Step 1: Prepare the image

1. Collect photos: Prepare 10 to 20 selfies or photos of yourself in different poses and backgrounds.

2. Naming convention: For easy retrieval, please name each photo using the same prefix (for example: "Heather1, Heather2").

3. Optional step: You can put all the photos into a folder or zip file.

Step 2: Start training

1. Access the model: Go to the model page and click the Train a FLUX LoRA button.

2. Add images: In the training interface, click the "Add Images" or "Pick a .zip file" button to add your photos to the training.

3. Set a trigger word: Enter a keyword that matches the one in your image file name—for example, "Heather."

4. Start training: Click the Start button to start training. The training process takes about 5 minutes.

Step 3: Generate prompts

1. Run Inference: When the training status shows "Completed", click the "Run Inference" button to start generating prompt words.

Fees and other information

• Training Fees: The cost of training FLUX LoRA on Fal is $2.

• Create a repository: You can create a Hugging Face repository on the training page to save your model.

🔥

High performing proxies with up to 1Gbps