LogoLogo
HomeExploreDocsAPIBlogContact
  • 🗃️Gooey.AI Docs
  • Changelog
  • 📖Guides
    • 🤖How to build an AI Copilot?
      • AI Prompting: Best practices
      • Curate your Knowledge Base Documents
      • Advanced Settings
      • Prepare Synthetic Data
      • Conversation Analysis
        • Glossary
      • Building a Multi-Modal Copilot
      • Frequently Asked Questions about AI Copilot
      • How to Automate Data Export?
    • 🚀How to deploy an AI Copilot?
      • Deploy to Web
      • Deploy to WhatsApp
      • Deploy to Slack
      • Deploy to Facebook
      • Broadcast Messages (via web or API)
      • Add buttons to your Copilot
    • ⚖️Understanding Bulk Runner and Evaluation
      • 💪How to set up Bulk Runner?
      • 🕵️‍♀️How to set up Evaluations?
      • How to use Bulk Run via API
    • 👄How to use AI Lip Sync Generator?
      • Lip Sync Animation Generator (WITH AUDIO FILES)
      • LipSync videos with Custom Voices
      • Set up your API for Lipsync with Local Folders
      • Tips to create great HD lipsync output
      • Frequently Asked Questions about Lipsync
    • 🗣️How to use ASR?
      • 📊How to create language evaluation for ASR?
    • How to use Compare AI Translations?
      • Google Translate Glossary
    • How does RAG-based document search work?
    • 🧩How to use Gooey Functions?
      • ✨LLM-enabled Functions
      • How to use SECRETS in Functions?
      • 🔥How to connect FirebaseDB to Copilot
    • 🎞️How to create AI Animations?
    • 🤳How to make amazing AI Art QR Codes?
      • API tips on AI Art QR Codes
    • 🖼️Create an AI Image with text
      • AI Image Prompting
      • API Tips for AI Image Generator
    • 📸AI Photo Editor
      • Build your avatar with AI
    • 🧑‍🏫How to use Gooey.AI’s Image Model Trainer?
    • 🔍Generate “People Also Ask” SEO Content
    • 🌐How to create SEO-Optimized content with AI?
    • How to use Workspaces?
      • How to use Version History?
      • How to add SECRETS in your Workspace?
    • 🍟How can I get free credits?
  • 😇CONTRIBUTING
    • Contributing
    • Documentation Style Guide
  • 🤓API REFERENCE
    • Getting started
    • API Generator
    • Rate Limits
    • Error Codes
  • 🍭ENDPOINTS
    • Copilot
    • Lipsync
    • Lipsync TTS
    • AI Art QR Generator
    • AI Animation Generator
    • Compare AI Image Generator
    • Gooey.AI on GitHub
Powered by GitBook
LogoLogo

Home

  • Gooey.AI
  • Explore Workflows
  • Sign In
  • Pricing

Learn

  • Docs
  • Blog
  • FAQs
  • Videos

Developers

  • How-to Guides
  • Get your Gooey.AI Key
  • Github
  • API Endpoints

Connect

  • Book a Demo
  • Discord
  • Team
  • Jobs

@Dara.network / Gooey.AI / support@gooey.ai

On this page
  • What is fine-tuning?
  • Why is fine-tuning important?
  • How to fine-tune Flux.1 on Gooey.AI?
  • Select your training model
  • Prepare your data
  • Select a model type
  • Choose a trigger word
  • Advanced settings
  • Run the training
  • Good Practices for LoRA Image Training
  • How to get a consistent character?

Was this helpful?

Edit on GitHub
  1. Guides

How to use Gooey.AI’s Image Model Trainer?

Make your custom image models with Gooey.AI

Last updated 21 days ago

Was this helpful?

What is fine-tuning?

Fine-tuning teaches the AI Image Generation model to recognize and generate new concepts. This is done by training/showing it a small set of example images. This allows you to customize the model's output for specific styles, characters, or objects.

Why is fine-tuning important?

As a creative practitioner, you might find that AI images all have a similar look/aesthetic - they lack originality, and feel flat. Fine-tuning can be a great way to bring your own style or concepts to AI Image generation, without struggling with a lot of prompt engineering.

Here, the artist Archana Prasad, who also goes by arcnoid, shared her original sketches in black ink. Notice that the style of the drawings is specific, with long faces and noses, and the textures and colors are very sparse. The use of human figures is also created in a highly abstracted and minimalist style. We used 12-15 of her images and created a fine-tuned model called arcn0id.

Here you can see the difference between the image generated with Flux.1 (with no fine-tuning) and the one on the right, which is fine-tuned with arcn0id style.

See this example, where we used Mughal miniature paintings (sourced from Creative Commons) to generate our AI images in the style of Mughal Miniature Paintings.

How to fine-tune Flux.1 on Gooey.AI?

Training an image model on Gooey.AI workflows is very easy. We use LoRA Image Training, which stands for Low-Rank Adaptation, a mathematical technique to reduce the number of parameters that are trained.

  • Smaller datasets: You only need 6-10 images to train a good model.

  • Faster training: Training a new concept with LoRA takes just a few minutes.

  • Smaller outputs: Trained LoRA outputs are small, easy to reuse, and share.

  • Combine concepts: You can combine several models to create unique outputs.

You can start here:

Select your training model

Select the training model you want to use.

Currently, we host:

  • Flux Lora Fast: Train styles, people, and other subjects - very fast!

  • Flux Lora Portrait: optimized for portrait generation, with bright highlights, excellent prompt following, and highly detailed results.

For this demo, we have selected “Flux Lora Fast”

Prepare your data

To start fine-tuning, you'll need a carefully selected collection of images. These images should represent your concept, or style, and these will “train” the model to create similar images.

These images should be diverse enough to cover different aspects of the concept. For example, if you're fine-tuning on a specific character, include images in various settings, poses, and lighting. Or if you are fine-tuning an artistic style, make sure you have diverse set of images in that style.

To fine-tune your model, you need:

  • 12-20 images for best results

  • Use large images if possible

  • Use JPEG or PNG formats

Upload the images to the “Input Images” section.

Here we are training a model for the artist Archana Prasad. These images have been shared with her permission.

Select a model type

With Flux Lora Fast, you need to select the type of model you are training. There are two options:

  1. Style: this refers to an artistic style (like Van Gogh style), or a general aesthetic style (like 90s analog film)

  2. Concept: this refers to an object, character, or clothing

Choose a trigger word

The trigger word refers to the object, style, or concept you are training on. Pick a word that isn’t a “real word”; most commonly used is TOK.

You can also use numerals or symbols, for example, instead of “mughal”, you could use mugh4l. The trigger word you specify will be associated with all images during training. Then, when you run your fine-tuned model, you can include the trigger word in prompts to activate your concept.

Advanced settings

  • Learning Rate: This is the rate at which the machine “learns” about the training. We recommend not to change this unless you’ve worked with machine training before

  • Steps: This is the number of times the machine “learns” and “trains” itself on the provided images. We recommend you try between 1000-1500. Please note: the higher the steps, the more credits you will use.

Run the training

  • Click on "Run" button

  • Once the training is ready, you will get a few options:

    • Generate Image: This will take you to a new tab with Gooey’s image generation workflow. Your model will be preloaded there, and you can start prompting to see if your new style is working.

    • Copy Model URL: This will allow you to copy the Model’s URL to use in our Image Generation workflow or other AI image generation tools like Fal.AI.

    • Download Model: This will download the model locally, and is useful for locally run workflows like ComfyUI.

Good Practices for LoRA Image Training

  • Prepare your dataset with care

  • Caption your images with good descriptive details

  • Make sure the images you are using are ethically sourced,

    • If you are using an artist's work, please get their permission

    • If you are using images from the internet, please check the copyrights

    • Try to use images that are in Creative Commons or open domains.

  • Training models can be expensive, so please check all your settings and data before you run the model

How to get a consistent character?

📖
🧑‍🏫
Model Trainer • Gooey.AI
Logo