
Apoplanesis
Based on Initium, so it has finetuned clips.
Improves some things, like color, base style and amount of bloom appearing around characters. Thanks @Manityro for comfy workflow, even though i didn't use it :lmao:
Consider this a style tune. But style loras are still working. Well, ones that are trained on Initium, that is. Didn't test other, but should still be compatible.
Sampler/Scheduler: Euler A/Align Your Steps
Steps: 20-24 (Significantly higher will cause your image to look like shit, don't do that.)
CFG: 6-9. Adjust based on how much loras you're using.
Also don't be a degen and don't use pony loras.
VAE: decent one is baked in already.
Would suggest against using this model in your merges at high values, especially of tencs, since it can break compatibility with older unet.
Don't feel like really saying much more.
描述:
训练词语:
名称: AAA Anime VAE SDXL v1_4.safetensors
大小 (KB): 163429
类型: VAE
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
名称: apoplanesis_v1.safetensors
大小 (KB): 6775553
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success