You Can Now Generate AI Images Using Canva, Here Is How

Canva has introduced its first in-house AI image generation model, Canva Craft, which lets users create images from text prompts. The new model is designed to generate detailed visuals with readable text and give users more control over how specific elements appear in an image. Read on to know how you can now generate AI images using Canva.

How To Generate AI Images Using Canva

Canva Craft is available through Dream Lab, Canva's AI image generation feature. You can use it to turn a written description into an image without having to create the design from scratch.

To get started, simply open Canva and sign in to your account. Head to Dream Lab, where you can access the image generation feature. Enter a prompt describing the image you want to create. Include details such as the subject, background, colours, lighting and overall style to help the AI understand what you have in mind.

Once you submit the prompt, Canva Craft generates an image based on your instructions. You can then review the result and make changes to your prompt if the image does not match what you wanted.

Canva says its new model is designed to make targeted edits without changing other parts of an image. This could be useful when you want to adjust one object or detail without having the AI alter the entire composition.

How is This Different From Other AI Image Generators?

Canva says Craft uses a two-stage process to generate images. First, the model interprets the user's prompt and works out how the different elements should fit together. This includes their size, position and importance within the image.

It then generates the final image based on that plan. According to TechCrunch, this approach is intended to improve the accuracy of objects and make text within generated images easier to read.

The model can generate images at resolutions of up to 2K and supports around 30 aspect ratios. This gives users more flexibility when creating visuals for different platforms and formats.

Another difference is the data used to train the model. Canva trained Craft using designs from its library, including posters, social media posts, presentations, infographics, menus and layouts that use a lot of text, rather than relying only on photographs.

This approach could make the model useful for people who want to create finished design assets rather than standalone images. Canva also plans to let users turn generated images into editable, layered designs using Canva Magic, its image-to-design tool.