Optimizing Product Images for Virtual Try-On

Introduction

The quality of your virtual try-on results depends heavily on the quality of your input images. By following these best practices, you can ensure that huhu.ai's AI models produce the most photorealistic and accurate results possible.

This guide covers everything from image resolution and lighting to garment positioning and file format selection. Whether you are photographing products in-house or working with a professional studio, these guidelines will help you achieve consistent, high-quality results.

Image Resolution Guidelines

We recommend a minimum resolution of 1024x1024 pixels for all product images. For the best results, use 2048x2048 or higher. Higher resolution images allow our AI to capture fine details like fabric texture, stitching, and patterns.

Keep in mind that extremely large images (over 8192x8192) will be automatically downscaled during processing. The sweet spot is between 2048x2048 and 4096x4096 pixels.

Lighting Requirements

Even, diffused lighting produces the best results. Avoid harsh shadows or strong directional lighting, as these can confuse the AI model about the garment's true shape and color. Studio lighting with softboxes or a well-lit natural environment works best.

Color accuracy is critical for e-commerce. Use a color calibration card during your photoshoot and ensure your white balance is correctly set. This helps our AI maintain accurate color reproduction in the generated try-on images.

Background Preparation

While our AI can work with any background, a clean white or light gray background produces the most consistent results. If you are using a mannequin or flat-lay setup, ensure the background is free of wrinkles, shadows, and color casts.

For garments on hangers, use a consistent hanger style and position across your catalog. This helps maintain visual consistency across your product listings.

Garment Positioning

Front-facing product shots produce the best virtual try-on results. Ensure the garment is fully visible with no parts cropped out of frame. Lay flat garments should be smoothed out with no significant wrinkles or folds that could obscure the design.

For mannequin shots, use a form that closely matches standard body proportions. Ghost mannequin photography (where the mannequin is removed in post-production) works particularly well with our AI.

Recommended File Formats

For upload, we recommend PNG for images with transparency (like ghost mannequin shots) and JPEG at quality 90+ for standard product photos. WebP is also supported and offers excellent compression without quality loss.

Avoid using heavily compressed JPEGs (quality below 70), as compression artifacts can degrade AI output quality. If you must compress images for storage, keep the originals and compress copies separately.

Common Mistakes to Avoid

• Over-editing source images: Heavy retouching or filters on input images can confuse the AI. Provide minimally edited photos for the best results.
• Inconsistent lighting across catalog: Mixing images with different lighting setups leads to inconsistent try-on results. Standardize your photography setup.
• Low-resolution uploads: Upscaling a low-resolution image does not add detail. Always start with the highest resolution original available.
• Cropped garments: Ensure the complete garment is visible in the frame, including sleeves, hem, and collar.

Pre-Upload Checklist

• Image resolution is at least 1024x1024 pixels
• Lighting is even and diffused with no harsh shadows
• Colors are accurately represented (white balance checked)
• Background is clean and consistent
• Complete garment is visible with no cropping
• File is in PNG, JPEG (90+), or WebP format
• No heavy post-processing or artistic filters applied