The Ultimate 2024 Guide to Mastering Image Guidance by Leonardo AI

Hey there, AI art lovers! Ready to Level Up Your Leonardo AI Game? This time we will discuss Image Guidance by Leonardo AI. It is a functionality that will be your friend for life.

Whether you are new to the fast-growing world of AI art or an expert at it, this guide will help get your output images looking simply glorious in no time. Well, grab your drinks and comfy cushions people as we are going on a creative ride together!

Introduction to Image Guidance in Leonardo AI

 

Source: Flux.ai

Let’s get into what image guidance is all about in Leonardo AI. You know how sometimes you have this amazing image in your head but the AI just doesn’t seem to understand what you’re going for? Yeah, that’s where image guidance by Leonardo AI comes in.

Imagine having an extremely smart art assistant who can see! Image guidance does exactly that, it allows the AI to be directed by images instead of text prompts when generating the output. Cool right?

In this tutorial, we’ll be looking at different techniques of image guidance available within Leonardo AI like the legacy system and new Image Gen v2. This guide will cover everything from basic image-to-image transformations to advanced character referencing so that these tools become second nature by the end!

Accessing Image Guidance in Leonardo AI

Leonardo AI offers two ways to access image guidance: the new Image Gen v2 and the legacy mode. Let’s explore both, with a focus on why the new version is particularly beneficial for beginners.

Image Gen v2: Streamlined and User-Friendly

The way users get and use image guidance has been changed by the new Image Gen v2 in Leonardo AI. The new version focuses on beginners, making things easier and more intuitive.

Here’s how to access image guidance in Image Gen v2:

• First thing first, open the Leonardo AI interface then go to the Image Gen v2 section.
• Look at the main interface for an image icon. This single icon gives access to all available options of image guidance.

 

Source: Leonardo.ai

• Click this button so that a wide range of techniques can be displayed in relation to different types of image guidance.
• Every image guidance is described clearly in what it does so that you can understand it faster.

 

Source: Leonardo.ai

• All you have to do is choose one of these methods which will suit best for your project.

Legacy Mode: The Traditional Approach

While Image Gen v2 is recommended for its ease of use, it’s worth understanding the legacy mode for comparison:

• To get into the legacy mode of image guidance, you need to turn on the Legacy Mode option in the Image Gen tool. After that, you should click on the Image Guidance tab which will allows you to access the image guidance options.

 

Source: Leonardo.ai

• Choose what kind of image guidance you want from a drop-down menu.
• The legacy mode doesn’t provide descriptions for the image guidance techniques, assuming a level of prior knowledge from the user.

Why Image Gen v2 is Recommended

• Improved User Interface: The sleek, modern design of Image Gen v2 makes navigation a breeze, even for first-time users.
• All-in-One Access: Having all image guidance options under a single icon reduces complexity and saves time.
• Better Presets: Image Gen v2 comes with improved preset options, giving users a head start in their creative process.

Image To Image: The Art Of Transformation

Let’s start with what I think is the most important part of an image guide – Image to Image Guidance. Think of it as giving one photo for another creation, telling them “Hey take this and run with it!”. So basically, it keeps the overall look of an input image and color pattern in your output image.

Here’s how it works:

1. Upload a base image or use the previous generation.
2. Set strength (how much input has on output).
3. Add text prompt(s) which should be used as guiding light throughout the transformation process.
4. Click Generate & watch the magic happen!

Pro tip: a strength of 30-40% usually gives a good balance between new ideas and input image; too high and you basically just get a copy of your input, too low and the input won’t have much influence at all.

For example, let’s try Leonardo AI’s “Anime” preset:

Input image: A calm landscape with mountains and a lake.

 

 

Prompt: “Turn the landscape into a busy futuristic city watercolour painting in anime style.”

Strength: 35%

Try it out, see what amazing things you can come up with!

Output Image:

 

 

As you can see, the output image kept the color pattern and overall look of the input image.

Depth to Image: Adding Dimension to Your Creations

Next on the list of things to talk about about images, is depth to image. This is a really cool tool that uses the depth within an inputted image and uses it as a guide to create another one. It’s like giving your 2D pictures a 3D effect and is useful for keeping the input image’s spatial layout and 3D structure.

Here’s how you use Depth to Image:

1. Pick an input image with recognizable depths (landscapes are perfect for this).
2. Adjust the strength – remember more strength means more of the original picture’s depth.
3. Make sure your prompt is consistent with the input image for best results.
4. Generate then prepare yourself for mind-blowing!

For example, let us try using Leonardo AI’s “Lifelike Vision” preset:

Input Image: A forest path leading towards a faraway mountain

 

 

Prompt: “Transform the forest path into a magical crystal cavern, glowing crystals, ethereal atmosphere”

Strength: 1.2 (Depth to Image often needs a bit more value)

Output Image:

 

 

As you can see, the output kept the 3D structure and spatial layout of the input image.

Remember – consistency between your input image and prompt is key here; if I give the prompt to display a beach scenery where the input image only shows the mountains might not lead to the expected results. Although sometimes those unexpected outcomes are what make everything worth it… don’t you agree?

Pose to Image: Strike a Pose!

All right, let’s get into some body language with Pose to Image! This feature is about taking the pose from your input image and putting it into a new creation. It’s like digital charades, but better.

Here’s how you can make the most out of Pose to Image:

1. Choose an input image that has a clear pose.
2. Set the strength – often 1 works well here.
3. Craft a prompt that describes what you want in the new image but do not forget about the pose!
4. Generate and see your character strike a pose!

Let’s give it a try with Leonardo AI “Cinematic Kino” preset:

Input image: A ballerina in graceful arabesque pose

 

 

Prompt: “Futuristic robot mimicking ballerina’s pose, sleek metallic body, neon accents, cyberpunk style”

Strength: 1

Output Image:

 

 

Pro tip: This feature works best when your input image and prompt are in the same ballpark – trying to get a sitting character from a standing character might leave you scratching your head at the results!

Edge to Image: Sharpening Your Vision

Now let’s talk edges! Edge to Image is like giving the AI a coloring book page saying “Fill this in with something awesome”. It detects edges on your input image and uses them as guides for generating new images.

Here’s your Edge to Image:

1. Pick an input image with clearly defined edges.
2. Set strength – default 1 often works well here.
3. Craft a prompt that describes what you would like within those edges.
4. Generate and watch the edges come alive!

Let’s do an example using “Lifelike Vision” preset which you can find by turning on the legacy mode:

Input image: simple line drawing of cityscape

 

 

Prompt: “Transform cityscape into a vibrant colorful cartoon world, whimsical buildings, flying cars, and happy citizens”

Strength: 1

Output Image:

 

 

Edge to Image is great for keeping the composition of your input image while changing its style or theme entirely. It’s like giving a total makeover to the image but leaving its bone structure!

Style Reference: Channeling Your Inner Artist

All right art snobs, it’s time for style! Style Reference is like having styles from your Favorite Artist– you give it an image with a style you love, and it applies that style to your new creation. Pretty cool huh?

Here’s how you can rock Style Reference:

1. Choose a reference image with the distinctive style you want to emulate.
2. In Image Gen v2 you can use up to 4 style references – mix and match for some real unique results!
3. Craft a prompt describing what you want to create in that style.
4. Generate and be amazed by your stylish new masterpiece!

Let’s do an example using Leonardo AI “Concept Art” preset:

Style reference: Van Gogh’s painting

 

 

Prompt: “Modern city skyline in Van Gogh’s style, swirling sky, vibrant colours, bold brushstrokes”

Strength: Mid

Output Image:

 

 

Pro tip: Try different combinations of style references – you might stumble upon a totally new art style which is yours only!

Content Reference: Bringing Ideas to Life

Next up is Content Reference. It’s useful when you have a specific subject or composition in mind, but you want to change up the details. This generates the output image keeping the content and composition of the input image.

Here’s your Content Reference cheat sheet:

1. Choose a reference image with the content or composition you want to emulate.
2. Remember, this doesn’t transfer colors or textures – just the general shapes and layout.
3. Craft a prompt that describes your new vision, using the reference image as a guide.
4. Generate and see your new take on the reference content!

Let’s give it a try with the Leonardo AI “Diffusion XL” model present in Legacy mode:

Content reference: A photograph of a cosy living room

 

 

Prompt: “Transform the living room into an underwater scene, coral furniture, fish swimming through seaweed curtains, and bluish lighting”

Strength: Mid

Output Image:

 

 

Remember, Content Reference works best with regular SDXL models rather than Lightning models. And don’t forget to describe what you want to see in your prompt – the more detail, the better!

Text to Image: When Words Paint Pictures

Now let’s get literal with Text to Image. This feature is like having a really artistic friend who takes everything you say way too literally -– in the best possible way!

Here’s how to make Text to Image work for you:

1. Upload an image with text in it.
2. Craft a prompt that describes how you want that text to be represented visually.
3. Generate and watch your words come to life!

Let’s try an example using the Leonardo AI “Leonardo Graphic Design” preset:

Input image: A sign that says “Summer Vibes”

 

 

Prompt: “Visualize ‘Summer Vibes’ as a beach scene, sun-soaked sand, crashing waves, colorful beach umbrellas, and people enjoying the sunshine”

Strength: 1

Output Image:

 

 

This feature is great for creating visual representations of quotes, book titles or any text-based concept. Get creative with it!

Character Reference: Bringing Consistency to Your Cast

Last but not least, let’s talk about Character Reference. This is your go-to tool for creating consistent characters across multiple images. It’s like giving the AI a character sheet for your digital RPG!

Here’s how to use Character Reference like a pro:

1. Choose a reference image of your character.
2. Craft a prompt that describes the new scene or situation for your character.
3. Generate and see your character in a whole new light!

Let’s try an example with the Leonardo AI “Portrait Perfect” preset:

Character reference: A portrait of a stern-looking businessman in a suit

 

 

Prompt: “Transform the businessman into a space colony administrator, wearing a futuristic uniform, standing on an observation deck overlooking a Martian landscape”

Strength: Mid

Output Image:

 

 

Remember, Character Reference works best with humanoid subjects and generated characters. For real people or pets, you’ll need to be extra descriptive in your prompt.

When to Use Each Image Guidance By Leonardo AI

Let me explain when to use each image guidance by Leonardo AI. Think of it as an AI art compass – pointing you toward the right method for your creative vision.

1. Image to Image:

• When to use: You have a specific image in mind but want to change or improve it.
• Best for: Changing scenes; transforming art styles; adding elements to existing images.
• Example: Day cityscape transformed into a night scene with neon lights.

2. Depth to Image:

• When to use: Useful for preserving the spatial arrangement of an image when changing its content.
• Best for: Changing landscapes, reimagining architecture or adding depth to flat images
• Example: Making a forest path look like a futuristic city street while keeping the perspective and depth of the image.

3. Pose to Image:

• When to use: Works best if you want to keep a certain pose or composition but change what it is about.
• Best for: Redesigning characters, transferring styles, generating variations of one pose
• Example: Turning a ballet dancer’s pose into a martial arts stance.

4. Edge to Picture:

• When to use: Perfect when you would like to keep general composition and shapes but completely change style or content
• Best for: Dramatic changes in style; abstract interpretations; colorizing line art
• Example: Evolving simple sketch into detailed colorful illustration

5. Style Reference:

• When to use: If you want to apply a particular artistic style to your creation
• Best for Emulating famous artists, Creating consistent style across series, experimenting with new styles
• Example: Doing a portrait in Van Gogh’s “Starry Night” style

6. Content Reference:

• When to use: You want to retain an image’s general content and layout but alter its details or context.
• Best for: Scene reimaginings, thematic transformations, creating variations on a theme.
• Example: Changing the cityscape into an underwater scene with a similar layout.

7. Text to Image:

• When to use: Best suited for visualizing text-based concepts or creating literal interpretations of words.
• Best for Book covers; Logo designs; Visual representations of quotes
• Example: surreal clock-winged birds visualizing the phrase “Time flies”

8. Character Reference:

• When to use: Ideal for keeping character consistency through multiple pictures/stories/scenes/poses etc..
• Best for: Comic book artists; Character designers; and Anyone creating series with recurring characters
• Example: Putting your original character in different settings/outfits while keeping their distinctive features.

Models and Presets for Image Guidance

Now that we’ve covered the basics of our image guidance techniques, let’s dive into some models and presets to get you started. But keep in mind this is just the beginning – experiment!

Almost all models and Presets support Image Guidance except the Phoenix Model,  the image guidance features are not supported yet in this model which is a little bit sad because this model is versatile and gives really good output quality. For now experiment with the other models and Presets given below.

1. Image to Image:

• Works with most models and presets.
• Try “Leonardo Diffusion XL” for versatile results.
• Lifelike Vision, Cinematic Kino Preset or Portrait Perfect will give you photorealistic transformations.

2. Depth to Image:

• Use the DreamShaper V7 model or Lifelike Vision preset if you have depth information, it often gives excellent results.
• “3D Animation Style” can create interesting dimensional effects

3. Pose to Image:

• If you want to preserve realistic poses while changing other aspects use PhotoReal Model, Cinematic Kino Preset or Portrait Perfect Preset.
• For stylized character poses try ”Leonardo Anime XL”

4. Edge to Image:

• If you want interesting interpretations of edge maps use the “Leonardo Vision XL” or DreamShaper V7 model.
• “Anime Pastel Dream” creates soft beautiful renditions from edge information

5. Style Reference:

• For style transfers try ”Leonardo Diffusion XL”, it is more versatile
• When working with artistic style interpretations go for “Concept Art Preset”

6. Content Reference:

• Often Diffusion XL Model works better than the Lightning models.
• “Absolute Reality v1.6” can create content-based generations quite well.

7. Text to Image:

• To generate intriguing visual interpretations of text use the “Graphic Design Preset”
• When generating anime-style scenes based on text “Leonardo Anime XL” works fine.

8. Character Reference:

• The Preset “Leonardo Anime XL” is great for maintaining the overall style of an anime character.
• “Portrait Perfect” or Cinematic Kino Preset is recommended if you want your characters to look more realistic.

What do I do with AI-generated Art Knowledge Now?

 

 

Finally, we are here. We started with simple Image-to-Image transformation and ended up making complex Character References. But this is only a fraction of all possible image guidance methods. The magic happens when different models are combined!

What can you do? Anything! Why not try mixing Style Reference with Pose to Image and make characters imitating works of your favorite artists? Or maybe Content Reference along with Edge to Image could help you remake famous paintings from scratch?

Be as creative as possible – there are no limits. So go ahead, give them a shot but most importantly have fun while doing it! Who knows – maybe some AI is waiting for your idea to create a new masterpiece right now.

Also if something amazing (or terribly wrong – yes it also happens) comes out during this process why not share it with friends? The community would be interested in seeing what people can do with these tools!

Remember that every professional was once an amateur so don’t hesitate to ask questions and break the rules because we believe in you!

Now let the magic flow through you. Excited about the results!

Frequently Asked Questions about Image Guidance

Q1: Can I use multiple image guidance methods at once?

A1: Yes, you can. New Image Gen v2 allows using several Style References (up to four) during one generation process. For other methods usually it’s one per generation however who says we cannot run more than one or combine findings afterward?

Q2: Why doesn’t my output exactly look like the input image?

A2: It happens because the text prompt serves as a reference rather than a template; strength setting and the random nature of AI also contribute.

Q3: Why do Character Reference generations differ sometimes?

A3: Different prompts, random seeds or model interpretations can cause variations even though the main goal is to keep main features constant. To get more consistent results be detailed in character descriptions.

Q4: Can I use these techniques for commercial projects?

A4: Yes, you can. According to Leonardo AI’s terms of service generated images may be used commercially but always check the for most recent terms and consider reference image copyright.

Q5: Which strength setting should I use?

A5: It depends on how much impact you want the ant input image to have. Start with default settings then adjust as needed; remember higher strength values may be required when working with methods like Depth to Image.

Q6: Does Text to Image work with handwritten text?

A6: No it does not. Handwriting tends to confuse AI therefore clearly typed texts are preferred when using the Text to Image technique.