# Shitty SDXL loras
[TOC]
## B-LoRA—inspired styles (Pony Diffusion V6 XL)
> Styles baked with a custom preset using Lycoris and then stripped down with blora_slicer.py for compatibility. First use takes pretty long at least on my forge setup but it'll be normal afterwards.
> The goal was to keep a good balance of style and compatibility with character/concept loras and base pony characters etc.


>Made in Abyss manga style lora
>Trained with score_9, source_anime
- [Made in Abyss v5 e10 out0to5_in12356_te1te2](https://pixeldrain.com/u/zH3zswq4)
	- [img1](https://files.catbox.moe/xs67mn.png) [img2](https://files.catbox.moe/kuctjk.png) [img3](https://files.catbox.moe/fp625o.png) [img4](https://files.catbox.moe/x8n331.png) [img5](https://files.catbox.moe/eh03vo.png) [img6](https://files.catbox.moe/lhxpry.png)
## B-LoRA—inspired styles workflow (Pony Diffusion V6 XL)
>Just the basic workflow, I have no clue what dataset sizes, learning rates, algos etc. are actually the best.
>This set of sliced traits is just what I ended up with, it may not be optimal for everything. Do some experiments!
**Training (sd-scripts):**
1. Copy this [blora_conv_ffnet_outallbut0_inall_mid.toml file](https://files.catbox.moe/xqm49q.toml) to use as a preset during training, you'll need [lycoris](https://github.com/KohakuBlueleaf/LyCORIS).
2. Example [.ps1 file using LoKr algo](https://files.catbox.moe/75yunp.ps1),Uses roughly 15-18gb VRAM, could be optimized more. You can increase the factor to 3 or 4 if you're desperate, I saw improvements going from 4 to 2 in early testing but I haven't tried again with final settings.
3. The .ps1 above uses immiscible noise, so you'll have to merge this pull request https://github.com/kohya-ss/sd-scripts/pull/1395:
``` bash
git fetch origin pull/1395/head:immis_noise
git merge immis_noise
```

**Slicing:**
1. Clone https://github.com/ThereforeGames/blora_for_kohya to your sd-scripts folder, the repo has more detailed install/usage instructions.
	1.1 Replace the slicer with this modified one [blora_slicer.py](https://files.catbox.moe/6zprgi.py) if you want the lora to retain it's metadata.
2. Add this to your blora_traits.json
``` json-object
"out012345_in12356_te1_te2":
	{
		"whitelist": ["lora_te1", "lora_te2", "lora_unet_output_blocks_0_1_", "lora_unet_output_blocks_1_1_", "lora_unet_output_blocks_2_1_", "lora_unet_output_blocks_3_1_", "lora_unet_output_blocks_4_1_", "lora_unet_output_blocks_5_1_", "lora_unet_input_blocks_1_1_", "lora_unet_input_blocks_2_1_", "lora_unet_input_blocks_3_1_", "lora_unet_input_blocks_5_1_", "lora_unet_input_blocks_6_1_"],
		"blacklist": ["proj_in", "proj_out", "alpha"]
	}
```
3. Run the slicer on your chosen epochs using the above --traits and you're done!
>python blora_slicer.py --loras ./input/{file} --traits out012345_in12356_te1_te2 --debug --output_path=./output/{no_safetensors}_out0to5_in12356_te1te2.safetensors

  Personally I add "input" and "output" folders to my blora folder and use this handy python script [folders_blora_slicer.py](https://files.catbox.moe/u943om.py) to run the script on all the loras in the input folder. It's very useful when you're testing multiple presets and multiple epochs.
  ``` cmd
  in your sd-scripts venv:

  cd .\blora_for_kohya\

  python .\folders_blora_slicer.py

-> grab the results from the output folder
  ```

## Styles (Pony Diffusion V6 XL)
>Trained with score_9, source_anime
>
>Mosha with new oven settings for baking, sketch is a good tag to use. V9 with shittier tagging was somehow superior at 1girl, standing but had some other issues with prompting. Hands often have issues in both.
- [Mosha v11 e25](https://pixeldrain.com/u/SeRJssRo)
	- [img1](https://files.catbox.moe/rog7h6.png) [img2](https://files.catbox.moe/ykr6rf.png) [img3](https://files.catbox.moe/wtsx43.png)
- [Alt. bake Mosha v9 e28](https://pixeldrain.com/u/KccKDeiX)
	- [v9 - v11 img](https://files.catbox.moe/ouzg7y.jpg)

>Trained with score_9, source_anime, eu03
>
>Quick bake with the 5th eu03 dataset from anon, mostly for science. e16 is probably the one you want.
- [eu03 e16](https://pixeldrain.com/u/AjF9jLWQ)
- [eu03 e28](https://pixeldrain.com/u/hVChbWPN)
	- [grid1](https://files.catbox.moe/e7d3tm.jpg)

>Trained with score_9, source_anime, tsukushi akihito
>
>9th epoch is probably the best for normal use. Last epoch colors/shading go a bit too hard but it's great for spooky stuff. 1.0 weight works fine but going down to 0.8 brings back more details to the backgrounds.
- [Made in Abyss 9th epoch](https://pixeldrain.com/u/pMcX6cwW)
	- [img1](https://files.catbox.moe/xpeubg.png) [img2](https://files.catbox.moe/w90bcc.png) [grid1](https://files.catbox.moe/874z7g.jpg) [grid2](https://files.catbox.moe/mbvbzf.jpg)

- [Made in Abyss last epoch](https://pixeldrain.com/u/LKvweCwE)
	- [img1](https://files.catbox.moe/3fybwe.png) [img2](https://files.catbox.moe/5wuo8y.png) [grid1](https://files.catbox.moe/iqe2xf.jpg) [grid2](https://files.catbox.moe/f8j9y7.jpg)

- [autismDPO extract lora](https://pixeldrain.com/u/spNUBH8x)

>Trained with score_9, source_anime, moshimoshibe
- [Moshimoshibe (token: moshimoshibe)](https://pixeldrain.com/u/8n7UpQt4)
	- [example](https://files.catbox.moe/qtyezp.jpg)
	- [example](https://files.catbox.moe/rxse7z.jpg)
# Shitty loras

## Styles (LoKr + Pivotal)
>Requires webui 1.7.0 / dev branch
>Feels more forgiving to bake while retaining prompt responsiveness & less spaghetti fingers with pivotal? Including the token in prompt seems beneficial but might be cope.

- [Raita (token: raita)](https://files.catbox.moe/zqnh5u.safetensors)
	- Has some issues with random impact effects at 1 weight, cba fixing the tags and rebaking.
	- [example](https://files.catbox.moe/vijy8k.png)

## Styles (LoKr)
- [Asanagi](https://files.catbox.moe/crfk5s.safetensors)
	- [example](https://files.catbox.moe/souox3.png)

## Styles
- [Youjo Senki manga](https://files.catbox.moe/r71fxm.safetensors)
	- [example](https://files.catbox.moe/d0kha4.png)
- [Isutoshi](https://files.catbox.moe/54qk8v.safetensors)
	- [example](https://files.catbox.moe/515e37.png)
- [Mosha](https://files.catbox.moe/03g93e.safetensors)
	- [example](https://files.catbox.moe/881hrq.png)
- [Shokugeki no Soma manga](https://files.catbox.moe/87a2jr.safetensors)
	- [example](https://files.catbox.moe/2zu2ru.png)
- [Kimura Takahiro (Code Geass official art)](https://files.catbox.moe/5s6rxj.safetensors)
	- [example](https://files.catbox.moe/pudypy.png)

## Styles (hires)
>Styles baked at 1200+ resolution, using these will generate mustard dolphin gas
- [ShindoL](https://mega.nz/file/L6gimABD#WZbRoZ8MOAqbJmuOTxI2Dt_b9ehxVK_IDOPbkwaFSE8)
	- [example](https://files.catbox.moe/yndjb0.png)
- [Made in Abyss manga](https://mega.nz/file/Sy4CnKTT#iCwCjPuqp0nLBio7vuTQH_nbl9-Ik5bH27w-gDzmKJ0)
	- [example](https://files.catbox.moe/d2ou3t.png)
- [Mizuryu Kei](https://mega.nz/file/uuxy3R7b#o29rFueoiVfU-j6s7UrNZS9gp8bdw1qpZQjyyAUgoGs)
	- [example](https://files.catbox.moe/qsxkgn.png)
- [Usagi Nagomu](https://mega.nz/file/vnoUEIKC#Xxjwd1_8YJmROQX7Y2Ama5ekc6jLRq44fusdpRAxOUM)
	- [example](https://files.catbox.moe/hgh18h.png)
	- [example](https://files.catbox.moe/lon58y.png)
- [Kubo Tite](https://mega.nz/file/i65j0L4L#N9dd0HsyaSOAOCa6MaYK-zSEX0Lzzrw_OxHZ6PPv-sA)
	- [example](https://files.catbox.moe/63u13w.png)
	- [example](https://files.catbox.moe/m2jg5u.png)
	- [example](https://files.catbox.moe/latqcz.png)
	- [example](https://files.catbox.moe/wmahsg.png)
	- [example](https://files.catbox.moe/2c10dm.png)

## Concepts
>Instant loss 2koma mating press, 2e4 adam and 2e6 ada. Which one is better? It's all gacha bullshit that depends on model/prompt/loras. 
>2e4 feels gentler on style and 2e6 feels a bit more consistent. I uploaded both so you can share the pain of indecision. More notes in metadata, tldr: controlnet good.
- [Instant loss 2koma mating press 2e4](https://files.catbox.moe/sq8ko3.safetensors)
	- [example](https://files.catbox.moe/gbyypr.png)
- [Instant loss 2koma mating press 2e6](https://files.catbox.moe/oxs8a4.safetensors)
	- [example](https://files.catbox.moe/6akgyc.png)
- [Instant loss 2koma mating press embed](https://files.catbox.moe/6kn1tg.pt)

## Questionable
[Kaiji](https://files.catbox.moe/lw21u6.safetensors)