YuzuLemonTea / README.md
thiros's picture
Update README.md
3b464f0
|
raw
history blame
1.95 kB
---
license: other
---
# YuzuLemonTea Mix models ☕
List of my experimental merge models
- [Recommended Settings](#recommended-setteings)
- [YuzuLemonMilk](#yuzulemonmilk)
- [YuzuLemonChaiLatte](#yuzulemonchailatte)
- [YuzuGinger](#yuzuginger)
# important notice(Jan 15/23)
According to bbc-mc's note, there is a possibility of bug that some token(prompt) can be ignored, when merge with "add difference" option.
Milk and ChaiLatte models are now replaced with bug-fix ver.
https://note.com/bbcmc/n/n12c05bf109cc
# Recommended Setteings
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 8, Clip skip: 2, ENSD: 31377, Hires upscale: 2, Hires upscaler: Latent (bicubic antialiased),Denoising strength: 0.54~0.7
Negataive Prompt: (worst quality:2), (low quality:2),inaccurate limb,lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, normal quality, jpeg artifacts,signature, watermark, username, blurry, artist name
*(worst quality:2), (low quality:2) are adjustable between 1.4~2.0
# Sample prompt
4girls,(a 3d reader of:0.8) (teenage loli children:1.2), (wearing intricate casual camisole, cute hair ornament,crop jacket,hot pants, tighhigh:1.1),
shiny brown skin,
looking at viewer, (alluring smug:1.2),
dynamic angle,
(onomichi street:1.2),fisheye
<img src="https://i.imgur.com/2JiZwFU.jpg" width="" height="1000">
# YuzuLemonMilk
Block merged model of Anything v3 and some real models.
Rather photo realistic.
<img src="https://i.imgur.com/qYK8DKn.jpg" width="" height="1000">
# YuzuLemonChaiLatte
Combination of a weight merge of ACertainModel and Anything-V3.0, and a block merge of several realistic models.
Rather anime-ish style with realistic background.
<img src="https://i.imgur.com/Tim9BZK.jpg" width="" height="1000">
# YuzuGinger
Add more anime models to YuzuLemonChaiLatte. Can be very anime looks.
<img src="https://i.imgur.com/4vc4HSL.jpg" width="" height="1000">