I am an assistant professor at Oklahoma Christian University. I’m sharing a breakthrough I’ve been working on that moves digital imaging beyond the discrete pixel grid.
ImageCFN (.cfng) is based on my Compositional Function Networks (CFN) framework. It represents images as continuous analytical manifolds rather than grids of dots.
The Benchmarks:
Efficiency: 3.7x higher compression than PNG.
Quality: 47.4 dB PSNR (visually indistinguishable from source).
Speed: 1.2s average encoding time on the Kodak dataset (3080 Ti GPU).
Resolution Independence: Decodes to any resolution in real-time. No pixels, no aliasing, just a continuous mathematical signal.
I've included an "Effects Lab" in the demo to show how HDR recovery and filters work when applied directly to the manifold before rasterization.
The technology is patent-pending, and I have a C++ SDK ready for evaluation. I'm hosting the demo on my own backend, so please be patient if we hit a traffic spike!
I am an assistant professor at Oklahoma Christian University. I’m sharing a breakthrough I’ve been working on that moves digital imaging beyond the discrete pixel grid.
ImageCFN (.cfng) is based on my Compositional Function Networks (CFN) framework. It represents images as continuous analytical manifolds rather than grids of dots.
The Benchmarks:
Efficiency: 3.7x higher compression than PNG.
Quality: 47.4 dB PSNR (visually indistinguishable from source).
Speed: 1.2s average encoding time on the Kodak dataset (3080 Ti GPU).
Resolution Independence: Decodes to any resolution in real-time. No pixels, no aliasing, just a continuous mathematical signal.
I've included an "Effects Lab" in the demo to show how HDR recovery and filters work when applied directly to the manifold before rasterization.
The technology is patent-pending, and I have a C++ SDK ready for evaluation. I'm hosting the demo on my own backend, so please be patient if we hit a traffic spike!
Demo: https://web-demo-ten-navy.vercel.app
I’m eager for feedback from the community!