What is GAN Evolution?
TL;DR
The technological evolution of GANs from image generation to video and 3D applications.
GAN Evolution: Definition & Explanation
GANs (Generative Adversarial Networks), introduced by Ian Goodfellow in 2014, are generative models where a Generator and a Discriminator compete against each other to produce high-quality data. Early GANs were limited to low-resolution image generation, but the technology evolved through DCGAN, ProGAN, StyleGAN, and StyleGAN2/3 to achieve photorealistic human image generation. Applications subsequently expanded to conditional generation (Conditional GAN), super-resolution (ESRGAN), image translation (Pix2Pix, CycleGAN), video generation, and 3D object generation. While diffusion models have recently surpassed GANs in image quality — as seen in Stable Diffusion and DALL-E 3 — GANs maintain an advantage in real-time generation speed and continue to be actively used in gaming and film production.