Image Quality
Optimize product images for best virtual try-on results. High-quality images lead to better AI processing, higher download rates, and increased conversions.
Why Image Quality Matters
Good Images = Accurate Try-Ons = Happy Customers = Sales
Poor Images = Bad Results = No Downloads = Wasted Credits + Lost Sales
The AI needs clear, well-lit model photos to generate convincing try-on results. Poor quality inputs produce poor quality outputs.
Requirements for Product Images
Must-Have: Model Photo with Visible Face
Critical: At least ONE image in your data-images array must show:
✅ Model wearing the product
✅ Model's face clearly visible
✅ Front-facing view
Why: The AI swaps the customer's face onto the model photo. No visible face = no try-on possible.
Example:
Image Quality Checklist
✅ Resolution
Minimum: 800×800px Recommended: 1200×1200px or higher Optimal: 1500×1500px
Why: Higher resolution = better AI training data = more accurate results
Check: View image properties to verify dimensions
Too Small (<800px):
Blurry results
Poor face detection
Pixelated output
Customer dissatisfaction
Too Large (>4000px):
No benefit (AI downscales anyway)
Slower upload times
Higher bandwidth costs
✅ Lighting
Good Lighting:
Even, soft lighting (no harsh shadows on face)
Natural daylight or studio lighting
Consistent across all products
Face well-illuminated
Poor Lighting:
❌ Dark/shadowy face
❌ Extreme backlighting (silhouette)
❌ Harsh flash creating hard shadows
❌ Inconsistent lighting across images
Example:
✅ Good: Soft, even studio lighting ❌ Bad: Face in shadow, product backlit
✅ Face Visibility
Requirements:
Face must be clearly visible
Front-facing or slight angle (max 45° turn)
Eyes, nose, mouth visible
No hair covering entire face
No hands covering face
No accessories obscuring face (large sunglasses, masks)
Optimal Face Position:
Looking at camera or slight off-center
Neutral or slight smile expression
Natural head position (not tilted excessively)
Not Acceptable:
❌ Profile view (90° turn)
❌ Face looking down/away
❌ Hair completely covering face
❌ Hat/accessories hiding face
❌ Extreme close-up (only face, no product visible)
❌ Group photos
✅ Model Position & Framing
Ideal Framing:
Full upper body visible (head to waist minimum)
For dresses: Full body (head to knees/feet)
Product prominently displayed
Centered in frame
Model Pose:
Natural, standing position
Facing camera (front view primary)
Arms at sides or slightly away from body
Product clearly visible (not obstructed)
Avoid:
❌ Extreme angles (bird's eye, worm's eye)
❌ Sitting/lying poses (inconsistent with most customer photos)
❌ Action shots (running, jumping) unless relevant
❌ Cropped faces (incomplete face in frame)
✅ Background
Recommended:
Plain white or neutral background (white/gray/beige)
Minimal distractions
Consistent across product line
Professional look
Acceptable:
Subtle lifestyle backgrounds (clean, not cluttered)
Outdoor shots with good lighting
Avoid:
❌ Busy, cluttered backgrounds
❌ Competing visual elements
❌ Low contrast (model blends into background)
Why: Clean backgrounds help AI isolate the model/product for better processing.
✅ File Format & Compression
Formats:
✅ JPEG/JPG (recommended, smaller file size)
✅ PNG (supports transparency, larger files)
✅ WEBP (supports modern file type)
❌ GIF (low quality, animated)
Compression:
Balanced: Good quality, reasonable file size
Max file size: 5MB per image
Avoid over-compression (artifacts, pixelation)
Optimization Tools:
TinyPNG (https://tinypng.com)
ImageOptim
Squoosh (https://squoosh.app)
✅ Product Visibility
The Product Must Be:
Clearly visible and in focus
Worn naturally on model
Not obscured by hands, hair, accessories
Showing front view primarily
For Tops/Shirts:
Show full garment (not partially off-frame)
Visible neckline, sleeves, hem
For Dresses:
Full dress visible (head to at least knees)
Fit and flow apparent
For Outerwear:
Worn closed or open (as customer would wear)
Visible details (buttons, zippers, pockets)
Testing Your Images
Before Going Live
Quality Indicators:
✅ Face blends naturally
✅ Product fits proportionally
✅ Lighting matches customer photo
✅ No artifacts or glitches
✅ Result looks convincing
Red Flags:
❌ Distorted face
❌ Product looks "pasted on"
❌ Misaligned eyes/features
❌ Weird artifacts around edges
❌ Unconvincing result
If results are poor, replace the product image before launching!
Common Issues & Solutions
Issue: Low Download Rate (<15%)
Likely Cause: Poor try-on quality due to image issues
Diagnose
Check product images: Are faces clearly visible?
Test yourself: Do results look good?
Compare to high-performers: What do their images have that yours don't?
Solutions:
Re-shoot products with better model photos
Ensure models look at camera
Improve lighting (eliminate shadows on face)
Use higher resolution images
Choose different model photos from your library
Issue: Face Detection Fails
Symptoms: "No face detected" error or try-on doesn't start
Causes:
Face turned away from camera
Face obscured (hair, hands, accessories)
Image too low resolution
Extreme lighting (silhouette, overexposed)
Solutions:
Use front-facing model photos
Ensure face is fully visible
Increase image resolution
Fix lighting issues
Issue: Unnatural Results
Symptoms: Final image looks "off", uncanny valley effect
Causes:
Model and customer face angles don't match
Lighting mismatch
Extreme model pose
Poor image quality leading to AI confusion
Solutions:
Use natural, straightforward model poses
Neutral facial expressions on models
Even lighting
High-resolution source images
Best Practices by Product Type
Tops (T-Shirts, Blouses, Shirts)
Framing: Chest up to head Model: Front-facing, natural pose Key: Show neckline clearly
Optimal Shots
Front view (primary for try-on)
Side/angle view (supplementary)
Detail shots (fabric, pattern close-ups)
Dresses
Framing: Full body (head to knees or feet) Model: Standing, front-facing Key: Show dress fit and flow
Optimal Shots
Full-length front view (primary)
Detail of bodice/face (clear face view)
Back/side views (supplementary)
Outerwear (Jackets, Coats)
Framing: Head to waist (minimum) Model: Standing, front-facing Key: Show how garment sits on shoulders
Optimal Shots
Front view, jacket open or closed
Side view (show fit)
Detail shots (zippers, buttons)
Accessories (Glasses, Hats)
Framing: Head and shoulders Model: Close-up but face still clear Key: Product worn naturally, face visible
Note: Some accessories may not work well with AI try-on (very small items, transparent items). Test thoroughly.
Professional Photography Tips
Hiring a Photographer
Brief them on requirements:
Need clear, front-facing model faces
Even, soft lighting
High resolution (2000px+)
Plain white/neutral background
Natural poses
DIY Photography
Equipment:
DSLR camera or high-quality smartphone (iPhone 12+, Samsung S20+)
Tripod (for consistency)
Softbox lighting or natural window light
White backdrop (seamless paper or wall)
Setup:
Position model 6-10 feet from backdrop
Camera at model's eye level
Soft lighting from 45° angles (two lights)
Test shots to check face visibility and lighting
Post-Processing:
Adjust exposure/brightness if needed
Crop/frame consistently across products
Sharpen slightly if soft
Don't over-edit (keep natural)
Model Selection
Diverse Model Range
Benefits:
Represent your customer base
Customers see themselves in your models
Inclusive brand image
Considerations:
Various skin tones, ages, body types
Different hair styles/colors
Gender representation
Cultural diversity
Consistent Models
Benefits:
Brand consistency
Easier for AI (same face structure)
Professional catalog look
Strategy: Use 2-4 primary models, rotate based on product type/season.
Image Management
Naming Convention
Use consistent, descriptive file names:
Benefits:
Easy to identify
Organized library
Quick retrieval
Storage & CDN
Recommendations:
Use CDN for fast loading (Cloudflare, AWS CloudFront)
Serve images in optimized formats (WebP with JPEG fallback)
Enable lazy loading
Cache aggressively
TrialClouds-Specific:
Ensure images are publicly accessible (no auth required)
HTTPS URLs required
CORS headers properly set (if hosting on separate domain)
Quality Assurance Checklist
Before launching products with try-on:
✅ Every product has at least ONE image with clear model face
✅ Images are minimum 800×800px (preferably 1200×1200px+)
✅ Lighting is even and faces are well-lit
✅ Models are front-facing or near-front-facing
✅ Backgrounds are clean and uncluttered
✅ File formats are JPEG or PNG
✅ Files are optimized (not too large, not over-compressed)
✅ Self-tested: Try-on results look good to your team
✅ Images are hosted on fast, reliable servers
✅ URLs are HTTPS and publicly accessible
Ongoing Optimization
Monitor Analytics
Track by product:
Try-on completion rate
Download rate (KEY metric for quality)
Customer feedback
Low download rate? → Image quality issue likely
Update Images
Replace underperforming product images
Seasonal updates (new model photos quarterly/annually)
Test new photography styles
A/B test different model photos for same product
Related Guides:
Need help evaluating your images? Email examples to [email protected]
Last updated