Leonado

About Image Guidance

Enhanced Image Guidance Options: Access a wide array of ControlNet options including Depth, Sketch, Canny, Pose, Normals, Pattern, QR, Lineart, HED, and Text Image Input for diverse image manipulation.

Quadruple Image Uploads for Premium Users: Premium users benefit from the ability to upload four separate reference images, each with the flexibility to apply distinct ControlNet enhancements and independent weight adjustments.

Revamped User Interface: Navigating Ai Images Guidance is now more intuitive than ever with a newly designed section on the AI Image Generation page, where you can effortlessly control both Image2Image and Image Prompt features.

Image credit: leonardo.ai

Benefits

Image Guidance elevates your creative control to new heights. All users gain access to an expansive toolkit of Image Guidance options for tailored image composition, while premium users enjoy exclusive access to three additional options and the capability to manage four reference images at once.

Dive into a wealth of finely-tuned options including Depth, Sketch, Pose, and QR, providing unparalleled command over aspects such as depth illusion, sketch-like artistry, accurate poses, and unique pattern translations. With a user-friendly new UI, effortlessly steering your artistic vision to reality is simpler than ever.

Image credit: leonardo.ai

Feature Breakdown

Depth to Image utilizes depth cues to amplify the three-dimensional quality of visuals. This enhancement bolsters the image's depth perception, making it an invaluable tool for tasks that require depth-sensitivity, such as sophisticated image editing and 3D rendering applications.

Image credit: leonardo.ai

Edge to Image adeptly captures and repurposes an image's layout by pinpointing object outlines and converting them into a line art sketch. This acts as a blueprint for the final artwork, where the AI interprets these edges in a manner akin to completing a coloring book. Increasing the strength preserves more of these defining lines, perfect for those aiming to maintain the original image’s composition while altering its artistic style. This feature is particularly effective for reimagining existing line art or harnessing the structural essence of a reference image

Image credit: leonardo.ai

Line Art: This feature distills an image to its essential outlines, creating a streamlined line art version that serves as a blueprint for the generation process. By employing Line Art alongside a Reference, you harness a powerful combination to guide the AI, ensuring that the fundamental structure of your reference image is translated into the final creation with clarity and precision.

Image credit: leonardo.ai

Edge to Image (HED Variant): Differing from Canny and standard Lineart, the Holistically-Nested Edge Detection (HED) option excels in capturing complex details and creating smoother, cleaner outlines of an image. It adeptly retains pertinent details from the reference image, offering a refined edge depiction for generating your artwork.

Sketch to Image: Tailored for compositions, the Sketch feature utilizes a rough sketch or scribble as a preliminary framework for the final artwork. Ideal for scenarios that demand a precise layout or structure, like architectural renderings, this function acts as a guiding draft, ensuring the AI follows your compositional direction.

Image credit: leonardo.ai

Pose to Image: This feature specializes in translating the postures of human figures from a reference image to the generated image. It detects the positioning of people or human-like figures and strives to mirror their stances. Particularly useful for character design and positioning, it's important to consider that more intricate poses may present challenges to the accuracy of replication.

Image credit: leonardo.ai

Normal Map: A tool well-known among 3D graphics professionals, Normal Mapping provides detailed information on an object's surface contours and structure, allowing for nuanced control over its visual texture and enabling dynamic lighting effects. While similar to depth maps, normal maps capture a finer level of detail, which is particularly advantageous for highlighting and enhancing the main subject of an image.

Image credit: leonardo.ai

Pattern to Image: The Pattern feature excels at reading and applying contrasting designs to generate compelling visual outputs. Best used with stark, high-contrast patterns, it effectively captures the style and form of the original design, faithfully transforming the source pattern into an artistic rendition within the generated image. This function is particularly adept at converting diverse pattern styles into eye-catching compositions that enhance the overall impact of the artwork.

Image credit: leonardo.ai

QR Code to Image: This feature is specifically designed to elevate the functionality of Quick Response (QR) codes by refining their design, improving scan-ability, and integrating them with unique visual styles or branding elements. It enhances the aesthetic appeal of QR codes, transforming them from mere data carriers into engaging, branded graphics that stand out while retaining their core purpose.

Image credit: leonardo.ai

Text to Image Input: This feature lets you create artistic interpretations of text by uploading an image with white text against a black backdrop. The generated result is a stylized textual artwork, enhancing the visual flair of the original written content.

Image credit: leonardo.ai

Kindly be aware that compatibility of new Image Guidance tools may vary with different base models. Inapplicable tools will be automatically dimmed and will display instructions for activation upon hovering over them. For your convenience, a reference table is provided, detailing which ControlNet options are available for each Stable Diffusion version, including those exclusive to premium users.

Image credit: leonardo.ai

How to use Image Guidance

For a swift start with Image Guidance, follow this straightforward guide:

1. Go to the AI Image Generation page. Beside the Generation History, you'll find the new Image Guidance feature—click to open it.
2. Upload your reference image into the Image Guidance panel. Premium users have the benefit of uploading up to four images.
3. From the dropdown, pick your preferred ControlNet option. Those not suited for your selected base model will appear greyed out. Look for the icon for tips on enabling greyed-out options. If your intent is to use a reference image as a starting point, opt for the Image to Image feature. To utilize the Image Prompt function, make sure Prompt Magic is activated from the left-hand sidebar.
4. Set the weight strength to your preference. Note: A stronger weight ties ControlNet closely to your reference image, which can be both good and bad, so it's wise to play around with this setting to strike the right balance.
5. Hit the Generate button to proceed!

Frequently Asked Questions

How much does it cost to use the new ControlNet features?

Each task utilizing an Image Guidance Option incurs a charge of 2 tokens. However, the Image to Image feature is exempt from additional token costs.

Which models and pipelines are optimized for Image Guidance?

Image Guidance is compatible with and enhances all models and pipelines, including Alchemy, PhotoReal, and Prompt Magic. Do note, however, that only Image Prompt is supported by Prompt Magic V2.

I'm experiencing inconsistent results. How can I improve them?

For more consistent outcomes, consider fine-tuning the weight settings or opting for a different reference image that may be more in line with your desired result.