File size: 2,283 Bytes
db91c3a f6825f0 c62597e db91c3a ad4c524 f7860cd ad4c524 25f1683 ef23870 b6bb49a 621c060 42ef175 bf649a5 83851b8 98c3fe8 c3b274f 76259f3 3d81d83 98c3fe8 95e9765 5f9fd91 d321f5d 3b464f0 54be28f 34efe95 81a4486 73b980f 9b9fda0 c23ed38 e568cdb 8467018 2f01300 5deb74f 81a4486 37c4258 f43f506 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 |
---
license: cc0-1.0
tags:
- stable-diffusion
- text-to-image
---
# 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
VAE: "kl-f8-anime2" and "vae-ft-mse-840000-ema-pruned" are suitable
Steps: 20-30, Sampler: DPM++ SDE Karras or DPM++ 2M 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), (low quality) are adjustable between 1.4~2.0
- If you don't want 3DCG-ish paint, you can add (3d:0.8)~1.0 in Negative Prompt
# 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.
Works fine with positive (realistic) and (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">
<img src="https://i.imgur.com/96ka6hd.jpg" width="" height="1000"> |