Watch neural networks come to life as I try to build an image generator inside Scratch.
Text generation is already hard enough, but image generation is on another level. It involves ideas like stable diffusion, backpropagation, training data, and neural networks actually learning patterns. So naturally, I decided to make it harder by building it in Scratch — a coding language made for 8-year-olds.
In this video, I’ll break down how image generation works in a simple way while attempting to code my own version from scratch. You’ll see the process, the problems, the experiments, and whether Scratch can actually handle something this ambitious.
Can I make a real image generator in Scratch… or will the project completely fall apart?
This is version 2 beta, it has many possible picture, prompt following, and video generation:
turbowarp: https://turbowarp.org/1306130950
scratch: https://scratch.mit.edu/projects/1306130950/
Thanks to Nate Ziller for letting me use his song "Artificial Average Intelligence"
00:00 - Introduction
01:01 - how images work
01:27 - what we CAN'T do
02:17 - the game plan
02:48 - Neural networks
03:08 - why we use Neural Networks
03:44 - Coding
04:37 - training
06:13 - testing
06:53 - Part 2 and final version
Text generation is already hard enough, but image generation is on another level. It involves ideas like stable diffusion, backpropagation, training data, and neural networks actually learning patterns. So naturally, I decided to make it harder by building it in Scratch — a coding language made for 8-year-olds.
In this video, I’ll break down how image generation works in a simple way while attempting to code my own version from scratch. You’ll see the process, the problems, the experiments, and whether Scratch can actually handle something this ambitious.
Can I make a real image generator in Scratch… or will the project completely fall apart?
This is version 2 beta, it has many possible picture, prompt following, and video generation:
turbowarp: https://turbowarp.org/1306130950
scratch: https://scratch.mit.edu/projects/1306130950/
Thanks to Nate Ziller for letting me use his song "Artificial Average Intelligence"
00:00 - Introduction
01:01 - how images work
01:27 - what we CAN'T do
02:17 - the game plan
02:48 - Neural networks
03:08 - why we use Neural Networks
03:44 - Coding
04:37 - training
06:13 - testing
06:53 - Part 2 and final version
- Категория
- AI Картинки
Выполните вход или зарегистрируйтесь для отправки комментария.
Комментариев нет.
