USE HIGHER SAMPLING STEPS TO GET GOOD RESULTS
(i usually use 32, but here to get something less abstract you need at least 50)
CHANGING THE STEPS MAKE DIFFERENT RESULTS so you can try this if you want to
v1 is better for low-step gen (for example - 32), v2 is better for high step gen (i used 57) and gives more detailes.
Was made just for fun. Not that good, but its enough to make moody pictures
Used same pics on v1 (800 train steps) and v2 (1200 steps) with all the same generation inf. so you can choose for yourself which one do you preferer. (all pics were made with 57 steps of Euler-a)
Maybe ill make an overhauled version later, but this is good for now.
Made it 1200 steps training
Used same pics on v1 and v2 with all the same generation inf. so you can choose for yourself which one do you preferer. Some times v1 better, some times v2
Maybe ill make an overhauled version later
1. สิทธิ์ของโมเดลที่โพสต์ซ้ำเป็นของผู้สร้างต้นฉบับ
2. ผู้สร้างต้นฉบับที่ต้องการรับรองโมเดล โปรดติดต่อเจ้าหน้าที่ SeaArt AI ผ่านช่องทางทางการ คลิกเพื่อรับรอง
