Philip Hofmann

models. All my self trained & released AI upscaling models. After gathering and applying over 600 different upscaling models, I learned how to train my own models, and these are the results.

upscale. My VitePress website I made to visually compare the output I created with over 600 different AI upscaling models. Running under https://phhofm.github.io/upscale/

aethernet. AetherNet: High-Performance Single Image Super-Resolution (SISR) architecture in PyTorch, optimized for quality, speed, and deployment.

LegalHistoryTimelinePrototype. This is a web application prototype programmed by Philip Hofmann with THREE.js, during his student job as programmer at the Faculty of Law at the University of Zurich (UZH). The development was requested and happened in collaboration with Antonia Hartmann from professorship Thier at UZH.

ParagonSR2. ParagonSR2: Hybrid - Efficient Super-Resolution with Dual-Path Architecture

aethernet-train-new. Python

ParagonSR. ParagonSR is a reparameterizable CNN-hybrid for restoration, integrating Transformer-style components. It uses an Inception-style depthwise conv for multi-scale context and a powerful Gated-FFN for feature transformation, while fusing into a simple CNN for fast inference.

open-model-database. An open and free database for AI models

spandrel-aethernet. Spandrel gives your project support for various PyTorch architectures meant for AI Super-Resolution, restoration, and inpainting. Based on the model support implemented in chaiNNer.

aethernet-train. Training code for the AetherNet SISR Network on https://github.com/Phhofm/aethernet/

PersonalAnalytics-FocusSession. My masterthesis code I programmed. The goal of FocusSession is to support knowledge workers focus on a specific task by reducing distractions and context switches. https://youtu.be/RsMGO8-sHPs

studybuddy-web. Astro

ius-teaching-privacy. Privacy policy for the IUS Teaching App (Rechtswissenschaftliches Institut, UZH)