GENERATE.md

Generate Images

Video to follow, for now please follow instructions below

Generate

Warning

You cannot run this while training is still ongoing.

Step 1: Run the gradio interface

Note

Option 1: If you are using the same runpod for training and generation, follow this instruction

Start the gradio interface using the following command:

./gen.sh

Note

Option 2: If your are NOT using the same runpod for training and generation, follow this instruction.

1. Setup the workspace by running following command:

   curl -s https://geocine.github.io/flux/gen.sh | sh
   

2. Paste your huggingface token on the token file.

3. Start the gradio interface using the following command:

   python inference.py
   

Step 2: Ensure you have a LoRA file

Upload your LoRA file to the /workspace/output/lora folder. If you need to, else you can just use the existing LoRA files.

Step 3: Generate images

Note

First generation will be slower since it requires to load the model into the VRAM. The subsequent generations will be faster.

1. You will get a URL to access the Gradio interface.
2. Enter a prompt and click generate.
3. Your files will be saved to the /workspace/output/generations folder.