About HDRPop
HDRPop restores the vibrant HDR look to photos that have been flattened by AI editing tools.
The Story
HDRPop was born from a simple frustration: every time I edited photos with AI tools like ChatGPT or Midjourney, they came back looking flat. The HDR magic my iPhone captured was gone.
The problem? AI editors process images in standard dynamic range and strip out the "gain map" that tells HDR displays how to make photos pop. It's a technical limitation, not intentional—but the result is the same: dull photos.
So I built a solution using state-of-the-art AI research (GMNet, published at ICLR 2025) to restore what was lost. Now I'm sharing it with everyone who wants their photos to look as good as they remember.
The Technology
HDRPop uses the GMNet deep learning model to intelligently predict HDR gain maps from standard images. The model was trained on thousands of HDR photos to understand how light, color, and tone should map to HDR enhancement data.
The output is an Ultra HDR JPEG (ISO 21496-1 standard) that works seamlessly with iOS Photos, macOS Preview, Android 14+, and other compatible apps.
Privacy-First Philosophy
Your photos are processed and automatically deleted within 24 hours. We don't store, analyze, or train on your images. Ever. See our privacy policy for full details.
Contact
Questions, feedback, or feature requests? Join our Discord
Who Made This
HDRPop is built by David Rose, a software engineer interested in ML/AI applications and building useful tools.