Image Generation — Generasi Gambar

AI yang menghasilkan gambar dari prompt, conditioning (gambar lain, pose, dll), atau random. Era modern: DALL-E 2, Stable Diffusion, Midjourney, Imagen, Flux.

Image generation: VAE, GAN, diffusion. Era modern didominasi diffusion (Stable Diffusion, DALL-E 3, Imagen, Midjourney). 2024+: text-to-video, multimodal.

Also known as: generasi gambar
Print

Image Generation

Definisi

Image generation adalah sub-bidang AI yang menghasilkan gambar baru dari prompt, conditioning, atau noise.

Era

VAE (2013-2017)

  • Variational Autoencoder (Kingma, 2013)
  • Blurry, low-res
  • DALL-E 1 (2021) — autoregressive

GAN (2014-2022)

  • GAN (Goodfellow, 2014)
  • StyleGAN (2018-2024) — wajah realistis
  • BigGAN (2018)
  • Masih digunakan untuk beberapa aplikasi

Diffusion (2020-sekarang)

  • DDPM (2020)
  • Latent Diffusion (2022) — Stable Diffusion
  • DALL-E 2 (2022)
  • Imagen (2022)
  • Midjourney v5 (2023)
  • SDXL, SD3 (2023-2024)
  • Flux (2024)
  • Sora (2024) — text-to-video

Aplikasi

  • Seni & desain
  • Advertising
  • Game assets
  • Film previsualization
  • Personalisasi
  • Pendidikan

Connected to

Not yet written

The following pages are referenced but don't exist yet — they'd make good future additions.

  • /concepts/diffusion-model

References

  1. Wikipedia

Type at least 2 characters to search.

Press to navigate, to open, esc to close.