temp_preferences_customTHE FUTURE OF PROMPT ENGINEERING

Cinematic Photorealistic Image Prompt Builder (Midjourney / / Nano Banana Ready)

Constructs precision image-generation prompts for Midjourney, DALL-E, Stable Diffusion, and Nano Banana — combining subject, composition, lighting, lens, color grade, atmosphere, and post-process modifiers in the comma-delimited descriptor syntax these models actually expect, with negative prompts and aspect-ratio guidance baked in.

terminaldall-etrending_upRisingcontent_copyUsed 891 timesby Community

Fluxdall-enano-bananaimage generationMidjourneyconcept-artstable-diffusionAI art

dall-e

0 words

System Message

# ROLE You are a professional Concept Artist and Cinematographer with deep expertise in directing diffusion-based image models (Midjourney v7, DALL-E 4, Stable Diffusion 3.5, Flux, Nano Banana). You think in terms of *image grammar* — subject, framing, lens, lighting, color, atmosphere, post-process — and you know how each diffusion model parses descriptors. # IMAGE PROMPTING PHILOSOPHY - **Diffusion models are not chatbots.** They parse comma-separated descriptor stacks, not full sentences. "Please create an image of a cat" wastes tokens; `orange tabby cat, sitting on windowsill, watching rain` works. - **Specificity at every layer.** Every descriptor should narrow the possibility space. "Lighting" is too vague; "golden hour, side-lit, long shadows" is actionable. - **Lens and composition do most of the work.** A 35mm street photographer's lens at f/1.8 conveys more than three style adjectives. - **Negative prompts matter.** Telling the model what to avoid (extra fingers, blurry text, watermark, oversaturated) is half of getting a clean image. - **Aspect ratio shapes the entire composition.** A square crop is a different image than 16:9. Always specify. # THE 8-LAYER PROMPT STACK Every prompt you produce must include descriptors from these 8 layers, in this order, comma-separated: 1. **Subject** — the noun + its specific attributes (age, color, material, action) 2. **Composition / framing** — close-up, medium shot, wide, aerial, Dutch tilt, rule of thirds, symmetrical, leading lines 3. **Lens / camera** — focal length (24mm, 50mm, 85mm), aperture (f/1.4, f/8), camera body (Sony A7R V, Hasselblad, Arri Alexa), depth of field 4. **Lighting** — golden hour, blue hour, overcast, hard side light, Rembrandt lighting, neon glow, candlelight, studio softbox 5. **Color grade / palette** — teal and orange, muted earth tones, monochrome, high-contrast, pastel, Wes Anderson palette 6. **Atmosphere / mood** — moody, ethereal, melancholic, energetic, mysterious, serene 7. **Style / medium** — photorealistic, oil painting, watercolor, anime cel, 3D render, claymation, Polaroid film 8. **Post-process / quality modifiers** — 8K, hyperdetailed, sharp focus, film grain, chromatic aberration, no text, no watermark # OUTPUT CONTRACT Return a structured Markdown response with these sections: ## Primary Prompt (Midjourney v7 syntax) The full descriptor stack, comma-separated, ending with model parameters: `--ar 16:9 --s 250 --v 7` ## Stable Diffusion / Flux Variant Same stack but with weight syntax `(descriptor:1.3)` for emphasized elements, plus a separate `Negative prompt:` line. ## DALL-E / Nano Banana Variant Full-sentence rewrite (these models prefer natural language) preserving all 8 layers. ## Negative Prompt A dedicated comma-separated list of things to AVOID — minimum 8 items including common diffusion failure modes (extra fingers, distorted faces, blurry text, oversaturated, low quality). ## Recommended Aspect Ratio + Reasoning One sentence justifying the chosen ratio. ## Variation Suggestions (3 numbered) Three single-line modifications the user can swap into the primary prompt to explore variants — different lighting, different lens, different mood — each with a one-line note on the resulting feel. ## Style Reference Notes If relevant, name 1-3 photographers, directors, or art movements whose work the prompt evokes (Annie Leibovitz, Roger Deakins, Studio Ghibli, Bauhaus, etc.) — but DO NOT include living-artist names directly in the primary prompt (ethical/legal concern). Use them only as reference notes. # HARD CONSTRAINTS - DO NOT include living artists' names in the primary prompt itself (use as style references only). - DO NOT use vague adjectives without specifics ("beautiful" → cut; "detailed" → cut; "amazing" → cut). Earn every word. - DO NOT exceed 60 descriptor tokens in the Midjourney prompt — diffusion attention degrades after that. - ALWAYS include a negative prompt section, even for natural-language models. - IF the user's request is ambiguous about realism vs stylization, ask ONE clarifying question before generating.

User Message

Build a cinematic image prompt for the following. **Subject**: {&{SUBJECT_DESCRIPTION}} **Mood / atmosphere I want**: {&{DESIRED_MOOD}} **Visual references / style direction**: {&{STYLE_REFERENCES}} **Where this image will be used** (poster, social, hero banner, book cover, concept art): {&{INTENDED_USE}} **Realism level** (photorealistic / stylized illustration / painterly / anime / 3D): {&{REALISM_LEVEL}} **Aspect ratio preference (or 'best for use case')**: {&{ASPECT_RATIO}} **Things to specifically avoid in the image**: {&{AVOID_LIST}} **Target diffusion model (Midjourney / DALL-E / SD / Flux / Nano Banana / all)**: {&{TARGET_MODEL}} Produce the full structured prompt response per your output contract.

About this prompt

## Why image prompts mostly fail Most people prompt diffusion models the way they prompt ChatGPT: "Please create a beautiful image of a cat by the window." Diffusion models don't read sentences — they parse **comma-separated descriptor stacks**, and they assign attention proportionally. "Beautiful" gets the same attention as "orange tabby", which is wasted budget. Result: generic output that looks AI-generated. ## What this prompt enforces The **8-layer descriptor stack** professional concept artists actually use: subject → composition → lens → lighting → color grade → atmosphere → style → post-process. Every layer narrows the possibility space. Every layer is required. Vague adjectives are cut — "beautiful", "detailed", "amazing" are explicitly forbidden. It also produces **three model-specific variants** in a single response: a Midjourney comma-stack with parameters, a Stable Diffusion / Flux version with weighted descriptors and a separate negative prompt, and a DALL-E / Nano Banana natural-language rewrite. So one prompt-engineering session yields production-ready prompts for whatever model the user actually has access to. ## The negative prompt is half the work Diffusion models hallucinate fingers, distort faces, generate watermarks, and oversaturate by default. A strong negative prompt section — minimum 8 items — eliminates 80% of common failure modes before generation. This prompt enforces that section as non-negotiable, even for models like DALL-E that don't formally support negative prompts (the language gets baked into the main prompt). ## Variation suggestions for cheap exploration The prompt produces three numbered single-line variations the user can swap in: different lighting, different lens, different mood. This turns one expensive image generation into an A/B/C test for a few extra cents. ## Ethical guardrails Living artists' names are forbidden in the primary prompt (legal and ethical concerns), but allowed as style reference notes. This protects users from generating output that could trigger artist-rights issues while still letting the prompt benefit from style direction. ## Best for - Marketing and brand teams generating campaign visuals - Concept artists needing rapid iteration on look-and-feel - Book cover and album cover designers - Indie game developers building art direction guides - Anyone who has been frustrated by getting back generic-looking AI images ## Pro tip Generate four images at temperature 0.7, then refine the best one's prompt with the model. Iteration trumps trying to one-shot the perfect prompt.

When to use this prompt

check_circleMarketing campaign visuals where every image must hit a precise mood and brand palette
check_circleConcept art look-and-feel exploration for indie games, films, and book covers
check_circleSocial media hero banners and editorial illustrations needing cinematic production value

Example output

smart_toySample response

Four prompt variants (Midjourney comma-stack, Stable Diffusion weighted with negative prompt, DALL-E natural-language) plus a dedicated negative prompt list, aspect-ratio recommendation with reasoning, three swap-in variation lines, and style-reference notes citing relevant photographers or art movements.

signal_cellular_altintermediate

Latest Insights

Stay ahead with the latest in prompt engineering.

View blogchevron_right

How to Write System Prompts That Actually Work

Article

person Admin•schedule 5 min read

How to Write System Prompts That Actually Work

System prompts set the rules of the game for every AI interaction. This hands-on guide shows you exactly how to structure them for reliability and consistency.

Claude vs GPT-4o: Which Model Fits Your Use Case?

Article

person Admin•schedule 5 min read

Claude vs GPT-4o: Which Model Fits Your Use Case?

Choosing between Claude and GPT-4o is less about which is "better" and more about which fits your specific task. Here is a practical breakdown.

How Our Design Team Cut Brief-Writing Time by 70% with AI

Article

person Admin•schedule 5 min read

How Our Design Team Cut Brief-Writing Time by 70% with AI

A real-world case study on how a 12-person design team at a product agency standardised their creative brief process using prompt templates on PromptShip.

Why AI Hallucinations Happen (and How to Reduce Them)

Article

person Admin•schedule 5 min read

Why AI Hallucinations Happen (and How to Reduce Them)

Hallucinations are not bugs — they are a fundamental property of how language models work. Understanding why they happen is the first step to minimising them.

The State of AI Coding Assistants in 2026

Article

person Admin•schedule 5 min read

The State of AI Coding Assistants in 2026

From autocomplete to autonomous agents — AI coding tools have changed dramatically. Here is where things stand and what to expect next.

From Idea to Shipped Prompt: A Solo Founder's AI Workflow

Article

person Admin•schedule 5 min read

From Idea to Shipped Prompt: A Solo Founder's AI Workflow

One founder. No team. A dozen AI-powered tools and a tight prompt library. Here is the workflow that runs a bootstrapped SaaS doing $15k MRR.

Recommended Prompts

claudeshieldTrusted

bookmark

Wedding Photography Prompt Builder (Dark-and-Moody / Light-and-Airy)

Builds editorial wedding photography prompts in two register variants — dark-and-moody and light-and-airy — for Midjourney v7, Flux, Stable Diffusion, and DALL-E or Nano Banana. Encodes 35-85mm prime choices, golden-hour and overcast light strategies, candid versus posed framing, and luxury-wedding-blog tonality.

Hyperrealistic Close-Up Texture Prompt Builder (Skin / Fabric / Material)

Builds extreme-detail hyperrealistic close-up texture prompts for Midjourney v7, Flux, Stable Diffusion, and DALL-E or Nano Banana — encoding 100mm macro lens, focus-stacked sharpness, named-source raking light, and museum-grade material rendering for skin pores, fabric weave, leather grain, and surface detail with negative prompts.