dissolves into grey smoke and a flock of birds, the coat merges into dark turbulent water, telephone poles fade into fog on a low horizon. Moody teals and deep blues, dramatic side lighting, gritty film grain. Built for Gemini image editing: hard constraints lock bone structure, eyes, skin texture and expression so the likeness stays exact.

35 0 0by Andrei Badulescu
Preview

Guides on image generation prompts

Image Prompting: Midjourney, DALL·E, Flux & More

Use the Image use case to turn a rough idea into a plain-text prompt tuned for Midjourney, DALL-E 3, Flux, and six other generators.

Questions about image generation prompts

Why are these plain text instead of the XML used elsewhere?
Image models do not parse XML; they read stacked constraints. So image prompts come back as plain text in the grammar of the target tool, with no preamble and no markdown, ordered subject, medium, style, lighting, composition, then parameters.
Do they carry tool-specific syntax?
When they were written for a tool that has it, yes: aspect and version flags for Midjourney, weighted terms for Stable Diffusion, plain descriptive phrasing for DALL-E 3, Flux and Imagen. A negative-prompt block appears only for the tools that accept one, which is Stable Diffusion and Leonardo.
Can I use a Midjourney prompt in a different tool?
The ordering transfers; the syntax does not. Strip the flags, or paste the idea back into the optimizer with your own tool selected and let it re-emit the same intent in that grammar.