Imageprompt: The Complete Reference Library

An imageprompt is a natural-language instruction sent to a text-to-image model. The compound spelling resolves three overlapping things: (a) the generic concept, (b) ImagePrompt.org, the standalone utility, and (c) the broader category of prompt-engineering tools.

imageprompt

Single-word form. Generic noun for any text instruction sent to a text-to-image model (DALL-E, Flux, SD3, Imagen).

image prompt

Two-word form. Same meaning, more common in academic writing (see Oppenlaender's taxonomy).

image-to-prompt (i2p)

Inverse process: upload an image, get a text prompt back via CLIP / BLIP / vision-LLM.

ImagePrompt.org

Specific commercial tool. Capital I, capital P. Offers prompt generation and i2p against a small set of model presets.

Per Oppenlaender's taxonomy in Behaviour & Information Technology (2023), imageprompts decompose into six modifier categories — subject terms, image prompts, style modifiers, quality boosters, repeating terms, and "magic terms" — and the choice of style vocabulary is the largest lever on output quality.^[1]

Each row was chosen because real searches for "imageprompt" surface it. ZeroTwo is included in its honest column — multi-model side-by-side is its only meaningful differentiator.

The seven-part structure works across DALL-E 3, Stable Diffusion 3, Flux, Imagen, Ideogram, and Midjourney. The order matters less than coverage: leaving any slot blank is the most common failure mode in our review of public prompt galleries.

• 030.01
  Subject
  Who or what — concrete noun phrase
• 030.02
  Style
  Artistic register — e.g. 35mm film, watercolor, isometric
• 030.03
  Composition
  Framing — wide shot, close-up, rule of thirds
• 030.04
  Lighting
  Source + quality — golden hour, softbox, neon rim
• 030.05
  Medium
  Material reference — oil on canvas, octane render
• 030.06
  Modifiers
  Style boosters per Oppenlaender taxonomy
• 030.07
  Negatives
  What to exclude — blurry, extra fingers, watermark

The DALL-E 3 paper by Betker et al. (OpenAI, 2023) attributes much of the model's improvement to a caption re-writing pipeline that systematically expands short prompts.^[2] Stability's SD3 research paper shows that rectified-flow transformers respond strongly to structured token sequences.^[3]

• 040.01Portrait

  Studio portrait of a violinist mid-performance, Rembrandt lighting, 85mm f/1.4, charcoal backdrop, sweat-lit cheekbones, editorial color grade. Negative: HDR, plastic skin.

• 040.02Landscape

  Vast Patagonian valley at blue hour, granite spires, drifting fog, wide 16:9 panorama, Sigma 14mm, long exposure water, National Geographic palette. Negative: people, text.

• 040.03Product

  Matte-black wireless earbuds floating above a slate plinth, three-point softbox, subtle rim light, seamless cyclorama, commercial product still, hyper-clean reflections.

• 040.04Anime

  Anime keyframe of a cyber-samurai under neon rain, cel-shaded, thick line art, Studio Trigger color palette, dynamic dutch angle, motion lines. Negative: 3D, photoreal.

• 040.05Isometric

  Isometric tiny home cross-section, axonometric projection, soft pastel palette, paper-cut layers, ambient occlusion, Monument Valley game aesthetic.

• 040.06Cinematic

  Cinematic still: lone astronaut walking a Martian ridge at dusk, anamorphic lens flare, 2.39:1 aspect, Denis Villeneuve color science, dust haze, score-worthy framing.

• 040.07Logo

  Minimalist vector logo for a coffee roaster named 'Ember', single weight stroke, geometric sans-serif wordmark, warm copper accent, on cream background, flat 2D.

• 040.08Architecture

  Brutalist concrete library at golden hour, low-angle exterior, raw board-formed concrete, deep shadows, Tadao Ando influence, architectural photography, 24mm tilt-shift.

• 040.09Food

  Overhead flatlay of handmade ramen, soft-boiled egg cross-section, chashu, scallion confetti, marble counter, natural window light, food editorial styling.

• 040.10Sci-Fi

  Derelict generation ship interior, bioluminescent moss creeping over corroded bulkheads, volumetric god-rays, concept-art painterly style, Simon Stalenhag mood.

"Prompt engineering is a creative skill that requires style-specific vocabulary, not just description."

• 060.01An imageprompt is any text instruction sent to a text-to-image model — DALL-E 3, SD3, Flux, Imagen, Midjourney.
• 060.02ImagePrompt.org is one specific tool inside the broader imageprompt category. Both meanings coexist in search.
• 060.03The seven-part formula — Subject, Style, Composition, Lighting, Medium, Modifiers, Negatives — works across every modern model.
• 060.04Style modifiers, per Oppenlaender's 2023 taxonomy, are the single largest lever on output quality.
• 060.05A multi-model approach (running one imageprompt across 8+ generators) reveals which model best fits a given style — ZeroTwo does this in one tab for $19.99/month.

1. Oppenlaender, J. "A taxonomy of prompt modifiers for text-to-image generation." Behaviour & Information Technology, Taylor & Francis (2023).
2. Betker, J. et al. "Improving image generation with better captions" (DALL-E 3 paper). OpenAI (2023).
3. Stability AI. "Stable Diffusion 3 — Research Paper." stability.ai (2024).
4. "Prompting AI Art: An Investigation into the Creative Skill of Prompt Engineering." International Journal of Human–Computer Interaction, Taylor & Francis (2024).
5. Wang, Z. et al. "PromptCharm: Text-to-Image Generation through Multi-modal Prompt Engineering and Refinement." arXiv:2403.04014.

• → AI Image Prompts (Library)
• → How to Write Image Prompts
• → Best AI Image Generator
• → AI Image Generator from Text

Check out a ZeroTwo card.

60+ frontier models, 8+ image generators, one $19.99/mo account. Bring your imageprompt — leave with eight images.