F L U X.1 dev Lo R A Cinematic 1940s

Introduction

Cinematic-1940s is a LoRA model developed to capture the vintage aesthetics of 1940s cinema. It enhances images with a monochrome, high-contrast palette typical of classic film noir, providing a nostalgic yet precise rendering of that era. The model is particularly adept at depicting charismatic characters, dynamic cityscapes, and dramatic lighting.

Architecture

The model is finetuned with the base model "black-forest-labs/FLUX.1-dev" using the LoRA technique. It focuses on delivering a cinematic experience by leveraging the capabilities of the diffusers library and safetensors format for efficient image generation.

Training

The training process involved finetuning the base model to specialize in 1940s cinematic elements. The model's training emphasized achieving a high level of detail and authenticity in its portrayal of the era's unique aesthetic qualities.

Guide: Running Locally

1. Install Dependencies: Ensure you have Python installed. Use pip to install the diffusers library and PyTorch with CUDA support.

   pip install diffusers torch
   

2. Import Libraries: Import the necessary modules in your Python script.

   import torch
   from diffusers import FluxPipeline
   

3. Load the Model: Use the FluxPipeline to load the model and the LoRA weights.

   pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)
   pipe.load_lora_weights('aixonlab/FLUX.1-dev-LoRA-Cinematic-1940s', weight_name='cinematic-1940s.safetensors')
   pipe.fuse_lora(lora_scale=0.8)
   pipe.to("cuda")
   

4. Generate an Image: Create an image using a prompt and save it locally.

   prompt = "a gorgeous woman and a man, cinematic_1940s"
   image = pipe(prompt, num_inference_steps=24, guidance_scale=3.0, width=768, height=1024).images[0]
   image.save(f"example.png")
   

5. Cloud GPUs: For optimal performance, consider using a cloud GPU service like AWS, Google Cloud, or Azure to handle the computation-intensive tasks.

License

The model is released under the flux-1-dev-non-commercial-license. For detailed licensing terms, refer to the license document.