My StableDiffusion automatic1111 v1.6.0 environment and notes

Other tools link

Training LoRA with Kohya's GUI

SET-UP automatic1111

1. Download StableDiffusion WebUI(automatic1111)

git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git

2. Install python v3.10

require 3.10 for xformers
https://www.python.org/downloads/

3. Modify webui-user.bat

• Add arguments --medvram --opt-split-attention --xformers --no-half in COMMANDLINE_ARGS
• Add path to python v3.10 C:\Users\USERNAMEHERE\AppData\Local\Programs\Python\Python310\python.exe

@echo off

set PYTHON="C:\Users\USERNAMEHERE\AppData\Local\Programs\Python\Python310\python.exe"
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--medvram --opt-split-attention --xformers --no-half

call webui.bat

4. Run(double click) webui-user.bat

5. Install automatic1111 extensions

Install these extensions from Extensions->Available tab.

Mandatory

• https://github.com/LEv145/--sd-webui-ar-plus
  • aspect ratio selector
• https://github.com/butaixianran/Stable-Diffusion-Webui-Civitai-Helper
• https://github.com/KohakuBlueleaf/a1111-sd-webui-lycoris.git
  • No longer needed for webui >= 1.5.0
• https://github.com/Bing-su/adetailer
  • after detailer
• https://github.com/Mikubill/sd-webui-controlnet.git
  • ControlNet models: https://huggingface.co/webui/ControlNet-modules-safetensors/tree/main
• https://github.com/pkuliyi2015/multidiffusion-upscaler-for-automatic1111.git
  • Tiled Diffusion & VAE
• https://github.com/hako-mikan/sd-webui-lora-block-weight.git
  • weights settings
  • LoRA Block Weight
• https://github.com/hnmr293/sd-webui-llul
  • Local Latent upscaLer
• https://github.com/toshiaki1729/stable-diffusion-webui-dataset-tag-editor
  • needs to tagging to train images
• https://github.com/a2569875/stable-diffusion-webui-composable-lora.git
  • Install from Extension index URL.
  • Use this extension with latent couple instead of RegionalPrompter.
    • RegionalPrompter happened some critical error in my env.
• https://github.com/ashen-sensored/stable-diffusion-webui-two-shot
  • Latent Couple extension
• https://github.com/hako-mikan/sd-webui-supermerger
  • LoRA merge tool: SuperMerger

Optional

• https://github.com/Coyote-A/ultimate-upscale-for-automatic1111.git
• https://github.com/nonnonstop/sd-webui-3d-open-pose-editor.git

6. Install VAE

Add files into \stable-diffusion-webui\models/VAE

• kl-f8-anime2.ckpt
  • https://huggingface.co/hakurei/waifu-diffusion-v1-4/tree/main/vae
• Counterfeit-V2.5.vae.pt
  • https://huggingface.co/gsdf/Counterfeit-V2.5/tree/main
• vae-ft-mse-840000-ema-pruned.safetensors
  • https://huggingface.co/stabilityai/sd-vae-ft-mse-original/tree/main
• orangemix.vae.pt
  • https://huggingface.co/WarriorMama777/OrangeMixs/tree/main/VAEs

7. Change automatic1111 settings

• User interface
  • Quicksettings list (setting entries that appear at the top of page rather than in settings tab) (requires restart)
    • Add sd_model_checkpoint, sd_vae, CLIP_stop_at_last_layers
    • Set Clip skip to 2 after UI reloaded

8. Download checkpoints, LoRA, LyCORIS

Add checkpoint models into \stable-diffusion-webui\models\Stable-diffusion
Add LoRA and LyCORIS models into \stable-diffusion-webui\models\Lora
https://civitai.com/

9. Install Textual Inversion

Add files into \stable-diffusion-webui\embeddings

• badhandv4
• EasyNegative
• negative_hand Negative Embedding

10. Install TensorRT Extension for Stable Diffusion

T.B.D.

TIPS, Another tools

Hires.fix settings & params

• Upscaler
  • 4x-UltraSharp, R-ESRGAN 4x+, R-ESRGAN 4x+ Anime6B
• Denoising strength
  • Recommended value is 0.25 to 0.50

SDXL and Refiner

• Refiner(and VAE) checkpoint link
  • https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/tree/main

TOOLS

• Kohya's GUI
• https://github.com/Sanster/lama-cleaner
  • Lama cleaner

ControlNet tile_resample upscaling params

Don't forget Disable ADetailer

• Common settings
  • Sampling method: Euler a, DPM++ 2M Karras, DDIM, etc
  • Sampling steps: Same as source image or 32-64 or higher
  • Upscaler: SwinIR_4x, R-ESRGAN 4x+, R-ESRGAN 4x+ Anime6B
  • Hires.fix : Enabled
  • Hires step: 0
  • Denoising strength: 0.3-0.5
• T2I
  • Set source image in Single Image tab
  • Enable Pixel Perfect (optional)
  • Enable Allow Preview
  • ControlType: Tile
    • Preprocessor: tile_resample
    • Model: control_v11f1e_sd15_tile
  • Down Sampling Rate: 1~3 (each individual case)
  • Control Mode: Default
  • Resize Mode: Default
  • Modify prompt (optional)
    • Change Lora weight from 0.7 to 0.4
    • Change Lora block weight
    • Change facial, hair colors, background and ...etc prompts
• I2I
  • Resize mode: Resize and fill
  • Denoising strength: 0.3-0.5
  • ControlNet Preprocessor: none

LoRA block weights format on a1111-v1.5.0

<lora:flat2:-0.3:lbw=1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0>

XX. settings, params

weights settings

F:0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0
NF:1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1,1

CHARA:1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0
BG:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,0,0

NONE:0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
ALL:1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
INS:1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0
IND:1,0,0,0,1,1,1,0,0,0,0,0,0,0,0,0,0
INALL:1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0
MIDD:1,0,0,0,1,1,1,1,1,1,1,1,0,0,0,0,0
OUTD:1,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0
OUTS:1,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1
OUTALL:1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1
ALL0.5:0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5

LC-Face:0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,0,1,1,1,0,0,1
LC-NCNF:1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,1,1,0,0,0,1,1,1,1,0,0
LC-NF:1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,1,1,0,0,0,1,1,1,1,1,1
LC-NC:1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0
LC-NPNBASE:0,1,1,1,1,1,1,1,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,0,0
LC-NP:1,1,1,1,1,1,1,1,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,0,0
LC-NIN:1,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1
LC-CT:0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1

Lora block weight

F,NF,CHARA,BG,NONE,ALL,INS,IND,INALL,MIDD,OUTD,OUTS,OUTALL,ALL0.5,LC-Face,LC-NCNF,LC-NF,LC-NC,LC-NPNBASE,LC-NP,LC-NIN,LC-CT