Prompt Library
AI Visual
Direction
Prompt Pack
50 tested image generation prompts for creative operators and brand strategists. Built for Ideogram, Adobe Firefly, and Higgsfield. Every prompt includes variables, tool recommendation, and pro tips from real production.
What's inside
Every prompt organized
by job type.
Not a dump of ideas. A library you navigate like a creative director would — by what you need to produce right now.
Category 01
Brand Identity Visuals
Hero images, logo concepts, color palette boards, mockups, and moodboard direction. Everything needed to build and present a visual brand system.
Category 02
Social Content Stills
Carousel backgrounds, quote cards, lifestyle stills, announcement graphics, and story highlights. Platform-ready across Instagram, LinkedIn, and TikTok.
Category 03
Campaign Concept Art
Launch visuals, event posters, ad creatives, email headers, and website heroes. Built for campaigns that need a strong visual center.
Category 04
Artist & Talent Visuals
Album covers, single artwork, promo visuals, music video concepts, EPK imagery, and merch design. Specific to music artists and independent creators.
Category 05
Cinematic Textures + Backgrounds
Dark atmospheric backgrounds, neon textures, fluid art, raw material surfaces, and cinematic skies. Build a reusable visual asset library.
Category 06
Creative Workflow Prompts
Claude prompts for briefs, captions, content calendars, email sequences, carousel copy, and brand voice guides. Text-based production tools.
Sample prompts
This is what you're buying.
Every prompt includes the variables to swap, the tool to use, and tips from real testing. Not generic. Not theoretical.
Art direction: [reference aesthetic — e.g. Caravaggio chiaroscuro / A24 muted realism / Saul Bass graphic reduction / Hype Williams fisheye era / Gregory Crewdson suburban noir]
Primary light source: single [sodium vapor orange / tungsten warm / blue hour ambient / practicals only — no fill]. Light direction: [45° camera-left / overhead / motivated from window right frame].
Color science: shadows crushed toward [hex — e.g. #0D0A14 blue-black / #100800 warm black]. Midtone hue shift: [warm +15 / cool -20 / neutral]. Highlights: [blown to paper white / retained at 90% / lifted toward #E8E0D0]. Saturation: [desaturated -40 / fully saturated / split-tone].
Grain: [Kodak 5219 35mm coarse / Ilford HP5 pushed 2 stops / clean digital]. Lens: [85mm f/1.4 shallow / 35mm f/8 deep / anamorphic 2x squeeze with horizontal streak flares].
Composition: [subject rule-of-thirds left, negative space right / centered symmetry / dutch angle -15°]. Typography clearance: 18% margin top, 12% bottom, 8% each side — keep zone free of detail. Aspect: 1:1.
Negative: no AI-typical lens flare, no symmetrical face, no beauty retouching, no HDR tone mapping, no stock-photo composition, no gradient backgrounds.
Why this works: Art direction reference locks the entire visual language before a single pixel generates. Hex codes for shadow density replace vague "dark" instructions. Typography clearance percentages are the difference between an album cover you can title and one where text fights the image. The negative prompt blocks the five most common AI failures by name.
Base: [deep field — #080810 / #0A0A0D / #050508]. No gradient. No vignette from center. Color field must be flat and even — like a perfectly processed matte C-41 negative base.
Atmospheric layer: [micro-particle fog at 8% opacity / directional smoke drift from bottom-left / heat distortion shimmer horizontal]. Layer must be visible but not dominant — texture, not effect.
Accent bleed: [#01DDFE cyan / #7C3AED violet / #FB923C amber] emitting from [upper-right corner / bottom-left edge / horizontal midpoint]. Spread radius: 15% of frame. Hard edge cutoff, no soft gradient bloom.
Grain: ISO 3200 analog noise — visible at 100% zoom, invisible at scroll size. Grain must be monochromatic, not color noise.
Typography readiness: image must hold white text [#F0F0ED] at opacity 100% without additional overlay. Test: if you placed 48px bold text at center, would it read? If not, darken the midtones.
Generate 4 variations: hold base hex constant, vary only atmospheric layer direction and accent bleed position.
Negative: no center glow, no light leak, no bokeh circles, no recognizable objects, no faces, no text.
Why this works: Specifying "C-41 negative base" gives the model a photographic reference for flat darkness — vague "dark background" produces inconsistent results. ISO 3200 grain instruction produces visible texture at 100% zoom that disappears at feed size, which is exactly what carousel backgrounds need. The typography readiness check is built into the prompt so you test before using, not after.
Subject: [describe with material precision — not "a bottle" but "a 750ml amber glass bottle with matte black label, no existing text visible"].
Lighting rig: [single overhead octabox 60cm, subject 2 stops brighter than background / Rembrandt triangle — key at 45° camera-left, fill at 1:4 ratio camera-right / motivated window light — large soft source camera-right]. No rim light unless specified.
Surface: [honed Calacatta marble, grey veining / matte gunmetal steel, brushed horizontal / oiled black concrete with micro-aggregate texture visible]. Material must catch specular highlight from key light.
Color science: primary palette [hex 1] + [hex 2]. Background must be [hex 3] ±10% — no gradient, no lighting contamination on background. Subject and background lit independently.
Depth of field: [shallow — subject sharp, background f/1.8 bokeh equivalent / deep — full scene sharp f/11 equivalent / hyper-shallow — only front surface sharp, everything behind out].
Text clearance: [left 40% of frame clear for headline / bottom 25% clear for subtext / full bleed — no clearance].
Commercial grade. No AI-typical plastic sheen. Photorealistic at 100% zoom.
Negative: no watermark, no text in image, no people, no hands, no lens flare.
Why this works: Naming the lighting rig by DP terminology (Rembrandt, octabox, key-to-fill ratio) produces dramatically more controlled light than "dramatic lighting." The surface material instruction forces the model to think about how light behaves on a specific material. Background hex ±10% with no gradient contamination is the instruction that separates professional product photography from generic AI output.
How it works
Buy once.
Use indefinitely.
No course to finish. No software to learn. Open the library, find the use case, copy the prompt, generate.
Purchase and get instant access
You pay once. You receive a Notion workspace link and PDF download immediately. No waiting. No manual fulfillment.
Duplicate the Notion workspace
One click and the full library lives in your own Notion account. Organized by category. Every prompt has its own entry with variables, tool recommendation, and tips.
Find your use case and copy the prompt
Browse by category. Find the job you need to do. Replace the variables in brackets with your project specifics. Read the tips before generating.
Paste into Ideogram, Firefly, or Higgsfield
Every prompt specifies which tool to use and why. Generate 4 variations. Select the strongest. The tips section tells you what to adjust if the first pass misses.
Deliver faster, charge the same
Work that used to take 3 hours of iteration now takes 30 minutes. The time savings compounds across every project. Your effective hourly rate increases from day one.
Who this is for
Built for operators.
Not beginners.
This pack was built for people who already know what they need to produce and want to get there faster. Not for someone learning AI tools for the first time.
Solo creative studios
One person running client projects across brand, content, and campaigns. You need output that looks like a team made it.
Brand strategists
You work upstream of execution. These prompts give you tangible visual direction to show clients before production starts.
Freelance designers
You use Adobe CC daily. Firefly is in your subscription. These prompts unlock what it can actually produce for real client work.
Artist managers
You need cinematic visuals for your artists without a photographer on retainer. This is the fastest path to professional output.
Content creators
You post 4–5 times a week and visual consistency is what separates a hobby account from a brand.
Small agency teams
You need a shared prompt library that produces consistent results across different team members.
Questions
Straight answers.
Do I need Midjourney?
No. Every prompt is written for Ideogram, Adobe Firefly, or Higgsfield. Ideogram has a free tier. Firefly is included in every Adobe CC subscription. Nothing extra to buy.
How is this delivered?
Immediately after purchase you receive a Notion workspace link and a PDF download. Duplicate the Notion template into your own account. It lives there permanently. The PDF is yours to keep and print.
Can I use these for client work?
Yes. The prompts are yours to use for personal and commercial projects. Adobe Firefly outputs are commercially safe by design. Check Ideogram and Higgsfield's current terms for your specific commercial use case.
What if new AI tools come out?
When new tools worth covering are added, you get the updated version free. The prompt structure works across most text-to-image tools — switching tools usually only requires adjusting the aspect ratio.
I'm not a designer. Can I still use this?
This pack was built for people already using AI tools for professional work. If you've never generated an AI image before, start there first. If you're already in Ideogram or Firefly and want better results faster, this is for you.
Stop wasting
time on bad
prompts.
50 tested prompts. Built from real studio production. Organized so you can find what you need and generate in under 60 seconds.
Get the Prompt Pack — $47