What to Type for AI

How to Generate Consistent Product Photos Across Your Etsy Listings Without Hiring a Photographer

The core problem: you photograph one ceramic mug beautifully, then the next listing looks like it came from a different store. Different background, different lighting angle, different mood. Your shop looks fragmented, which tanks trust and conversion.

The fix isn't to shoot everything over. It's to lock in your visual system once—then use AI + your phone to generate the variations that fit it. This works because you're not starting from nothing; you're starting from one real, on-brand photo you already made. AI fills in the 7 other angles and moods that match your actual product and your actual lighting setup.

Here's what separates working from broken: most people throw a product name at DALL-E and get chaos. You need to reverse-engineer your real photo into a prompt, then use that prompt as a template for variations. We show you the exact framework to do that.

Why One-Off Photos Break Your Shop's Visual Cohesion

When you photograph randomly—sometimes with window light, sometimes with a ring light, sometimes outdoors—your listings tell the customer that you're disorganized. More practically, Etsy's algorithm uses image quality and consistency as a ranking signal. A buyer scrolling your shop shouldn't feel like they're shopping three different sellers. Cohesion is especially critical for jewelry and ceramics, where material finish and color matter enormously. One metallic pendant lit warm; the next lit cool. One ceramic glaze photographed matte; the next reflective. The customer assumes the products are different. Manual reshoots are the obvious answer—until you realize you'd need to reshoot weekly. AI generation solves this by letting you lock one real photo as your "source of truth," then generate variations that inherit its lighting, tone, and material rendering. This only works if your AI prompt names the exact material, finish, and lighting angle of your real photo first.

The Three-Step System: Real Photo → Locked Prompt → Variations

Step 1: Choose your best real product photo—the one with the lighting and mood you want to repeat. Step 2: Reverse-engineer that photo into a detailed prompt (material name, finish type, specific light source, background tone, angle). This becomes your template. Step 3: Generate 7–12 variations by tweaking only the angle, background mood, or secondary styling—never the material or light type. This method works because AI respects specificity. Instead of "a ceramic mug," you write "handthrown ceramic mug, matte grey glaze, shot from 45° angle with diffused side-window light, warm cream linen background, shallow depth of field, no shadow." Each variation you generate changes only one variable—"shot from above," "shot from side," "minimalist white background," "styled with dried flowers"—while keeping the material and light constant. Result: 8 photos that are unmistakably from the same product line, shot with the same care.

Matching AI Output to Your Phone's Actual Lighting Setup

The mistake most people make: they generate an AI image of perfect backlighting, then try to fake backlighting with a phone flashlight. It doesn't match because AI backlighting and DIY backlighting render shadows and material differently. Instead, reverse the process: set up your phone's lighting first (we detail 5 real setups: window side-light, white bounce fill, DIY backlighting, ring light, three-point with flashlights), then write your AI prompt to match what you actually have. For example, if you're using window side-light, your prompt should say "diffused side-window light from the left, soft shadow on right, no artificial fill." If you use a white bounce card, say so. AI will render the material, reflection, and shadow exactly how that light would hit it—because you've named it. This is the bridge between AI and reality that keeps your variations looking like they belong in the same shoot.

When AI Breaks and How to Fix It Without Starting Over

Common failures: warped geometry (your ring looks melted), wrong material rendering (ceramic looks plastic), background contamination (trees appear behind your indoor product), lighting inconsistency (shadows are on the wrong side). Each has a one-line prompt fix. Warped geometry usually means your prompt is too vague about size or position—add "shot straight-on, no perspective distortion, symmetrical." Wrong material usually means you didn't name the finish—add "matte" or "polished" or "raw clay." Background contamination means your background description was unclear—remove "outdoor" language and add "clean studio background, no environment." Lighting inconsistency usually means you didn't anchor the light source—name it: "single side-window light, no other light sources, hard shadow on right." We include 4 detailed fail cases with exact edits so you can diagnose your own output in 30 seconds and fix it without regenerating.

Building Your Own Prompts for Products We Didn't Cover

The 20 included prompts cover jewelry, ceramics, glass, and textiles—but your shop likely has specific styles. Use the 5-layer prompt framework: Layer 1, name the object and material exactly ("brass signet ring, high-polished finish"); Layer 2, describe the angle and framing ("shot from above, 3/4 view, close-up on surface"); Layer 3, lock the light source ("warm window light from left, no fill, contrasty shadows"); Layer 4, set the background ("soft cream linen, no texture, completely clean"); Layer 5, add negative instructions to prevent AI from hallucinating ("no other objects, no background environment, no text, symmetrical composition"). Write these down once per product type. Then every time you generate, you only change the angle, mood, or styling detail—the core stays consistent. We show a fully worked example (a handblown glass bowl) so you can see exactly how to apply this to your first custom product.

The 2-Hour Workflow: From One Photo to Eight Listing Images

Hour 1: Set up your lighting once (30 min), shoot one hero photo with your phone (15 min), reverse-engineer that photo into a locked prompt (15 min). Hour 2: Generate 8–12 variations with different angles and moods using your template prompt (20 min), screenshot and cull to 8 keepers (15 min), upload to Etsy and add copy (25 min). This is faster than reshooting because you're not waiting for natural light, repositioning props, or adjusting cameras. You generate, pick, and upload. The checklist sequences this: hero shot first, then lifestyle variations, then real photo matching (pick the generated image that matches your real photo best—that's your confirmation the prompt is locked), then final quality check. Follow this order and you'll spend zero time on dead-end generations.

FAQ

Won't customers notice these are AI-generated?

Only if you don't anchor them to reality. Start with a real photo you actually took, then generate variations that match it. Customers see the real photo first, trust it, then the other 7 images look like professional angles of the same product. The moment you generate from nothing—no reference to a real photo—it looks fake. Always start with real.

Do I need to use Midjourney, or can I use DALL-E or Stable Diffusion?

All three work. Midjourney is sharpest for jewelry and reflective surfaces; DALL-E is best for natural styling and background flexibility; Stable Diffusion is fastest and cheapest. The prompts need translation between platforms (more descriptive for DALL-E, more technical for Midjourney), but the framework is identical. We include 3 worked translations so you can adapt.

What if my product has a lot of detail and AI keeps oversimplifying it?

Name every detail in your prompt: "filigree silver pattern on band, intricate, hand-engraved effect, not smooth, visible texture." AI responds to specificity. If you say "detailed," it ignores you. If you say "intricate filigree, hand-engraved, visible fine lines," it prioritizes detail. Also: AI struggles with very small objects. If your jewelry is tiny, zoom in by saying "macro photography, extreme close-up, 10cm from lens" and AI will render detail better than wide shots.

Can I use these images on Etsy and Shopify, or do I need to disclose they're AI-generated?

You can use them. Etsy and Shopify allow AI-generated images if they don't misrepresent the product. The key: pair AI images with one real photo (your hero shot) so customers see what they're actually buying. If all 8 photos are AI and don't match reality, that's misrepresentation. One real + seven AI variations is standard and compliant.

How long does it take to generate one batch of 8 images?

About 4–6 minutes with Midjourney (1 minute per image, some regeneration), 2–3 minutes with DALL-E (faster but sometimes lower quality), 30 seconds to 1 minute with Stable Diffusion (fastest). The bottleneck isn't generation—it's you picking which ones match your prompt. Expect 15–20 minutes from prompt to culled final set.

What if the lighting in my AI images doesn't match my phone's actual capability?

That's the point of matching your prompt to your lighting setup first. Don't describe backlighting if you only have a ring light. Don't say "three-point professional lighting" if you're shooting with a phone and one window. The 5 phone-feasible setups we include all have matching prompt language, so your AI output matches your real photo capability. This keeps everything believable.

Do I have to regenerate all 20 prompts, or can I just use a few and modify them?

Modify them. The 20 are templates. Swap in your actual product names, materials, and colors. A prompt for "oxidized silver ring" becomes "oxidized copper ring" if that's what you make. The structure and lighting language stay the same. The framework section teaches you to build from scratch for products totally outside these 20, so you're never trapped by what's included.

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 Batch Apply the Same Color Grade to 100+ Photos in Camera Raw →Learn the Sync Settings method to apply one color grade to 100+ photos at once in Adobe Camera Raw. Works for flat-lay, lifestyle, and catalog shots. Make All Your Lifestyle Photos Look Like One Cohesive Brand — Without Reshooting →Apply consistent color grading to mixed-lighting lifestyle shots in minutes, not hours. Copy-paste AI prompts + exact Camera Raw values. Why Your Lifestyle Photos Have Different Skin Tones—And How to Fix Them Fast →Correct skin-tone color shifts from mixed lighting without reshooting. Step-by-step Photoshop workflow for consistent brand portraits. Color-Grade 100 Lifestyle Photos: AI Prompt Workflows for Consistent Brand Look →Grade 100 product photos to a consistent brand look in one to two hours — without manual editing or a retoucher on retainer....