Veils — Dho Yee Chung

The Veils of Asian Women #GenerativeAdversarialNetwork #AIGeneratedImage #DeepLearningModel #3DPointCloudProcessing #3DPrint #AsianStereotypes

AI-generated images of Asian women are often shaped by deep, longstanding biases. For decades, popular representations of Asian women have been packaged and sold to consumers, reinforcing a narrow, market-driven aesthetic. The stereotypical faces produced by AI echo and amplify this commodification, reducing the Asian female body to a visual symbol of desire, sexuality, obedience, and the tensions between tradition and modernity. These images ultimately reflect the demands of the marketplace more than the diversity of real individuals.

For this project, I collected over 5,000 commercial images of Asian female celebrities and influencers—images saturated with highly preferred, stereotypical features. These were used to train a custom generative adversarial network (GAN), which produced new, fictional “Asian celebrity” faces that embody these biases. The generated images were then transformed into 3D forms using point cloud processing and 2D image–to–3D analysis techniques. The final outcomes include a series of GAN-produced images and a set of 3D-printed masks.

[STEP1] GAN Implementation: Creating Stereotype Asian Celebrity Faces that Don't Exist

[STEP2] Convert 2D Images to 3D Models with Python: Using 2D Image Data Analysis Model

Image Test from Epoch 356

Image Test from Epoch 439

Image Test from Epoch 781

Image Test from Epoch 1084

Critical Inquiry

→ The Veils of Asian Women
→ Black Rain White Light
→ Mirror Mirror
→ The Digitalization of the Creative Market

Biometrics / Emotions / Sensory

→ Pulse to Light
→ Graphic Language for Emotion-Based Interface
→ Sensitive Surface

Enhanced User Experience

→ Gesture as Inclusive Interface
→ Natural Sound Mixer