How AI image generators make a picture from your words
Guidance scale: what one number really changes in an AI picture
Some AI image generators let you set a number called the guidance scale. Stable Diffusion uses 7.5 if you do not change it. But this number is not a setting inside the diffusion model. So what does it really change?
-
1
Leave the number alone, and the software of Stable Diffusion 1.5 uses 7.5. Raise it, and the picture sticks closer to your prompt.
-
2
Yet the diffusion model only gives back a guess of the noise, and its parameters were fixed for good in training. Nothing in it is a setting for "closer".
The rest of this episode is for subscribers
Subscribe to read every premium episode. You can cancel when you want.
Paid yearly: $99 instead of $119.88