Skip to main content

Elements (LoRAs)

Introducing Elements - Our Low-Rank Adaption (LoRA) model system

Nick Lee avatar
Written by Nick Lee
Updated this week


Overview

Elements is Leonardo.Ai's custom implementation of Low-Rank Adaption (LoRA) models, offering a highly efficient and resource-friendly approach to nuanced image generation. It allows precise and intricate control over the style and visual outcome of generated images with the simple addition of dynamic modifiers.
​

With Elements, users can seamlessly blend various model styles to produce unique results that align perfectly with their creative vision.

About Elements

  • LoRA Integration: Leonardo.Ai has integrated an library of diverse Elements for Flux Dev, SDXL and the older SD1.5* and SD2.1* models (*Legacy Mode only).

  • Style Categories: Select one or combine up to 4 compatible Elements from a diverse library of finessing tools and artistic styles, such as "Infrared Photography" and "Oldschool Comic"

  • Adjustable Strength: The strength of each Element can be adjusted, allowing users to find the perfect balance for their creations. Do note that some Elements have an different effect when a negative value is used, something that can be explored by the user.

  • Element Training: In addition to the numerous Elements crafted by our team, Leonardo.Ai also offers Element Training for both Flux Dev and SDXL models to all premium users, allowing you to train your own Element on your specific style, object or character, unlocking even further customizability for your image generations. For more details, read the Element Training guide.
    ​

Benefits

Elements streamlines the creation of highly customized artwork and versatile workflows right from the Image Creation tool. Users can simply select their preferred styles and adjust the strength; the integrated LoRA models then influence the output behind the scenes, all while allowing the user the flexibility of being able to use their favorite models such as Leonardo Kino XL or Flux Dev.
​
Effectively, Elements allows users to create more unique outputs with their favorite model, providing more nuanced and granular control than can be achieved by using just a model alone.
​

List of Elements Available

SDXL compatible Elements

  • 3D Sculpt

  • Abstract Line Art

  • Celshaded Anime

  • CGI Noir

  • Coloring Book

  • Colorpop

  • Colorful Scribbles

  • Cosmic Retro

  • Cute Emotes

  • Cybertech

  • Dark Arts

  • Digital Painting

  • Dopamine Illustration

  • Dragon Scales

  • Dreamy Acrylics

  • Editorial Illustration

  • Fantasy Icons

  • Fiery Flames

  • Folk Art Illustration

  • Game UI

  • Glasscore

  • Glitch Art

  • Glowwave

  • Infrared Photography

  • Kids Illustration

  • Medieval Illustration

  • Modern Analog Photography

  • Oldschool Comic

  • Pop Surrealism

  • Psychedelic Art

  • Rainbowcore

  • Simple Flat Illustration

  • Simple icons

  • Soft Pastel Anime

  • Soft Retro Futurism

  • Solarpunk

  • Sparklecore

  • Toon & Anime

  • Vintage Christmas Illustration

  • Vintage Photography

  • White Ethereal

  • Woodcut Illustration

Flux Dev compatible Elements

  • Abstract Line Art

  • Cute Handdrawn

  • Folk Art Illustration

  • Medieval Illustration

  • Oldschool Comic

  • Toon & Anime

SD1.5 compatible Elements (Legacy Mode only)

  • Baroque

  • Biopunk

  • Celtic Punk

  • Crystalline

  • Ebony & Gold

  • Gingerbread

  • Glass & Steel

  • Inferno

  • Ivory & Gold

  • Lunar Punk

  • Pirate Punk

  • Tiki

  • Toxic Punk

SD2.1 compatible Elements (Legacy Mode only)

  • Surreal Collage

How to Use Elements

  1. Activation: In the Image Creation tool, change to an compatible preset (Flux Dev or an non-Phoenix preset). Then click on the button located on the left side of the prompt bar and click on View More to view all available Elements.

  2. Mix and Match: Select up to 4 Elements to combine them together. You can experiment to find different styles and combinations that you like.

  3. Strength: Adjust the strength for each Element by clicking on their thumbnail below the prompt bar to fine-tune and find the perfect balance for your needs. (Note: some Elements work well with negative weighting!)

  4. Generate: Click ‘Generate’ and see the results in seconds!

💡 Leonardo's tip:

  • Our recommended values for Element strengths are typically around 0.20~0.80.

  • When using more than 1 Element, it is best to avoid having all Elements set at high strengths. It is usually best to adjust them so that the overall weight of all Elements is 1.00 at most to avoid any visual glitches.


Frequently Asked Questions

Why can’t I use certain Elements together?

  • Due to the different base models that Elements are training on, some Elements are incompatible with certain models. This applies to user trained Elements as well.

How much does an Element generation cost?

  • Currently usage of Elements does not incur additional token costs.

Why are there unwanted artifacts and visual glitches when I use Elements?

  • If there are unwanted artifacts and visual glitches when Element are used, it may be likely that the strength of the Elements are set too high. It is always best to keep the strength at 1.00 or below.
    ​
    When using multiple Elements, it is best to keep the overall strength to an total value of 1.00. As a good practice, it is best to have secondary Elements at much lower strengths (0.10-0.20) and your primary Element at an higher strength (0.40-0.80).

I am unable to locate an Element I had previously used

  • In this case, it may be likely that you have an incompatible Preset selected. To view compatible Elements simply switch the active Preset.

Can I use Elements along with Image Guidance options such as Content Reference?

  • Yes, you may use Elements along with the various Image Guidance options for additional customizability.


​

What is the difference between an Element and a Model?

  • An Element essentially is like a mini model, which contains additional knowledge that is run on top of the model, allowing you to generate images with better specifiability. The influence can be easily adjusted, allowing you to finetune how much should the Element influence the end result.

  • An model has more generalized knowledge, which allows it to generate an large variety of subject matter. An finetuned model usually implies that the model has been further trained to be more specific in terms of aesthetics.

Feedback

We encourage and value user feedback. Please share your thoughts, experiences, and suggestions with us to help make Elements even better here.

Did this answer your question?