dreamlike photoreal 2.0

Introduction

Dreamlike Photoreal 2.0 is a photorealistic model based on Stable Diffusion 1.5, developed by Dreamlike.art. It is designed for generating high-quality, photorealistic images from text prompts.

Architecture

Dreamlike Photoreal 2.0 is built upon the Stable Diffusion framework, particularly utilizing the 1.5 version. It supports generating photorealistic images and is optimized for different resolutions and aspect ratios to enhance image quality.

Training

The model was trained using 768x768px images and performs well on resolutions such as 768x1024px or 1024x768px. It can adapt to various aspect ratios, improving the visual appeal of portrait or landscape images.

Guide: Running Locally

To use Dreamlike Photoreal 2.0 locally, follow these steps:

1. Install the Diffusers Library: Ensure you have the latest version of the Diffusers library installed.

2. Set up the Environment: Use a machine with CUDA-enabled GPU for best performance.

3. Load the Model:

   from diffusers import StableDiffusionPipeline
   import torch
   
   model_id = "dreamlike-art/dreamlike-photoreal-2.0"
   pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
   pipe = pipe.to("cuda")
   

4. Generate an Image:

   prompt = "photo, a church in the middle of a field of crops, bright cinematic lighting, gopro, fisheye lens"
   image = pipe(prompt).images[0]
   image.save("./result.jpg")
   

For enhanced performance, consider using cloud GPUs such as those offered by AWS, Google Cloud, or Azure.

License

Dreamlike Photoreal 2.0 is licensed under a modified CreativeML OpenRAIL-M license. Key points include:

• No hosting, fine-tuning, or inference on websites/apps without permission. Contact: contact@dreamlike.art.
• Hosting model card and files is allowed with proper credit.
• Outputs can be used commercially in small teams.
• No illegal or harmful content creation.
• Redistribution of model weights must include the same use restrictions and share the license.

For detailed terms, refer to the full license.