Lunaris is an all purpose model that combines a now total of 16(v1), 48(v2) and 72(v3) models, ranging from popular models to niche gems that I have found lying around somewhere on page 194. It listens to prompts well and has a distinct, 2.5D feeling that is adaptable with negative prompts and weights. It always produces some of my favourite results. Model updates are purely on a modernisation and for fun basis.
-Most of the images used to showcase the model had highres.fix used on them. The Cover image has it's unupscaled version showcased right after.
If you like the model, please feel free to leave a review, and like the model. It really helps me out!
Recommended Negative Prompt: (results may vary drastically with different negative prompts.)
worst quality, (low quality, normal quality:1.3)
Recommended Settings:
-DPM++ 2M Karras, Restart, DPM++ 3M SDE Karras and DPM adaptive are my favourite samplers. Warning, DPM adaptive takes longer than other samplers.
-I use 20, 35 or 60 steps on average.
-CFG from 4-12 recommended.
-Highly recommend a 2x highres.fix.
-I prefer NMKD-Siax, 0,35 denoise and 20 hires steps. BooruGan, Anime6B are also good options.
-I prefer NMKD-Siax upscaler for SD Upscale.
-Don't use Restore Faces.
描述:
more models, better dataset from the new models
训练词语:
名称: blessed2_fp16.pt
大小 (KB): 163462
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: lunaris_v2.safetensors
大小 (KB): 2082643
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
