Generate by Prompt using Flux-schnell

Introduction

We design a module that generates photo-realistic & high resolution images based on user-defined prompts. While preparing the module, we utilize the pretrained model Flux-schnell at Hugging Face provided by black forests labs.

Setting Up the Environment

Using Conda (recommended)

1. Install Conda, if not already installed.
2. Clone the repository:

   git clone https://github.com/byrkbrk/generating-by-prompt-flux-schnell.git
   

3. Change the directory:

   cd generating-by-prompt-flux-schnell
   

4. Create the environment:

   conda env create -f environment.yaml
   

5. Activate the environment:

   conda activate generating-by-prompt-flux-schnell
   

Using pip

1. Download & install Python (version==3.11)
2. Clone the repository:

   git clone https://github.com/byrkbrk/generating-by-prompt-flux-schnell.git
   

3. Change the directory:

   cd generating-by-prompt-flux-schnell
   

4. Install packages using pip:

   pip install -r requirements.txt
   

Generating Images

Check it out how to use:

python3 generate.py --help

Output:

Generate images by prompt using Flux-schnell

positional arguments:
  prompt                Prompt that be used during inference

options:
  -h, --help            show this help message and exit
  --num_inference_steps NUM_INFERENCE_STEPS
                        Number of inference steps used during generating
  --device {cuda,mps,cpu}
                        The device used during inference. Default: `None`
  --enable_sequential_cpu_offload
                        Enables sequential cpu offload during inference

Example usages

Execute the following code blocks to generate the corresponding images displayed below. The results will be saved into the folder ./generated-images.

python3 generate.py\
 "an image of a turtle in Picasso style"\
 --num_inference_steps 4\
 --enable_sequential_cpu_offload

python3 generate.py\
 "an image of a turtle in Camille Pissarro style"\
 --num_inference_steps 4\
 --enable_sequential_cpu_offload

python3 generate.py\
 "an image of a turtle in Claude Monet style"\
 --num_inference_steps 4\
 --enable_sequential_cpu_offload

Generating via Gradio

Check it out how to use:

python3 app.py --help

Output:

Generate image using Flux-schnell via Gradio

options:
  -h, --help            show this help message and exit
  --enable_sequential_cpu_offload
                        Enables sequential cpu offload
  --share               Allows Gradio to produce public link

Example usage

To run the app on your local device, execute the following:

python3 app.py\
 --enable_sequential_cpu_offload

Then, visit the url http://127.0.0.1:7860 to open the interface displayed below: