A full redesign concept for crop.photo — from marketing homepage to application dashboard. Turning an AI-powered bulk image editing platform into an experience that shows its intelligence without making users read about it.
A powerful bulk image tool is worthless if the person who needs it cannot tell, in ten seconds, what it does for them.
The approach
Redesign both halves — a marketing site that states the outcome plainly, and a dashboard that makes the batch workflow obvious.
The outcome
A full concept redesign taking crop.photo from homepage to application, with the AI capability made legible rather than implied.
2
Surfaces redesigned
1
Batch workflow
1
Marketing site
1
Dashboard
The Problem
AI power hidden behind complexity
Crop.photo processes thousands of images in minutes — reducing a 4-day editing job to
under 1 hour (as documented with Wonder Wheel). Yet the homepage buried this under feature lists, and the
application dashboard overwhelmed new users before they could experience the AI's impact.
The Opportunity
Brands like Lacoste are achieving 10x ROI from
crop.photo's bulk AI editing. The tool already delivers — the design needed to match that confidence.
Redesign the homepage to demonstrate value in 3 seconds, and the application to make bulk processing feel
intuitive rather than industrial.
8 AI features including Smart Resizer, Headless Face Cropper, Listing Analyzer, and
AI Fashion Model Generator — each needed a clearer, more discoverable interface.
Wonder Wheel cut image processing from 4 days to under 1 hour using crop.photo.
Yet the homepage never communicated this speed — visitors read feature labels instead of experiencing
the AI's actual capability, costing the platform conversions before the scroll began.
02 / 03
Feature Discovery Failure
Crop.photo has 8 distinct AI features — Smart Resizer, Headless Face Cropper,
Listing Analyzer, AI Fashion Model Generator, Background Remover, and more. None were surfaced intuitively
in the application. Users who found Batch Processing stayed; those who didn't, left.
03 / 03
Marketplace Friction
Sellers on Amazon, Shopify, Walmart, and eBay each need platform-specific image
dimensions. The existing export flow required manual dimension entry per platform — eliminating the very
time saving the AI had just delivered. A single friction point was reversing the product's core value.
05 — User Stories
What Users Need
As a...
I want to...
So that...
Priority
E-commerce Seller
Bulk crop and remove backgrounds in one workflow
I can process hundreds of product images in minutes
High
Marketing Designer
Export one image in all platform-specific sizes at once
I eliminate the repetitive resizing step from my workflow
High
Social Media Manager
Edit and export images from my phone
I can create content anywhere without needing a desktop
High
Photographer
Remove complex backgrounds from portraits accurately
I can deliver clean cutouts to clients without manual masking
Medium
Agency Creative
Access API for batch processing at scale
I can integrate AI cropping directly into my production pipeline
Medium
06 — Competitor Analysis
Market Landscape
Feature
Remove.bg
Canva
Adobe Express
Clipping Magic
Crop.photo
Batch Processing
~
✕
~
✕
✓
AI Background Removal
✓
✓
✓
✓
✓
Aspect Ratio Templates
✕
✓
✓
✕
✓
API Access
✓
✕
~
✕
✓
Mobile Support
✓
✓
✓
✕
✓
Quality Preservation
✓
~
~
✓
✓
Solution Exploration
Three decisions that made the AI visible.
Crop.photo's AI delivers extraordinary results. The redesign problem was that nothing in the interface communicated those results — or helped users trust them. Three decisions addressed that gap.
Decision 01
Feature-list homepage vs. Live before/after demo as homepage hero
Option A
Feature-list homepage: copy describing "AI-powered cropping," "background removal," "bulk processing" — accurate but abstract, forces visitors to imagine the result
Option B — Chosen
Before/after slider as the hero: a single drag demonstrates what 2000 words of feature copy can't — the AI's actual output in under 2 seconds
AI output is the product's most powerful asset. Describing it wastes the opportunity to show it. A before/after slider communicates "bulk AI image cropping" with zero reading required — the demo is the value proposition. Visitors who see the result convert; visitors who read the description consider it.
Decision 02
Manual marketplace dimension entry vs. Marketplace-aware preset selector
Option A
Manual dimension input: user enters width × height for each marketplace, each time — technically complete, but the most frequent task was also the most tedious
Option B — Chosen
One-click preset selector: Amazon, Shopify, Walmart, eBay, Macy's presets — choose marketplace, all dimensions resolve. Multi-platform parallel export in one download.
The AI saves 4 days of work, then manual dimension entry gives some of it back. The most common workflow was Amazon → Shopify → eBay every time. Presets don't add features; they remove the one friction point that made users feel like the AI hadn't actually saved them any effort.
Decision 03
Trust-all AI output vs. AI confidence score with flagged review
Option A
Trust-all: export all processed images directly — fastest path, but users processing 500 product shots are anxious about what they can't see
Option B — Chosen
Per-image AI confidence score: high-confidence → direct export; flagged → surface for manual review. Selective trust rather than blind or paranoid
Trust is calibrated, not binary. Users managing 500 product images for Amazon can't manually review every one — but they also can't risk a bad crop on a primary product image. Confidence scoring makes AI output auditable without making it laborious. It resolves the anxiety of trusting AI at scale.
Final Design
Let the AI do the talking
A full redesign spanning marketing homepage and application dashboard — homepage
leads with a live before/after demo, the app surfaces all 8 AI features through a unified, marketplace-aware
workflow that turns bulk processing into a delightful experience.
crop.photo
Crop.photo — AI Bulk Image Editing App Redesign
Design Decisions
Live Before / After Demo as Homepage Hero
Problem
The homepage described the AI's capability in text. Visitors had to imagine the result — and imagination is always less convincing than evidence.
Approach
A before/after slider as the centrepiece hero. A single drag communicates "bulk AI image cropping" in under 2 seconds without any reading required.
User Benefit
Visitors see the AI's actual output immediately. The "does it work?" question is answered before any feature copy is processed.
Business Benefit
Demo-as-hero reduces the distance between awareness and conversion. Visitors who see the result sign up; visitors who read about it consider it.
Unified Batch Dashboard with Real-Time AI Status
Problem
Active jobs, processing progress, and completion queue were scattered across tabs. Users processing 500-image batches couldn't tell where anything was without navigating.
Approach
Single scannable dashboard: all active jobs, real-time AI processing progress, and completion queue visible simultaneously. One screen tells the complete picture.
User Benefit
Batch status understood at a glance. No hunting across tabs to know where a 500-image job stands — operators can work on other tasks while monitoring progress peripherally.
Business Benefit
Reduced support tickets about job status. Users who can see progress are more patient with processing time — perceived performance is as important as actual performance.
Marketplace-Aware Preset Selector
Problem
Manually entering dimensions for each marketplace (Amazon: 1000×1000, Shopify: 2048×2048, eBay: 1600×1600) erased the time savings the AI created. The most common task was the most tedious.
Approach
One-click presets for Amazon, Shopify, Walmart, eBay, and Macy's. Choose marketplace, all dimensions resolve instantly. Multi-platform parallel export in a single download.
User Benefit
The AI's time savings are preserved end-to-end. What was previously the biggest friction point becomes the fastest step in the workflow.
Business Benefit
Higher multi-marketplace export rates mean users are getting more value per session. More marketplace outputs = more stickiness and higher retention.
Batch Preview Grid with AI Confidence Score
Problem
Users couldn't trust AI output blindly at scale — a bad crop on a primary Amazon product image has real revenue impact. But reviewing 500 images manually defeats the purpose.
Approach
Per-image AI confidence score in the review grid. High-confidence images export directly. Flagged images surface for selective manual review — calibrated trust, not blind or paranoid.
User Benefit
Audit without reviewing everything. Users focus attention on the images that need it, confident the rest have been correctly processed.
Business Benefit
Confidence scoring reduces the barrier to adopting AI for high-stakes product photography. Trust at scale is what converts trial users to production users.
Headless Model & AI Feature Surfacing
Problem
Crop.photo's most powerful features — Headless Face Cropper, Listing Analyzer, AI Fashion Model Generator — were buried in settings menus and not discoverable to new users.
Approach
Dedicated, contextually surfaced entry points for each feature — shown at the moment in the workflow when they're most relevant, with clear "what this does" copy above the CTA.
User Benefit
Power users discover capabilities that save them additional hours. New users understand the product's full scope without needing a demo or documentation.
Business Benefit
Feature adoption drives retention. Users who discover and use the Headless Cropper or AI Fashion Model Generator are significantly more likely to become long-term paying customers.
Design System
Built for clarity at speed
A blue-and-amber palette matched to crop.photo's actual
brand — clean SaaS typography, generous whitespace, and a component library built around batch processing
clarity, AI status communication, and marketplace-aware workflows.
Colour Tokens
--accent#0066FF
--accent-sec#f59e0b
--bg#060812
--accent-dimrgba(0,102,255,0.10)
Type Scale
72pxHero
Title
48pxSection
Italic
20pxBody Light
11pxLABEL
CAPS
Spacing Scale
Component Library
Start Free →
Learn More
Ghost CTA
E-commerceBulk
APIBG
Remove
INSIGHT CARD
Show,
Don't Tell
Visual demos
outperform feature text for AI tools — every time.
BEFORE
AFTER AI
Impact & Results
From hidden potential to visible power
The redesign addresses both conversion and retention —
homepage shows the AI's impact in seconds; the application makes bulk processing feel intelligent and
in-control, matching the 10x ROI that brands like Lacoste are already achieving.
0ROI — Lacoste with Crop.photo
Documented customer result: Lacoste achieved 10x ROI on image processing
workflows using crop.photo's AI bulk editing.
4 days → 1 hrProcessing Time — Wonder Wheel
Wonder Wheel reduced a 4-day image editing workflow to under 1 hour — the kind
of impact the redesign was built to make immediately visible.
0Behance Appreciations
Strong reception from the design and AI SaaS community — validating the
visual-first approach to communicating AI tool value.
0AI Features Made Discoverable
All 8 crop.photo AI features — from Smart Resizer to AI Fashion Model Generator
— given clear entry points in the redesigned application.
Key Learnings
What this project taught me
01
Show the work; remove the explanation
When the AI's output is demonstrably powerful, describing it is the wrong strategy. A before/after slider communicates more in 2 seconds than any amount of feature copy. For AI products, the interface's job is to get out of the way and let the output speak — the demo is the value proposition.
02
The last mile matters as much as the AI
The AI saved 4 days of work; manual dimension entry gave some of it back. The experience of an AI tool isn't just the AI — it's everything the user has to do before and after. Marketplace presets didn't improve the AI; they made the AI's time savings real by eliminating the manual work that surrounded it.
03
AI trust is calibrated, not binary
Users don't need to review everything; they need to be confident that what they don't review has been correctly processed. AI confidence scoring makes trust auditable — which is what allows trust to scale. Products that ask users to trust everything ask them to trust nothing.
04
Feature discoverability is a product design problem, not a documentation problem
The Headless Cropper and AI Fashion Model Generator were powerful features that most users never found. Documentation doesn't solve discovery — contextual UI placement does. Features that are surfaced at the right moment in the workflow get adopted; features that require users to look for them don't.
"Crop.photo already delivers results that sound impossible — 4 days to 1 hour, 10x ROI. The designer's job
was to make the interface match the AI's ambition. Every decision was about surfacing that power rather than
describing it. Show the work; remove the explanation."
