Illustrious v9:
Weight: 0.5-0.7
Flux v9:
Weight: 0.5-0.7
I also use 1.4-2.0 with Flux1-dev-bnb-nf4-v2
v9:
Tagging Method: "glowing-neon-your colour-your tag"
Examples: "glowing-neon-green-jacket" "glowing-neon-red-shorts"
Added Kimono
Added Tail (Fox)
Weight: 0.4-0.8
v8:
Tagging Method: "glowing-neon-your colour-your tag"
Examples: "glowing-neon-green-jacket" "glowing-neon-red-shorts"
Improved Glowing/Blacklight Effect
Weight: 0.4-0.8
v7:
New Tagging Method: "glowing-neon-your colour-your tag"
Examples: "glowing-neon-green-jacket" "glowing-neon-red-shorts"
New Training Data
Some colours and clothing types have a weaker glowing effect than others. When this happens, try increasing prompt weight and/or Lora weight. (Or leave as is if you don't care for the glowing effect.)
Improved Glow Effect for standard clothing: Shirt, Tank top, Shorts, Pants, Skirt, Jacket
Improved Colour Variety
Improved Background Variety
Added Hijab
Added Bikini
Improved Hair Glow
Improved clothing: Thong, Pantyhose
Improved Accessories: Glasses, Earrings, Choker
Weight: 0.4-0.6
v6:
Improved Accessories: Glasses, Sunglasses
Improved Hair Glow
Pony - Weight: 0.3-0.6
v5:
Improved Glowing/Blacklight Effect - Credit: Stapfschuh
Improved Skin Texture
Improved Accessories: Choker, Earrings
Pony - Weight: 0.3-0.6
SD 1.5 - Weight: 0.4-0.6
SDXL - Weight: 0.4-0.6
v4:
Improved Cosmetics: Fingernails, Face paint, Makeup, Lipstick
Improved Hands.
Pony - Weight: 0.4-0.6
SD 1.5 - Weight: 0.4-0.6
SDXL - Weight: 0.4-0.6
v3:
Added Footwear: Stiletto Heels, Platform Heels, Boot Heels
Pony - Weight: 0.4-0.6
SD 1.5 - Weight: 0.2-0.5
SDXL - Weight: 0.2-0.5
Description
FAQ
Comments (5)
since it's your first version, let me give you an image i made with the same theme that you can use to finetune your model.....
Like this isn't a real post, so incredibly helpful it looks lol xd, I hope I get a post like this someday too. :)
@happyhen i would do the same for you, but your three models are simple and therer is nothing that could be changed on them. if you are lost on something, you can still ask me. i create prompts for everything. my latest "victim" is this model JadeEpicPortraitWorldXL_V1.1 - v1.0 | Stable Diffusion Checkpoint | Civitai
@themancalledmojo999 to change? Not that I want to change anything, more like strengthen it. I mean, yes in general the dataset for the first model should be done on the "beta version" of the pony xl, the second model requires more of a change of approach if you want a more original style. Well, for the style I would not give up the images in the author's version, because everything that could be obtained in other ways is already done. So now it remains to generate quality neural versions, on which it will be possible to retrain the lora to improve the quality. At the moment I'm busy clearing out the excess from the current works, so no time to try to make the same style unfortunately.
@themancalledmojo999 there I see you mostly gave a prompt, I thought you meant generating images of a quality suitable for the dataset