Stable Diffusion & ComfyUI Pro

[Subject] [Action/Pose] [Environment] [Lighting] [Style] [Quality Tags]

# דוגמה מעשית:
"a young woman sitting in a cafe, reading a book,
warm afternoon sunlight through window,
cinematic photography, shallow depth of field,
high detail, 8k, award winning photograph"

"ugly, blurry, low quality, distorted, watermark, text,
extra limbs, bad anatomy, bad hands, deformed,
lowres, worst quality, normal quality, jpeg artifacts"

# פתח ComfyUI → Manager → Install Custom Nodes
# חפש: "ControlNet" → התקן: comfyui_controlnet_aux

# הורד ControlNet Models:
# https://huggingface.co/lllyasviel/sd_xl_controlnet
# sd_xl_controlnet_openpose.safetensors → models/controlnet/

[Load Image] → [Depth Anything V2] → [Apply ControlNet (depth, strength=0.6)]

import cv2
import numpy as np

def extract_canny(image_path: str, low: int = 100, high: int = 200) -> np.ndarray:
    img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
    img = cv2.GaussianBlur(img, (5, 5), 0)
    edges = cv2.Canny(img, low, high)
    return edges

# https://huggingface.co/h94/IP-Adapter
# ip-adapter_xl.bin → models/ipadapter/

!git clone https://github.com/kohya-ss/sd-scripts
%cd sd-scripts
!pip install -r requirements.txt -q
!pip install xformers -q

!pip install transformers -q
from transformers import BlipProcessor, BlipForConditionalGeneration
from PIL import Image
import glob, os

processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")

for img_path in glob.glob("training_data/**/*.jpg", recursive=True):
    image = Image.open(img_path).convert('RGB')
    inputs = processor(image, return_tensors="pt")
    out = model.generate(**inputs, max_new_tokens=50)
    caption = processor.decode(out[0], skip_special_tokens=True)

    # הוסף trigger word
    caption = f"[trigger_word], {caption}"

    txt_path = img_path.replace('.jpg', '.txt')
    with open(txt_path, 'w') as f:
        f.write(caption)
    print(f"{os.path.basename(img_path)}: {caption}")

accelerate launch train_network.py \
  --pretrained_model_name_or_path="stabilityai/stable-diffusion-xl-base-1.0" \
  --train_data_dir="./training_data" \
  --output_dir="./output_lora" \
  --output_name="my_lora_v1" \
  --network_module="networks.lora" \
  --network_dim=32 \
  --network_alpha=16 \
  --resolution="1024,1024" \
  --train_batch_size=1 \
  --learning_rate=1e-4 \
  --lr_scheduler="cosine_with_restarts" \
  --lr_warmup_steps=100 \
  --max_train_steps=1000 \
  --save_every_n_steps=200 \
  --mixed_precision="bf16" \
  --xformers

# הגדרות Load LoRA Node:
#   lora_name: my_lora_v1-000800.safetensors
#   strength_model: 0.8
#   strength_clip: 0.8

# Prompt לבדיקה:
# "[trigger_word], portrait photo, professional lighting"