Changelog

Image Editing

Powerful new editing capabilities:

• Inpainting: Edit specific regions of generated images
• Outpainting: Extend images beyond their original boundaries
• Style Transfer: Apply artistic styles to existing images
• Background Removal: One-click background removal

AI Upscaling

Enhance your images with AI:

• 4x Upscaling: Increase resolution without quality loss
• Detail Enhancement: Add fine details to upscaled images
• Noise Reduction: Clean up low-quality source images

Performance

• 50% Faster: Optimized inference pipeline
• Global CDN: Faster image delivery worldwide
• Real-time Preview: See generation progress in real-time

Quality Improvements

• Improved prompt understanding for all models
• Better handling of complex compositions
• Enhanced text rendering accuracy
• More consistent style application

Bug Fixes

• Fixed occasional timeout errors
• Resolved aspect ratio calculation issues
• Improved mobile touch interactions

New Models

Expanded our model lineup with two powerful additions:

• Nano Banana Pro: Google's fastest photorealistic model, 3-5x faster generation
• Recraft: Vector-first model for clean SVG assets and scalable graphics

Developer API

Introducing our public API for developers:

• Unified Endpoint: Single API for all models
• Code Examples: JavaScript, Python, and cURL examples
• Rate Limiting: Flexible limits based on your plan
• Webhook Support: Get notified when images are ready

Improvements

• Batch Generation: Generate multiple variations simultaneously
• History View: Browse and re-download your generated images
• Prompt Templates: Save and reuse your favorite prompts
• Improved Queue: Faster processing during peak times

Bug Fixes

• Fixed image download issues on Safari
• Resolved credit deduction timing
• Improved error messages for failed generations

AI Image Generation

We're excited to announce the initial release of AvatarMe with the following AI models:

• GPT Image 1.5: OpenAI's latest model with exceptional text rendering
• FLUX.2: Black Forest Labs' photorealistic image generation
• Midjourney v7: Industry-leading artistic style and composition
• HiDream-I1: 17B parameter open-source model

Core Features

• Text-to-Image: Generate images from natural language descriptions
• Multiple Aspect Ratios: Support for square, portrait, and landscape formats
• High Resolution: Generate up to 2048x2048 pixel images
• Download & Share: Easy download and social sharing options

Technical Stack

• Built with Next.js 14 and React Server Components
• Drizzle ORM for type-safe database operations
• Stripe and Telegram Stars payment integration
• Responsive design for all devices