Jasper Zheng (Shuoyang)
CV


All Research Artwork Music

Real-time interventions of DNNs and expressive feature exploration

In pursuing human-AI co-creation, we propose an alternative approach to reuse encoded knowledge in trained AI models and intervene in the generative process, allowing it to re-organise and re-write features that diverge from the original dataset.

Category Artwork Subject Creative AI Date Dec, 2022

Enabling pre-trained StyleGAN models to image-to-image translation

Rewriting Activation Maps proposes a training system appended to StyleGAN2 architecture, enabling a pre-trained model to perform image-to-image translation, even if the input images are not in the original domain.

Category Research Subject Creative AI Date Jun, 2022

Manipulated Network: quirks and oddities from corrupted generative models

I’ve come across plenty of quirks and oddities generated from corrupted models. These models produce impossible, distorted but realistic images, diverted from the original outputs and sometimes lead to aesthetic preoccupations.

Category Artwork Subject Creative AI Date Mar, 2022

↑↑↑% - aestheticise chaotic dynamics in the physical world

↑↑↑% (up percent) is a sound installation that reveals possible interactions of natural systems, sound, light and humans. We are invited to review the metrics of the 'organic collaborators' we are searching for.

Category Artwork Subject Sound Installation Date Dec, 2021

1 2 3 4