Sword Art Online - Girl Pack (40)
its my 100 Uploaded LoRA on Civitai so a Upload Special :)
Infos
Clip Skip: 1 - 2 wont work!
Trigger: See "Girl List" below
Train Image Size: 768x768
Samples in Gallery Below - Because 20 Image Limit :/
SD Version: 1.5
Clip-Skip: 1 - so no NAI!
Trained on: AnythingElse V4
1.0 Girl List
administrator
alice
alicia
argo
asuna-alfheim
asuna-rl
asuna-sao
cardinal
dil
fanatio
freyja
kyouko
leafa
linel
lipia
lisbeth
mito
natsuki
nori
rika
rinko
ronye
rosalia
sachi
sakuya
shino
silica-alfheim
silica-sao
sinon-alfheim
sinon-gungale
siune
sortiliena
suguha
tiese
urd
yolko
yui-fairy
yui-human
yuna
yuuki
██ Generate Fine
██ Need more Workaround
██ Bugged
█ and █ will be fixed in 1.1
But pleace search your perfect Strenght <3
Old Gen Infos
Version 0.2: Season 1 + 2
Version 0.1: Season 1 Episode 1-25 Girls
Added in 0.2 (All 0.1 Include)
asuna_alfheim
npc_urd
npc_freyja
sinon
sinon_alfheim
sinon_gungale
yuuki
Version 0.1 Girls:
Alicia
Asuna
Asuna_Elf
Leafa
Lisbeth
Rosalia
Sachi
Sakuya
Silica
Suguha
Yolko
Yui
Yui_Fairy
Trained on: W:480 H:720
Triggerwords:
"name" - see list
anime art style - Trigger for Anime Style
See Preview Images for my Prompts.
Preview Images generate:
LoRA Load over: additional-Networks (work without it)
Model: abyssorangemix2_Hard
VAE: Anything-V3.0.vae.pt
Txt2Img and Upscale settings:
DPM++ SDE Karras
Steps: 20
480x720
CFG: 7
LoRA Strengh: 0.6
Upscale: 2x
Denoise: 0.6
Upscaler: Latent
Description
FAQ
Comments (9)
something i highly recommended to us on these girlpacks:
instead of using only the characters name as a trigger, add
"character -trigger-" (replace trigger with name duh )
this way, you get the perfect character with nearly no problems.
Can you do a boy pack?
can i PM you somewhere please? twitter, discord or something
Nope i wait for PM system here on Civitai :/ and on patreon but its just for supporter so plz dont use Patreon when you just PM me its money waste :D sry (unless you want to support me xD)
@SysDeep
the situation is, i love to place more girls in one picture with the latent couple extension. but for your loras, it doesnt work, or very hard to do, and when it does, the outcome is not enough satisfying.
i beleive its because of the model trained on(?).
i tried it with different models (including the anything else v4), clip skip 1 and 2, but the issue still there.
could you re-train your dataset on NAI and upload it as a different version? maybe it could resolve the problem. (would be nice for another loras too, like fairy tail or to love ru, etc.)
of course, im not asking for free, i'd leave some bucks as a tip, or sub you on patreon as exchange.
@siegxrain586 I know the Problem. But for a LoRA pack its hard because i train each character in own train process, i need for group many more (dataset x2) so one epoch will done in 18hrs ^^ when i have more free time i will upgrade this Pack with single lora for each character.
can you do these two the best characters in all sword art please thanks
Strea from Sword Art Online: Hollow Fragment (video game)
Kizmel from Sword Art Online: Hollow Realization (video game)
At time many Commissions so can't promise that it will be something in the next weeks / months neither sd1.5 nor sdxl ^^"
BUT on my list
No Strea??
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.