GalaxyStarlight's AI Stuff Repository

Pixiv - CivitAI - Tensor.Art

Contact me on Pixiv via messages here I do not have a public Ko-fi/FANBOX/Patreon/etc account!

Table of Contents

LoRA Downloads

LoRAs made by me

Fervent Tempo Delta ∆ (Rozaliya Olenyeva) (Honkai Impact 3)

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (Release 3R)

Usage

Compatibility
Should work with most anime-like checkpoints. Tested with AOM3, BOM, Hassaku, Grapefruit, PVC-AOAOKO, Cornflower X, etc.
Weight
Recommended 0.825; 0.75~0.90 range works nicely depending on how much flexibility you need.
Prompts
No trigger words are required. However:
You can put (mismatched stockings, long hair:1.15) on prompt and (tail, mechanical tail, weapon, sword, holding, holding weapon, holding sword:1.15) on negative prompt for better consistency.

> < Eyes (Cartoon Wink / Inequality Sign Smiley Expression) LoRA

Preview Image

Download

CivitAI - Catbox (Release 1)

Usage

Compatibility
Should work with most anime-like checkpoints. Tested with AOM3 (works best), and MeinaHentai. more testing will be done later.
Weight
I recommend 0.75~1.0; however, should you use other LoRAs with this, try a higher value such as 0.85~1.1
Prompts
(> <:1.2), adding one eye closed may be able to generate only one eye doing the > < expression, but this is gacha

Yayoi (Kantai Collection / Kancolle)

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (v1)

Usage

Compatibility Rating

S - AnyLoRA [Base Model]
A - Hassaku [Recommended], ^MeinaUnreal [Recommended]
B - MeinaMix, MeinaPastel, MeinaHentai, ^PVC-AOAOKO
C - AbyssOrangeMix3, AnythingV5, *^ReV, AnyHentai, Grapefruit
D - NovelAI, BloodOrangeMix

^2.5D model
*Not an anime oriented model
Prompts and Weight
Positive Prompt: <lora:YayoiKancolle:0.85>, (short hair with long locks:1.1), crescent hair ornament, blue eyes, black serafuku, long sleeves, pleated skirt, hair between eyes, black sailor collar, white purple hair, (midriff:0.5)
Negative Prompt: (white clothes, crop top, dark purple hair:1.1), long hair
Tags were not pruned to improve flexibility of the LoRA (i.e. different attire).

If you get unwanted stockings, put legwear, stockings in negative prompt.
If hair color is too white, remove white from white purple hair prompt (may occur with non-Meina models).
If hair color is too dark, add white hair to positive prompt.

G11 (Girls' Frontline / 소녀전선)

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (v1)

Usage

Compatibility Rating

S - AnyLoRA [Base Model]
A - Hassaku [Recommended], ^MeinaUnreal [Recommended]
B - MeinaMix, MeinaPastel, MeinaHentai, ^PVC-AOAOKO
C - AbyssOrangeMix3, AnythingV5, *^ReV, AnyHentai, Grapefruit
D - NovelAI, BloodOrangeMix

^2.5D model
*Not an anime oriented model
Prompts and Weight
Positive Prompt: <lora:g11:0.9>, long hair, grey hair, hair between eyes, open clothes, jacket, shirt, shorts, scarf, hat
Negative Prompt: None are necessary!
Tags were not pruned to improve flexibility of the LoRA (i.e. different attire).

Ichika / イチカ (Blue Archive / 블루 아카이브)

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (v1)

Usage

Compatibility Rating

S - AnyLoRA [Base Model]
A - Hassaku [Recommended], ^MeinaUnreal [Recommended]
B - MeinaMix, MeinaPastel, MeinaHentai, ^PVC-AOAOKO
C - AbyssOrangeMix3, AnythingV5, *^ReV, AnyHentai, Grapefruit
D - NovelAI, BloodOrangeMix

^2.5D model
*Not an anime oriented model
Prompts and Weight
Positive Prompt: <lora:IchikaV1:0.75>, swept bangs, sidelocks, black serafuku, black sailor collar, pleated skirt, hairclip, black choker, red neckerchief, red armband, long sleeves, (black wings:0.8), halo, blue eyes
You can replace blue eyes with eyes closed to get her signature closed eyes smile~
Negative Prompt: thigh length stockings, mismatched stockings, white serafuku
Tags were not pruned to improve flexibility of the LoRA (i.e. different attire).

Grace (Sound Voltex / SDVX) [5 Outfits]

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (v1)

Prompts and Weight

Weight: I recommend 0.9, like this: <lora:SDVX_Grace_E5_prod1:0.9>
Exceed Gear (variant with "double belt cover")
Replication Success Rate: B-
POS: pink eyes, (exceed gear), black elbow gloves, navel, headgear, (EG double belts)
NEG: (shorts, black tightsuit, black panties:1.2), garter belt, exposed crotch, gun, mechanical wings
Exceed Gear (variant without "double belt cover")
Replication Success Rate: B
POS: pink eyes, (exceed gear), black elbow gloves, navel, headgear, (exposed crotch)
NEG: (shorts, black tightsuit, black panties, belts, ribbon panties:1.2), garter belt, (EG double belts), covering crotch, gun, mechanical wings
NOTE: Ideally for both above it's black fingerless elbow gloves but ai is TERRIBLE at that.
Mixxion
Replication Success Rate: A-
POS: purple eyes, (mixxion), elbow gloves, navel, revealing clothes
NEG: (exceed gear), gun, mechanical wings, microphone, bra
SuddeИ Death
Replication Success Rate: C+
POS: pink eyes, (sudden death:1.15), (SD triple belts:1.15), (white shirt, short sleeves)
NEG: (exceed gear:0.9), microphone, gun, mechanical wings, (EG double belts:1.1), garter belt
NOTE: ideally fingerless gloves but yeah as explained above
Gravity Wars
Replication Success Rate: A
POS: pink eyes, (gravity wars), navel, black thighhighs, middle strip
NEG: shorts, garter belt, microphone, gun, mechanical wings, gloves
Vivid Wave (Experimental and not thoroughly tested)
Replication Success Rate: ???
POS: pink eyes, (vivid wave), black skirt, garter straps, navel
NEG: (exceed gear:0.9), microphone, gun, mechanical wings

Compatibility
  • MeinaHentai [Top Pick]
  • AbyssOrangeMix3-A1
  • MeinaUnreal
    VAE: blessed2
    Should work on most 2D and 2.5D models
    Tags were only partially pruned to improve flexibility of the LoRA (i.e. different, non-training data attire).
    Therefore, you can mess with the prompts given here if you want to change attire. Not officially tested, but should be feasible.

Geopelia (Phigros)

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (v1)

Usage

Prompts and Weight
Positive Prompts: <lora:geopelia-v100:0.75>, 1girl, long hair, black hair, bangs, hair ornament, yellow eyes, hair between eyes, hat, ahoge, long sleeves, (white dress:0.85), black pleated skirt, (red thighhighs:1.1), detached sleeves
Negative Prompts: chibi, red skirt, red dress, white skirt, black thighhighs, white thighhighs, mismatching stockings
Tags were not pruned to improve flexibility of the LoRA (i.e. different attire).

Yomiga (Rizline)

Preview Image
Release announcement post on pixiv here

Download

CivitAI - Catbox (v1)

Usage

Prompts and Weight
Positive Prompts: <lora:yomiga-v100:0.8>, 1girl, long hair, bangs, blue eyes, blue hair, bare shoulders, buttoned white dress, long sleeves, headphones, chokerr
Negative Prompts: animal
Tags were not pruned to improve flexibility of the LoRA (i.e. different attire).

Guides made by me

(Scuffed) 4-Step Image Decensor Guide [R-18]

Available At

CivitAI - Catbox (PDF port)

Stuff no longer available on CivitAI that I have on my disk

Those are not my works!

They were reuploaded here for preservation. I do not take credit.

OrangeJuiceMix [R-18]

Download

Catbox (v1.0)

Usage

Weight: 0.1~0.5
Prompts: peeing, idr if there's anything else

Defecation [R-18]

Download

Catbox (v1)

Usage

Weight: 0.3~0.6
Prompts: scat, feces pile, idr if there's anything else
Note: prone to causing body collapse.

Model Recommendations

HuggingFace links are preferred over CivitAI links if possible!

Checkpoint Recommendations (for generation)

S: MeinaUnreal, MeinaHentai [R-18], AbyssOrangeMix3
A: Hassaku [R-18], MeinaMix
B: PVC-AOAOKO [R-18], BloodOrangeMix

Checkpoint Recommendations (for training LoRAs)

S: AnyLoRA
A: NovelAI (Magnet Link) -> Refer here for installation help

VAE Recommendations

S: Blessed2 (Inbetween of NAI-VAE and kl-f8-anime2)
A: NAI-VAE (OrangeMixsVAE, AnythingVAE etc), kl-f8-anime2

Quality Control TI Recommendations

bad_prompt_version2, bad-artist/bad-artist-anime, bad-hands-5, EasyNegative, verybadimagenegative_v1.3

See also: this repo I found

Style LoRA Recommendations

add_detail (Controls detail level of images)

Other Stuff

For first timers: Minimum requirement for running SD locally is a NVIDIA 16XX GPU + 4GB VRAM, but this would have horrible performance. I recommend a 20xx GPU + 6GB VRAM or better.
Voldy Retard Guide - Entry level guide for first timers
catto / AI Porn Guide [R-18] - Has a massive LoRA repository

My SD Setup

Running locally (with this repo)
CMD Arguments: --port 6969 --opt-sdp-attention --no-half-vae
CPU: i7-13700H
MEM: 16GB, 6000 MHz
GPU: RTX 4050 Mobile, 6GB

Edit
Pub: 01 Mar 2024 16:57 UTC
Edit: 05 Mar 2024 14:54 UTC
Views: 318