stableface2.mp4
A set of simple notebooks to create a clear understanding on Stable Diffusion aspects like sampling, architecture, CFG, attention layers, inverse scheduler, pivotal inversion, image reconsutrction and prompt2prompt editting which we utilise to achieve hyperreaslitic results to edit any real-world face and build a pipeline to build your own face app.
(You can skip to Chapter 7 if you are already familiar with the basics of SD)
| Chapter 1: Sampling but not Inference A Simple example of how to create/sample an image with diffusers pipeline |
 |
| Chapter 2: Stable Diffusion Dissected Exploring the individual elements that results in image generation |
|
| Chapter 3: Spicing it up with CFG Exploring Classifier-Free Guidance to enhance image generation. |
|
| Chapter 4: Attention, Please! Overriding attention layer tomanipulate image generation process |
|
| Chapter 5: Attention Switcheroo Shared cross-attention layers between multiple images. |
|
| Chapter 6: Reality Check DDIM-inversion and pivotal tuning for accurate image reconstruction. |
|
| Chapter 7: FaceApp, Who? Synthesise learnings to create an advanced facial feature editing system. |
|
| Appendix 1: FaceApp, XL!! SD XL based face editing with corrective actions. |
|
