Huggingface Repo: https://huggingface.co/lex-hue/Fluffity
Use e621 tags (no underscore), Artist tag are very effective (SD/e621 artist)
ClipSkip: Stop at CLIP layers=1 (SD=1, NAI=2) [Settings\Stable Diffusion\Clip skip]
Example Setting
Txt2img Setting
Steps = 15~35
Sampler = Any
CFG scale = 5~25
Negative embeddings = fluffity-neg ※[SD-WebUI\embeddings]
Hires. fix
Hires steps = 15
Denoising strength = 0.45~0.5
Hires upscaler = 4x-UltraMix_Smooth ※[SD-WebUI\models\ESRGAN]
Img2img Resize
Scale to = 1.5~2.0
Denoising strength = 0.45~0.5
ControlNet0 = softedge_hed, control_v11p_sd15_softedge
ControlNet0 Weight = 0.4~0.5
ControlNet0 Pixel Perfect = true
ControlNet1 = Reference
ControlNet1 Weight = 0.2~0.3
ControlNet1 Pixel Perfect = true
SD Setting
Stop at CLIP layers = 1
Eta noise seed delta = 0
Txt2img Animated Diff
Steps = 20~50
Sampler = Any
CFG = 7~25
ImageSize = 512~768
Number of frames = 8~16
Description
Better art-style and needs less emphasis for rare tags like gynomorph or such other stuff.
More adaptability with Loras.
It started to like animating stuff.. Dont ask me why..
There was an Beta before, What changed?
Trained it more as it wasnt as good as I expected
The Beta made weird images 30% in times
The Beta didnt even understand what some things even meant, that was weird as I had trained it to do that..
The new Version has an faster generation (3.8% Faster at some things (at 50 Steps))
FAQ
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.