Traditional product photography costs $50-150 per image. With AI, that drops to $2-8 per image. For a catalog of 500 products, that difference is $25,000-75,000.
I'm a marketing agency owner. I've watched clients hemorrhage budget on product photography that should have been done in-house with AI tools two years ago. The workflows exist. Most brands just haven't made the transition yet.
This guide covers the exact process we use: base product shots, background removal, AI enhancement, final polish, and the three tools that do the work.
The ROI Case (In Numbers)
Before the workflow, the numbers that justify the investment in setting this up.
Traditional product photography cost breakdown:
- Professional photographer: $150-300/hour
- Studio rental (if needed): $100-300/half day
- Post-processing: $15-40/image
- Total: $50-150 per finished image
AI product photography workflow cost breakdown:
- Adobe Firefly subscription: $20.99/month (included in Creative Cloud for most businesses already paying for it)
- Canva Pro: $15/month (or free tier)
- Midjourney Standard: $30/month
- Your time (after the learning curve): 5-10 minutes per image
- Total: $2-8 per finished image including amortized tool costs
Break-even: a team generating 20+ product images per month recovers the tool costs immediately. At 100 images/month, you're saving $4,000-12,000 compared to outsourcing.
The time savings compound too. Traditional photo shoots mean scheduling, coordination, turnaround time. AI workflows run when you need them. New product launch on Tuesday? Images ready by Monday morning.
What You Actually Need to Start
A few things that make the difference between usable output and frustrating garbage:
A decent base photo. AI enhancement works with what it gets. A product shot with consistent lighting, sharp focus, and a simple background (white, gray, or any single color) gives AI something to work with. Your phone camera in good natural light is sufficient for most products. Professional-quality input gives professional-quality output.
Remove.bg or Photoshop's AI background removal. The background removal step is where most people's DIY attempts fall apart. Don't try to do this manually. Remove.bg handles most clean-background product shots in seconds, costs $0.20-0.80 per image, and the output quality is better than everything except Photoshop's Generative Remove with proper setup. If you're already on Creative Cloud, use Photoshop's AI tools -- they're the best quality and already paid for.
Adobe Firefly, Canva AI, and Midjourney. I'll explain each role below.
The Four-Step Workflow
Step 1: Capture the Base Product Shot
Photograph your product against a white or neutral background with even lighting. The exact background doesn't matter -- it gets removed. Lighting matters.
Consistent, diffused lighting from multiple angles eliminates harsh shadows that make background removal difficult. A $30 lightbox works for small products. Natural light near a window works for most consumer goods. Avoid direct sunlight or single-direction lighting.
Minimum resolution: 2000px on the short side. You're cropping and resizing later, so give yourself room.
One tip I give every client: take more photos than you think you need. 20 angles of a product is easier to work with than 3. AI enhancement can do a lot, but it can't generate a side angle if you only photographed the front.
Step 2: Background Removal
Use Remove.bg (batch API if you have volume) or Photoshop's Select Subject + Remove Background.
For 10 images or fewer: Remove.bg's web interface is fastest. Upload, download the transparent PNG, move on.
For 50+ images: set up the Remove.bg API with a batch script, or use Photoshop's batch automation for background removal. The per-image cost on Remove.bg's API is $0.20-0.80 depending on plan. At scale, the Creative Cloud option becomes more economical.
Quality check: zoom into edges before proceeding. Products with complex edges (hair, transparent packaging, loose fabric) need manual touch-up in Photoshop. Products with clean edges (bottles, electronics, packaged goods) usually come through automatically.
Step 3: AI Background and Environment Enhancement
This is where the visual transformation happens. Three tools, different use cases:
Adobe Firefly — for professional product contexts
Firefly's Generative Fill is the most practical tool for e-commerce work because the output stays commercially safe (trained on licensed content, no copyright concerns for commercial use).
Workflow: open your background-removed product PNG in Photoshop, use Generative Fill to add an environment. Prompts that work well for product photography:
- "White minimalist studio background, subtle shadow below product"
- "Modern kitchen counter, natural light, clean aesthetic"
- "Lifestyle bathroom shelf, warm morning light"
- "Outdoor wooden table, garden background, blurred depth of field"
Generate 4 variations, keep the best 1-2. For catalog images and advertising, Firefly's output is consistently usable without much post-processing. Typical time: 5-8 minutes per product.
Canva AI — for quick social and marketing variations
If you're creating social media posts, story templates, or marketing assets rather than pure product catalog images, Canva AI's background generator is faster and easier to work with than Firefly.
Upload your background-removed product, use the Background Generator to create a scene, adjust the layout in Canva's editor. The output quality is slightly lower than Firefly for technical photorealism, but for social use at 1080x1080 or story format, it's more than sufficient.
Time advantage: Canva's editing environment is faster for creating multiple format variations of the same image (square, vertical, horizontal, story). If you're making 5 different social formats from one product photo, Canva wins on workflow efficiency.
Midjourney img2img — for lifestyle and brand editorial
For premium lifestyle contexts -- the product placed in an aspirational scene, dramatic lighting, high-production-value environment -- Midjourney's image-to-image feature produces output that Firefly and Canva can't match aesthetically.
Workflow: provide your product image as a reference, set image weight (--iw 0.8 to 1.2 depending on how much you want the product to dominate vs blend with the scene), write a detailed environment prompt.
This takes more prompt engineering than Firefly and the output isn't always consistent. But for hero images and brand storytelling -- the shots that lead your campaigns -- Midjourney's aesthetic quality is worth the extra work.
Cost note: each Midjourney generation uses a GPU minute from your plan. Don't use Midjourney for catalog images where you need 200 consistent product shots on white backgrounds. Use it for the 5-10 hero images where aesthetic quality is the priority.
Step 4: Final Polish
Even with good AI output, 10-15 minutes of post-processing improves every image. What to check:
- Shadow consistency — AI backgrounds sometimes put shadows in the wrong direction. Fix in Photoshop with a manual shadow layer.
- Color accuracy — AI-generated environments can shift product color. If your red product looks slightly orange, use Selective Color to correct.
- Edge quality — Zoom in at 200% and clean up any edge artifacts from the background removal step.
- Compression for web — Export at 85% JPEG quality for product pages. WebP format if your platform supports it for faster page load.
Final quality check: does the product look like it actually belongs in the scene? The tell is lighting direction and shadow placement. If those are correct, the image reads as real.
What This Looks Like at Scale
A client of mine runs an outdoor gear brand with about 300 SKUs. Before implementing this workflow, they were spending roughly $18,000/year on product photography -- quarterly shoots with a photographer, plus post-processing.
After the transition:
- Base photos: in-house on a $200 lightbox setup
- Background removal: Remove.bg API at batch pricing, ~$60/month
- Firefly for catalog images: included in their existing Creative Cloud subscription
- Midjourney for campaign hero shots: $30/month
- Part-time VA trained on the workflow: $15/hour, ~5 hours/week
Annual cost of new workflow: approximately $4,200. Annual savings vs. previous approach: approximately $13,800.
The images aren't identical to professional photography for every use case -- hero campaign shots still occasionally go to a photographer for the highest-stakes placements. But 80% of the catalog volume now runs through the AI workflow, and the client has more assets (more angles, more lifestyle contexts, more seasonal variations) than they could afford before.
Common Mistakes and How to Avoid Them
Starting with bad base photos. AI can't fix blurry, poorly-lit, or incorrectly-exposed product shots. The 30 minutes learning basic product photography lighting (a lightbox, natural window light, or two cheap LED panels) is time well spent.
Using one tool for everything. Firefly for catalog backgrounds, Canva for quick social variations, Midjourney for hero lifestyle images -- they're tools for different jobs. Trying to force Midjourney into catalog volume work is expensive and inconsistent.
Skipping the polish step. AI background generation isn't finished product photography. The 10 minutes of shadow correction and edge cleanup separates professional output from "this looks AI-generated."
Ignoring licensing. For commercial use, make sure your base photography is yours (not from a stock site with commercial restrictions) and use commercially-safe AI tools (Firefly is the safest bet here). Midjourney commercial use requires a Pro or Mega plan. Read the terms for anything you're using in paid advertising.
Getting Started This Week
If you're managing a product catalog and not using AI photography yet, the lowest-risk entry point:
- Pick 5 of your existing product photos with clean backgrounds
- Run them through Remove.bg ($1-2 total)
- Bring the transparent PNGs into Photoshop Generative Fill or Canva AI
- Generate 3 background variations per product
- Compare the output to what you're currently using
If the output is usable for your product category, you have your answer. Most categories pass this test easily. A few (luxury jewelry, food with freshness cues, anything requiring precise color accuracy) need more careful quality control.
The tools have gotten good enough. The economics are clear. The main blocker at this point is just doing the initial setup.
For context on the AI image tools covered here, our best AI image generators roundup covers the full landscape. The DALL-E 3.5 review is relevant if you're evaluating OpenAI's tools specifically for product visualization. And if you need consistent text in your product graphics, the Ideogram 2.0 review covers that use case directly.
The bottom line: if you're spending more than $5/image on product photography at scale, you're leaving money on the table.
Top comments (0)