Case study: Generative AI for image creation—a diffuse vision (May and November 2027)

Please click here for a PDF of the case study.

These are specific terms and terminology you must be aware of: 

1. Standard level and higher level 
   1. Bias mitigation 
   2. Character consistency 
   3. Convolutional neural network (CNN) 
   4. Dataset curation 
   5. Denoising 
   6. Denoising diffusion probabilistic model (DDPM) 
   7. Diffusion model 
   8. Embedding-based approach 
   9. Image generation 
      1. Conditional 
      2. Unconditional 
   10. Image-to-image translation 
   11. Noise injection 
   12. Segmentation map 
   13. Text-to-image generation 
   14. Training stability 
2. Higher level only 
   1. Adversarial dynamic 
   2. D-dimensional noise vector 
   3. Generative adversarial network (GAN) 
      1. Discriminator 
      2. Generator 
   4. Hybrid model 
      1. Flow-based model 
      2. Variational autoencoder (VAE) 
   5. Latent space 
   6. Mode collapse