stable diffusion 3.5 medium turbo

Introduction

TensorArt Stable Diffusion 3.5 Medium Turbo (SD3.5M Turbo) is a high-performance text-to-image model derived from StabilityAI's stable-diffusion-3.5-medium. It focuses on stability and efficiency, supporting a variety of art styles and creative scenarios.

Architecture

The model is designed to offer:

• Turbo Performance: Faster generation speeds for multitasking and high-demand scenarios.
• Versatile Styles: Capability to handle photorealistic and abstract art styles.
• High-Resolution Outputs: Produces images with exceptional clarity and details.
• Easy to Extend: Integrated with LoRA technology for ease of customization and experimentation.

Training

The model leverages the latest advancements in diffusion models and LoRA technology to optimize performance and adaptability across different artistic styles.

Guide: Running Locally

1. Download the Model

   • Obtain the latest model files:
     • SD3.5M LoRA 8Steps Final
     • SD3.5M Checkpoint

2. Environment Setup

   • Ensure Python 3.8+ and PyTorch 2.0+ are installed.
   • Install required libraries such as diffusers.

3. Model Loading

   • Load and use the model as per the repository instructions. Use the provided workflows for integration with ComfyUI.

4. Example Usage

   • Using ckpt:

     import torch
     from diffusers import StableDiffusion3Pipeline
     
     pipe = StableDiffusion3Pipeline.from_pretrained("tensorart/stable-diffusion-3.5-medium-turbo", torch_dtype=torch.float16,)
     pipe = pipe.to("cuda")
     
     image = pipe(
        "A beautiful bald girl with silver and white futuristic metal face jewelry...",
        num_inference_steps=8,
        guidance_scale=1.5,
        height=1024,
        width=768 
     ).images[0]
     
     image.save("./test4-2.webp")
     

   • Using lora:

     import torch
     from diffusers import StableDiffusion3Pipeline
     from safetensors.torch import load_file
     from huggingface_hub import hf_hub_download
     
     repo = "tensorart/stable-diffusion-3.5-medium-turbo"
     ckpt = "lora_sd3.5m_turbo_8steps.safetensors"
     
     pipe = StableDiffusion3Pipeline.from_pretrained("stabilityai/stable-diffusion-3.5-medium", torch_dtype=torch.float16,)
     pipe = pipe.to("cuda")
     
     pipe.load_lora_weights(hf_hub_download(repo, ckpt))
     pipe.fuse_lora()
     
     image = pipe(
        "A beautiful bald girl with silver and white futuristic metal face jewelry...",
        num_inference_steps=8,
        guidance_scale=1.5,
        height=1024,
        width=768 
     ).images[0]
     
     image.save("./test1.webp")
     

• Cloud GPUs: Consider using cloud GPU services like AWS, Google Cloud, or Azure for optimal performance.

License

The model is licensed under the stabilityai-ai-community license. For more details, refer to the LICENSE.md file.