Hans | Wavefunk

Hans Brouwer / Wavefunk

Neural networks & neuro basses.

Notes and experiments at the intersection of machine learning, generative art, and sound.

Writing & explorations

RSS ↗

  1. Optex: Replicating "Optimal Textures"

    Fast, robust texture synthesis & style transfer with optimal transport.

  2. Audio-reactive Latent Interpolations with StyleGAN

    Teaching GANs to dance! ♪┏(・o･)┛♪

  3. Lakspe: Building a Dataset

    Texture synthesis, polar transforms, and seam carving.

  4. Zeinu: Next Frame Prediction I

    Generating videos with next frame prediction and feedback loops.

  5. Rhodops

    My first GAN ᡣ𐭩 : progressively growing generative adversarial networks

Research & papers

  1. Master's thesis

    Self-supervised Audio-reactive Music Video Synthesis

    Measuring and optimizing audiovisual correlation.

  2. NeurIPS 2020 workshop paper

    Audio-reactive Latent Interpolations with StyleGAN

    Mapping musical features to the latent and noise inputs of a generative model.

  3. Research survey

    Survey on Multi-modal Generative Models

    Encoding, information exchange, and decoding across data modalities.

Shorter notes

  1. Portfolio

    Audiovisuals for KOAN Sound's Led by Ancient Light Tour

    Audio-reactive visuals using StyleGAN and AnimateDiff for KOAN Sound's Led by Ancient Light Tour, a live performance where they play a variety of instruments and orchestrate effects over the base visuals in TouchDesigner.
  2. Portfolio

    Realtime Audio-/MIDI-Reactive Visuals for Corpus Viderum

    A high-performance, audio-/MIDI-reactive inference pipeline for Ryan Walsh's Corpus Viderum. Using Apache TVM to compile a StyleGAN model and push realtime, full HD resolution visuals to three free-hanging LCD panels simultaneously on a single RTX 3090 GPU.
  3. Portfolio

    Audiovisuals for Frequent @ Tipper Red Rocks 2021

    An hour long, multi-GAN audio-reactive opus for Frequent's set in the legendary Red Rocks Amphitheatre near Denver, Colorado.
  4. Video

    It's Raining: Next Frame Prediction III

    Experimenting with CycleGANs, next frame prediction feedback loops, and cyclical temporal consistency.
  5. Video

    Tuoruen

    Writing my own progressively grown GAN in PyTorch.
  6. Image

    Mode Collapse

    GAN training instability, a curse and a blessing.
  7. Video

    Neural Lavalamp: Next Frame Prediction II

    Experimenting with CycleGANs, next frame prediction feedback loops, and cyclical temporal consistency.
  8. Image

    Rhodox

    A GAN trained on a dataset of images generated with recombinant style transfer.

Look around. Have a listen.