concept

Image prompt syntax

Image prompt syntax is the model-specific structure for steering diffusion outputs — Midjourney's `--ar`, `--style`, `::` weights; SD's negative prompts and prompt weights `()` `[]`; natural-language for Flux + Imagen. Each model has its own dialect.

Image prompting is not portable: a Midjourney prompt with `--ar 16:9 --style raw --niji 6` means nothing to SDXL; a Stable Diffusion prompt with `(beautiful:1.4)` confuses Flux. By 2026 the dialects: Midjourney v7 uses `--` parameters + `::` weights + natural language; SDXL / Stable Diffusion 3 uses bracket weights + negative prompts + structured tags; Flux + Imagen take fluent natural language (no special syntax — write like an art director); ComfyUI graphs add prompt scheduling across diffusion steps. Migrating prompts across models requires translation; the prompt that wins one model can flop on another. Cheatsheet pages (per-model) and prompt-translation tools are the practical workaround.

When to use image prompt syntax

Common mistakes

FAQ

What is image prompt syntax?

Image prompt syntax is the model-specific structure for steering diffusion outputs — Midjourney's `--ar`, `--style`, `::` weights; SD's negative prompts and prompt weights `()` `[]`; natural-language for Flux + Imagen. Each model has its own dialect.

When should I use image prompt syntax?

Picking the right prompt structure per model.

What are the most common mistakes with image prompt syntax?

Pasting an SD-style prompt into Flux — weights and tags get ignored or confused. Writing a paragraph for Midjourney — better with structured `subject, style, lighting --ar 16:9 --style raw`.

Last updated: 2026-06-01. Raw markdown: https://promtable.com/glossary/image-prompt-syntax.md.