Z-Coders
ATLAS is a single-session Al scrapbook tool that transforms a small batch of ordered photos into a continuous, ghostwritten story. You quickly validate Al-generated captions for each photo, and ATLAS uses that verified context to craft a cohesive, hallucination-free narrative. The result is an interactive digital flipbook where every page pairs your memories with rich, personalized storytelling.
The persistent, urgent need for unique, high-quality, and lightweight web animations (like the "floating photos" effect) that can be generated instantly without needing complex design tools like After Effects. We saw a gap between simple GIF-makers and professional motion graphics tools.
Atlas is an AI-powered service that takes a static image and a simple text/audio prompt (e.g., "photos slowly drifting in dark space") and outputs a lightweight, performant Lottie or CSS Keyframe Animation in under 15 mins, ready for website use.
We used a Nextjs. The core logic uses an AI model fine-tuned on common web animation requests (floating, parallax, pulsing) to translate text prompts into a JSON-based animation definition (Lottie/keyframe code) rather than generating a bulky video file.
..
Performance & File Size: We initially struggled with generating massive GIFs. We smashed this by switching our output from video/GIF to the lightweight Lottie JSON format, achieving a 90% file size reduction for the same visual quality.
Successfully implemented a one-click deployment feature that gives users the exact embed code for their chosen animation. Generated 15+ complex, production-ready animations within the 24-hour window.
The biggest lesson was the critical difference between generating context from image. Focusing on the latter solved the performance bottleneck and made the solution truly useful for web developers.
Users can share their memories publicly and it would be recommended to feed of those people who might have attended the same event.
