Sign In

ALX LoRA: Self-teaching myself Stable Diffusion — img2img — Part 2

1
Jan 19, 2024
musing
ALX LoRA: Self-teaching myself Stable Diffusion — img2img — Part 2

Catch up on the previous article: ALX LoRA: Self-teaching myself Stable Diffusion — First attempts — Part 1


In case I change my profile picture, this is the current profile picture that I have:

Unfortunately, I did not create it. About a year ago, I used my pictures on some profile picture generator website that generated them for me. I can say that it went good. Tho, it had some imperfections, caused by one of my pictures that had an Instagram filter.

Normally, I could have used the spot healing brush tool in Adobe Photoshop, but I wanted to challenge myself to something unique — using img2img to fix it.

To my surprise, it went perfect:


However, it was not perfect on the first try. In the img2img prompt, there is a slider for strength. I thought it was the strength of the current image (the one above), but no — it was the strength of the prompt. Before I realized that I have to set it around 30% or so, I used 70% (the default), and it made me look like someone that would have deep opinions on the government if you were to even touch any subject regarding politics.

1

Comments