v2 trigger: happy new year
experimental LoRA, may be inconsistent
A testing LoRA for PonyXL. The examples are cherry picked. It makes wrong number of character blocks quite often. Use larger CFG if necessary.
v1 trigger (utf-8): 謹賀新年
Using trigger happy new year will partially work in autismmix but not usable in pony base. It's a mystery.
It doesn't meet my expectation (maybe I set it too high based on my XL impression) after struggling for several days. Not sure if I was doing something wrong but I feel some pieces are still missing for the puzzle. It's quite frustrating to see lessons I learned in SD1.5 LoRA training not working out here, after dozens of hours of waiting (it's painful when you're used to 20 minutes lightning fast SD1.5 LoRA trainings?).
描述:
训练词语: 謹賀新年
名称: happy_new_year_xlpony_24re_4_d32_謹賀新年.safetensors
大小 (KB): 223090
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
