LoRA / README.md
Aotsuyu's picture
Update README.md
6f166b7
|
raw
history blame
6.6 kB
---
license: creativeml-openrail-m
---
Arbitrary epochs I use the most.
Sometimes different variants of strength, which are usually different epochs.
The 'resized' folders contain the same loras resized with the resize lora script and sv_fro, they're similar in strength but much smaller.
The token for most is either the name, is in the LoRA metadata or is not necessary.
<details>
<summary> Hogwart Uniforms </summary>
`hogsks`
Ex prompt: `hogsks, hogwarts school uniform, black robe, gray vest, slytherin, green tie`
</details>
<details>
<summary> WIP </summary>
Stuff that I needed to upload somewhere but I'm not sold on it being the best version.
</details>
<details>
<summary> archival </summary>
Mostly things that have a version that I find strictly better uploaded.
</details>
<details>
<summary> artstyles </summary>
Artstyle LoRAs.
Generated some basic previews so you get the gist of what the LoRA does. ~~They're not indicative of quality though, since they're just gens without any thought.~~
**WhiteMelancholy** is a mix between yagamishiro, dElysium, closer to death's depressed waifus and xlimo.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/whitemelancholy.png" height="12%"></details>
<br>
**bkubchibi** is 9948. It really wants the `chibi` keyword to function properly.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/bkub.png" height="12%"></details>
<br>
**crayoncirno** is trained on cute little crayon cirno drawings.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/crayoncirno.png" height="12%"></details>
<br>
**dElysium** is trained on disco elysium portraits.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/discoElysium.png" height="12%"></details>
<br>
**darkKukiv1** is a really decent noise offset (not multi-res) LoRA, you can use it to generate darker pictures. It's quite strong.
**darkv7** is another noise offset LoRA, one that doesn't affect overally quality of the image much, but being weaker as a result.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/dark.png" height="12%"></details>
<br>
**deaver** is 29999491.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/deaver.png" height="12%"></details>
<br>
**faintpaint** is a funky little LoRA mix between some pastelmix, whitemelancholy and diffLoRA to give it a more painty feel.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/faintpaint.png" height="12%"></details>
<br>
**gotagotay** is 30662116.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/gotagotay.png" height="12%"></details>
<br>
**hkv2** is 5691902, hollow knight art. It can generate some interesting sceneries.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/hk.png" height="50%"></details>
<br>
**kasuyabaian** is 78872596, trained on a dataset that hasn't been cleaned of text.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/kasuyabaian.png" height="12%"></details>
<br>
**omao-locon** is 29268484 though it's finicky.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/omao.png" height="12%"></details>
<br>
**pastelmix** stronger and weaker follow the formula of the original pastel mix model (andite's) but merged into a LoRA with different weights.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/pastelweak.png" height="12%"></details>
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/pastelstrong.png" height="12%"></details>
<br>
**rajawat** is 32622636. ~~I have no idea what this does.~~
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/rajawat.png" height="50%"></details>
<br>
**reverinth** is 66421798.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/reverinth.png" height="12%"></details>
<br>
**syokumurav4** is 5512911.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/syokumura.png" height="12%"></details>
<br>
**tekitoumidori** is 50544926.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/tekitoumidori.png" height="12%"></details>
<br>
**yagamishiro** is 625564.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/yagamishiro.png" height="12%"></details>
<br>
**zunusama** is trained on mostly his newer works.
<details><summary>preview</summary><img src="https://huggingface.co/Aotsuyu/LoRA/resolve/main/.previews/zunusama.png" height="12%"></details>
<br><br><br>
</details>
<details>
<summary> characters </summary>
Character LoRAs.
**SBKindred** is Spirit Blossom Kindred.
**bondrewd** is Bondrewd from MiA.
**erisallv2** is Eris Greyrat from Mushoku Tensei, with all of her outfits trained.
**gwen** is a League of Legends champion.
**kndrdsks** is Kindred, another LoL champion.
**lain** is Iwakura Lain. Lokr version included.
**mdtsk** is a 1dim Madotsuki from Yume Nikki.
**ozen** are Ozen from MiA.
**Qiyana** is yet another LoL champion.
**Yazawa Nico** is from Love Live.
**bangkisks** is the outfit/character from 暁Records's Rock 'n' Rock 'n' Beat PV.
</details>
<details>
<summary> concepts </summary>
Concept LoRAs.
**nnsks** is a nose removal LoRA. Pendin retrain.
**scuffedHorror** is a noise offset LoRA trained on a horror dataset.
</details>
<details>
<summary> diffLora </summary>
Magical little lora trained by overfitting on an image, then editing that image and training another lora and merging the loras with -1 and 1 weights.
This achieves a small lora that doesn't affect other stuff very much, mostly just providing the effect.
Most also yield interesting results at negative weights, and weights exceeding 1.
Due to how this method works, if the edited image is too different, the merge isn't that great and the original image might start showing.
</details>