Entagma | Tutorial: Stable Diffusion & OpenCV for Face Detection and Automatic Compositing @Entagma | Uploaded 1 year ago | Updated 2 hours ago
Support us on Patreon: www.patreon.com/entagma
More on: www.entagma.com

Ever wondered what to do with your overfitted neural networks that seem to be good at generating exactly one thing? In this video, Mo has a suggestion: Augment it with a bit of OpenCV and use it to automatically detect facial features in an input images, composite those into a first run of your network and then rerun. Et voila!

Download Project File: entagma.com/downloads/double_diffusion_script.zip

00:00 - Intro: Overfitted Neural Nets
01:47 - The Idea
03:00 - Demo
04:40 - Creating Our Custom Script: General Setup
05:45 - Importing Libraries
06:16 - Custom Parameter Interface
07:48 - Importing Haar Cascade Classifiers
08:40 - Eye/Mouth Detection & Compositing
14:00 - Main Loop
17:30 - Recap / Where To Find Module Info
19:12 - Finishing Our Script & Testing
20:33 - Infamous Last Words