What to Type for AI

Why Your AI Flat-Lay Looks Like a Furniture Catalog (And How to Fix It)

You write a detailed prompt, hit generate, and get back something that looks like a department store ad—props scattered randomly, no sense of intention, everything fighting for attention. The problem isn't your AI tool. It's that you're describing flat-lay composition in words when AI generators actually respond to spatial logic.

The clock-position system changes that. Instead of saying "arrange items artfully," you tell the AI exactly where each element goes: your main product at 12 o'clock, supporting props at 3 and 9, texture at 6. It's the same constraint professional stylists use, and AI generators understand it immediately. Most sellers who switch from free-form descriptions to positioned layouts see usable outputs on the first or second try instead of the eighth.

The Real Reason Flat-Lay Prompts Fail

AI generators have no spatial memory. You can describe 'elegant' and 'minimal' a dozen ways, but the model doesn't know that "minimal" means 4 props, not 12—or that the cream jar should be the visual anchor, not the background. What works is constraint. Professional photographers and stylists use a physical grid. Your prompt needs the same grid translated into language the AI can parse. That's not creative limitation; it's how you actually get control.

Clock-Position System: The Spatial Formula That Works

This is the core technique in the Shot Lists: a 6-position map for any flat-lay frame. 12 o'clock = main product (hero), 3 o'clock = first supporting element, 6 o'clock = texture/base surface detail, 9 o'clock = secondary prop, center = optional small accent. You write it into your prompt exactly that way. Example: "Top center (12 o'clock): white ceramic jar, slightly angled. Right side (3 o'clock): fresh green leaves. Bottom (6 o'clock): linen fabric fold." Generators respond to this directional language. You get composed shots, not scattered ones. The difference in generation success rate is immediate.

Category-Specific Surface and Lighting That Actually Work

A skincare flat-lay needs different modifiers than a jewelry layout, but most prompts just say "good lighting." The Shot Lists include tested pairings: skincare responds to 'soft diffused light, warm white marble' in ways that jewelry doesn't. Jewelry needs 'directional rim light, dark gradient background.' Home goods want 'natural window light, weathered wood.' These aren't preferences—they're the exact keyword combinations that make AI generators produce on-brand images without 10 failed attempts first. Copy them into your workflow and your generation time cuts in half.

The Documented Failure Modes (And Their Exact Fixes)

The Shot Lists catalog 10 common output problems: props that look photoshopped in, backgrounds that steal focus, 'floating' compositions that seem weightless, color wash-out, weird hand positions if you include a hand model. Each one has a documented rewrite fix. Props looking fake? Add 'shadows cast on surface' to your prompt—that alone grounds everything. Focus stolen by background? Swap 'blurred background' for 'solid neutral background, shallow depth of field, subject in sharpest focus.' These are small edits, but they're the difference between 'try again' and 'publish this.'

Pre-Generation Checklist: What to Verify Before You Hit Generate

Most sellers skip this step and then wonder why outputs fail. Before generating: Is your main product named specifically (not 'skincare bottle')? Are your positions assigned (clock system or directional)? Did you pick a specific lighting type, not just 'nice lighting'? Do you have 2-3 surface/texture modifiers? Did you exclude what you don't want ('no people, no text, no distracting elements')? A 30-second checklist prevents 90% of bad generations. The Shot Lists include a printable version you can use for every single prompt.

When to Use Ready-Made Templates vs. Custom Prompts

The Shot Lists provide 3 complete flat-lay templates (minimal editorial, warm lifestyle, luxury/high-contrast) plus 12+ category-specific examples. Use a template as-is when you're launching fast and your product fits the category exactly. Customize one when you have a specific aesthetic—swap modifiers, change the surface, adjust positions. The 6-part modular formula means you don't start from scratch; you modify working pieces. That's how you go from 15 minutes per prompt to 3 minutes.

FAQ

Will the clock-position system work with every AI image tool?

Yes. Midjourney, DALL-E 3, Stable Diffusion, and Adobe Firefly all respond to directional language. Generators that struggle with spatial precision benefit most from the system—it compensates for their weakness by being explicit. You might need to adjust phrasing slightly (some tools prefer 'left side' over '9 o'clock'), but the core logic is universal.

How many prompts do I need to write per product?

For a full launch, the Shot Lists recommend 8 image types: hero flat-lay, 2-3 styled flats, 1 detail close-up, 1 lifestyle scene, 1 lifestyle action, 1 context/environment. One prompt per type, so 8 total. If you're using templates, you're really just swapping product names and specific details—not writing from scratch.

What if my product doesn't match any of the 6 categories in the examples?

The 6-part formula works for any product. You follow the structure (subject + position + lighting + surface + style + exclusions), look up your closest category example for modifier reference, then customize. The formula is the core skill; the 30+ examples just show you what language works for common products.

Can I use these prompts for social media content, or just listing images?

Both. The shot-list templates are designed for listing images (standardized, neutral), but the mood keyword library and brand voice framework let you create lifestyle and social-specific variations. A candle brand, for example, uses one prompt for listing images and a modified version (swapping mood modifiers and framing) for Instagram content.

How much time does this actually save vs. hiring a photographer?

Per image: 3-5 minutes per successful prompt generation (including tweaks) vs. 1-2 hours of photographer time. For a 8-image product launch, you're looking at 30-40 minutes of prompt work plus generation time (usually 2-5 minutes per image in your tool of choice) vs. a day of shooting plus editing. The upfront learning curve is maybe an hour.

Related

AI Product Photo Editing Prompts →Prompt guides for product photo cleanup, batch editing, consistent catalog images, mockups, and studio-ready ecommerce visuals. How to Fix Blurry Product Photos in Photoshop Using AI Generative Fill →Learn to sharpen and rescale blurry product photos using Photoshop Generative Fill. Works on phone cameras and low-res shots. No retouching experience needed. Turn DIY Product Photos Into Studio Quality—Without a Studio →Learn to transform amateur product shots into professional studio photos using Photoshop's Generative Fill. Fast, no retouching skills needed. How to Make Product Photos Look Consistent When Shot on Different Cameras →Unify product photos shot on different cameras, days, and lighting. Use Photoshop Generative Fill to build catalog consistency in minutes per image. Photoshop Generative Fill Prompts: Product Photos to Studio-Ready in 3 Steps →Turn raw product photos into studio-ready catalog images at 15–20 per hour using Photoshop's Generative Fill — with prompts actually written for how Adobe... Pinterest Pin AI Prompts: 12 Copy-Paste Templates for 30-Day Growth →Stop losing 2 hours trying to get AI image tools to output something actually usable for Pinterest....