Stable Diffusion in 5 Levels of Difficulty: Level 3.5

Channel:
IMJONEZZ
Subscribers:
825
Published on ● Video Link: https://www.youtube.com/watch?v=evppsUX8cvI


Duration: 5:55
185 views
1

Tom Hanks bot makes his ML debut (I didn't want to do an actual deepfake, I don't like those a bunch), but he explains how to perturb latents (add extra noise) for img2img and how to regenerate similar images using those latents and starting from different steps within the image decoding process.

0:00 Tom says hi



Other Videos By IMJONEZZ


2023-03-18Run ChatGPT on a Raspberry Pi (not really, it's Alpaca)
2023-02-14Hogwars Legacy: Unforgivable Curses are not missable
2023-02-14Hogwarts Legacy: What do the Daedulus keys unlock?
2023-02-06High on Life is bananas
2022-12-01This video is fake: Stable Diffusion Level 5 - Textual Inversion & Dreambooth
2022-11-27Home alone 2
2022-11-19How to evolve Gimmighoul in Pokemon Scarlet & Violet (and why you want to)
2022-11-18How to evolve Bisharp into Kingambit in Pokemon Scarlet & Violet
2022-11-18How to evolve Charcadet into Cereuledge in Pokemon Violet
2022-11-12Stable Diffusion in 5 Levels of Difficulty: Level 4
2022-11-01Stable Diffusion in 5 Levels of Difficulty: Level 3.5
2022-10-28Stable Diffusion in 5 Levels of Difficulty: Level 3
2022-10-26Stable Diffusion in 5 Levels of Difficulty: Level 2
2022-10-25Stable Diffusion in 5 Levels of Difficulty: Level 1
2022-10-14Sometimes this game sucks my soul
2022-10-11Assassin's Creed Valhalla: What happens when you finish the Tombs of the Fallen?
2022-10-10How to make an RNN (LSTM) in Python (while True: learn() part 6 final)
2022-10-07How to make a model that uses Stochastic Gradient Descent in Python (while True: learn() part 5)
2022-10-07How to make a Genetic Algorithm System in Python (while True: learn() part 4)
2022-10-05Big Boy Legos
2022-10-01Transformers Encoder and Decoders: How To Make Your Own (And Why You Should)


Tags:
transformer
stable diffusion
midjourney
dalle2
dalle
latent diffusion
large language models
ai
ai art
text to image
image to image