Sofa And Light Sdxl Lora

🚀 SDXL LoRA DreamBooth - prampampam/sofa-and-light-sdxl-lora

This project provides LoRA adaption weights for stabilityai/stable-diffusion-xl-base-1.0, enabling users to generate images with specific concepts in a Scandinavian style.

✨ Features

• LoRA Adaptation: These are LoRA adaption weights for stabilityai/stable-diffusion-xl-base-1.0, allowing for more targeted image generation.
• Multiple UI Support: Can be used with various UIs such as AUTOMATIC1111, Comfy UI, SD.Next, Invoke.
• Diffusers Library Compatibility: Compatible with the 🧨 diffusers library, providing a programmatic way to use the model.

📦 Installation

Download model

Use it with UIs such as AUTOMATIC1111, Comfy UI, SD.Next, Invoke.

• LoRA: download sofa-and-light-sdxl-lora.safetensors here 💾.
  • Place it on your models/Lora folder.
  • On AUTOMATIC1111, load the LoRA by adding <lora:sofa-and-light-sdxl-lora:1> to your prompt. On ComfyUI just load it as a regular LoRA.
• Embeddings: download sofa-and-light-sdxl-lora_emb.safetensors here 💾.
  • Place it on it on your embeddings folder.
  • Use it by adding sofa-and-light-sdxl-lora_emb to your prompt. For example, a photo of living room with sofa-and-light-sdxl-lora_emb and in scandinavic style (you need both the LoRA and the embeddings as they were trained together for this LoRA).

💻 Usage Examples

Basic Usage

from diffusers import AutoPipelineForText2Image
import torch
from huggingface_hub import hf_hub_download
from safetensors.torch import load_file
        
pipeline = AutoPipelineForText2Image.from_pretrained('stabilityai/stable-diffusion-xl-base-1.0', torch_dtype=torch.float16).to('cuda')
pipeline.load_lora_weights('prampampam/sofa-and-light-sdxl-lora', weight_name='pytorch_lora_weights.safetensors')
embedding_path = hf_hub_download(repo_id='prampampam/sofa-and-light-sdxl-lora', filename='sofa-and-light-sdxl-lora_emb.safetensors', repo_type="model")
state_dict = load_file(embedding_path)
pipeline.load_textual_inversion(state_dict["clip_l"], token=["<s0>", "<s1>", "<s2>", "<s3>"], text_encoder=pipeline.text_encoder, tokenizer=pipeline.tokenizer)
pipeline.load_textual_inversion(state_dict["clip_g"], token=["<s0>", "<s1>", "<s2>", "<s3>"], text_encoder=pipeline.text_encoder_2, tokenizer=pipeline.tokenizer_2)
        
image = pipeline('a photo of living room with <s0><s1> and <s2><s3>').images[0]

Advanced Usage

For more details, including weighting, merging and fusing LoRAs, check the documentation on loading LoRAs in diffusers.

📚 Documentation

Trigger words

To trigger image generation of trained concept(or concepts) replace each concept identifier in you prompt with the new inserted tokens:

• To trigger concept TOK1 → use <s0><s1> in your prompt.
• To trigger concept TOK2 → use <s2><s3> in your prompt.

Details

All Files & versions. The weights were trained using 🧨 diffusers Advanced Dreambooth Training Script.

• LoRA for the text encoder was enabled: False.
• Pivotal tuning was enabled: True.
• Special VAE used for training: madebyollin/sdxl-vae-fp16-fix.

📄 License

This project is licensed under the openrail++.