Comments
Loading Dream Comments...
You must be logged in to write a comment - Log In
Used stable diffusion 2 my 1st attempt. The prompt used was made using a photo of a candle with melted wax used for the new stable diffusion 2 demo. I got a very wax candle look from this output which is what I wanted.
Settings on stability ai diffusion 2 were 4 image output. 50 steps of iteration. 49.9 guidance scale. Seed set to 1,643,202,343.
My opinion on stable diffusion 2: It's much better, and much worse depending on what you want. If you want a generic looking image that perfectly copy's someone else's style stay on 1 (i have gotten very good results on 1 just it can veer close to just copying others art at times, while other times it just has a touch of influence). If you want way more freedom to experiment with guidance and texture use 2. You get results that are way closer to what you describe in 2 but they remove being able to directly use artist as style altogether. Be warned if you set guidance really high even on good internet speeds, it can freeze in it's current form.