Personalize Your AI Art Like Never Before

Today, we introduce the magic of fine-tuning to NightCafe. But, what is fine-tuning? It’s a new feature that lets you further customize our AI, teaching it to replicate faces, animals, objects, and styles.

How does fine-tuning work?

You can begin with just chosen 20 images (we suggest more than that) to train your model. Our powerful base AI image algorithm, SDXL 1.0 (Stable Diffusion), takes these images and merges its vast knowledge with your one-of-a-kind dataset. The outcome? Astonishing, personalized artwork. ✨

Why try fine-tuning?

Whether you’re looking to generate images in your favorite style or place yourself in imaginary scenarios, fine-tuning is your ticket to a more immersive creative experience.

How to Get Started with Fine-Tuning

Step 1: Access My Models

PC: Find it in the main menu header.
Mobile: Open the dropdown menu by tapping on your profile picture (top right corner) and select “My Models”.

Step 2: Click on Fine-tune a new model.

Choose your Model Type: Options include face, object or animal, and style. Name your model for further use.

Note: Fine-tuning is a PRO-only feature, but free users get 1 free face-model tune and 10 generations with it.

Step 3: Choose images for model training

Click on Choose or Create Dataset – here you can upload or choose images from your existing library that you want to use for training your model.

Start with at least 20 images. For more effective learning, diverse and numerous images are preferred. Once set, scroll down, agree to the terms, and click on “start training”.

Note: A dataset is a set of images that are used to train a model.

Step 4: Wait for Training Completion

This typically takes 10 to 30 minutes. You will get a notification once it’s done. Access your trained model either from the “Model” picker in NightCafe Studio or the “My Models” page.

Step 5: Use Your Model

Include your model’s “token” in your prompt (guidance provided below the prompt field). Then, unleash your creativity!

A Few Things to Remember:

• Non-PRO users get 10 free PRO generations after training a model. To continue using the model, consider upgrading to NightCafe PRO.
• Our Fine-tuning feature is still in BETA. This means there might be hiccups, and we can’t guarantee perfect results every time.
• Fine-tuning a model doesn’t automatically produce images. Once your model is ready, you can use it as you please to make your creations.
• There are some Fine-tuning restrictions for residents of state of Illinois, USA.

Using Your Fine-Tuned Model in a Prompt

What is LoRA?

“lora” or “LoRA” is the type of fine-tuned model that you can train on NightCafe. It is an acronym for “Low Rank Adaption” and is a method for quickly fine-tuning models on a small dataset.

Simply said, the LoRA training model makes it easier to train Stable Diffusion on different concepts, such as characters or a specific style. 

What is a token and how do I use it?

When you use a fine-tuned model, you need to add the token for that model to your prompt so that it can be interpreted in the context of the rest of the prompt.

The token is in the format:

<{type}:{name}:{optional weight}>

An example is <lora:My Face:0.8>. A prompt using a token would look like the following:

“A photo of <lora:My Face:0.8> riding an elephant”

or

“A unicorn in the style of <lora:Dark Fantasy:0.5>“

The weight is optional and can be omitted. Weights should usually be between 0 and 1. If omitted, the lora weight will default to 0.8. E.g. <lora:My Face> will be interpreted as <lora:My Face:0.8>. In the future there might be more types of models, which is why it’s used as part of the token.

Tips & Tricks for Making the Most of Fine-Tuning

To help you get started on the right foot, here are some tried-and-true tips:

Experiment with Lora Weights

Tweak the lora weights to see what makes images resemble your dataset most. The rule of thumb? The higher the weight, the more pronounced its impact. But tread cautiously: weights above 1.3 might not yield the desired results. Decimal point weights, like 0.88, can also be your ally in refining images.

Diversify Your Dataset

When training, don’t just take multiple pictures in the same setting. Variety is the spice of life, and in this case, the secret to a diverse dataset. Different angles, backgrounds, and expressions will result in richer and more varied outputs.

Seek Inspiration

If you’re hitting a creative wall, our presets are here to give you that much-needed nudge. Explore them and get those creative juices flowing!

Privacy Matters

Your models are private and safe. You are the sole user who can use them.

To wrap up, fine-tuning on NightCafe is not just another feature. It’s a new way for you to co-create with AI, combining your personality and preferences into the art. Dive in, experiment, and let us know what you think!