Daniel Scott
@dan
In this post, we are going to explore Canva’s Artificial Intelligence (AI) Image Generator, also known as Magic Media. This powerful feature transforms basic text instructions (we call them prompts) into detailed images, cool graphics, and (in the near future) remarkable video clips (the last two I mentioned are still running on Beta testing versions). In this post we will be focusing on image and graphics generation as at the time I’m writing this, video mode is still a bit glitchy.
This post is based on my brand new Canva Design Essentials course! I’m really excited to share this new course with you and help you get started with this awesome design tool! When you become a BYOL member, you gain access to this course as well as my 30+ additional courses on Illustrator, Photoshop, Lightroom, InDesign, Figma, and more. As a BYOL member you will also enjoy personalized support, earn certificates, and tackle exciting community challenges. Head here to sign-up!
Ready to bring ideas to life? Let’s go!
AI Image Generator, or Magic Media, is part of the Magic Studio suite included in Canva. Not all AI tools in the suite are available in the Free version and some may have limited credits or features. Magic Media is open to all users, so hooray! We can all move on without any limitations.
However, if you want to sign up for my Canva Design Essentials and explore the full extent of Canva’s resources and tools, you can try the 30- day trial Pro account and, once you complete all the lessons and earn your certificates, decide if you want to keep the Pro subscription or go back to the Free version, considering your current needs and available budget. You can use this exclusive BYOL link to create your free or 30-day trial Pro Canva account.
Let’s see Magic Media at work. We’ll be creating a mockup Instagram post for a kindergarten classroom named Dragonflies, so we will be generating some lovely and fun images and graphics of, you guessed it, dragonflies!
From the Canva homepage, click on Create a Design and, inside the pop-up window, select Social media and click on the shortcut Instagram Post (Square) to create a new blank page.
Let’s create a new social media design to welcome our kindergarten children to their first day in class!
Let’s start filling our clean page with the cool AI images. Moving the mouse cursor to the left toolbar, we click on Elements. Inside the flyout menu, we scroll down until we find the AI image generator section and click on Generate your own.
We are looking for something very specific that matches our design goal, so let’s create our own images!
We are now inside a new panel, Magic Media. The first available line displays three categories:
Images: Highly detailed, pixel-based images and illustrations. These can be exported from Canva in multiple formats like JPEG or PNG.
Graphics (Beta): Detailed, vector-based artwork. Downloadable as pixel images in the Free version. To export these as scalable vector SVG files, you must subscribe to the Pro version.
Videos (Beta): A new Canva AI feature still under development, with some interesting results with landscapes and simple animated objects. Requires more prompt detail for better results, and video clips are best exported as MP4.
Below these categories, we find the Prompt box. This is where we describe what we want Canva AI to generate as an image.
There are lots of online resources about prompt design. We can give AI generators really basic descriptions or include enough detail to have as much control as possible on the final results. If you are generating a high detail photograph for print, you can instruct AI to consider technical details like angle, exposition, ISO values, lens type, etc. Knowledge is always a good thing!
Although there is not an ideal model to create an efficient prompt, styles may vary according to what you are trying to bring to the light, I like to follow a basic structure that usually delivers good results:
Subject (who or what), Description (what’s happening and where), Style or Aesthetic (realistic, steampunk, cartoon), Additional details (color, mood, composition).
In the prompt box, type in the description of the image you are looking for.
If you are new to generative AI and prompts, you can click the Inspire me button and Canva will display prompt examples that Magic Media can process into original images. Canva recommends a minimum of 5 words, but you can type something as simple as “cat” to get cool random results. We’ll keep things simple for this step-by-step guide, it’s best to focus on how the tool works, for now. Time and practice will bring us the rest.
After Canva provides the first prompt example, we can click Inspire again to see new options.
Even if we don’t use Canva’s examples, they are great for understanding how prompts are written.
I’ll type “dragonfly, resting on a leaf” as my first attempt. You can try your own ideas, that’s perfectly fine and it’s always fun to try alternative routes.
We’ll start with a basic description.
We’ll come back to Styles a few lines down and scroll down to the bottom of the Magic Media panel and have a look at Aspect ratio. We have three options to choose from:
Square: suited for Instagram feed and carousel posts, Facebook posts, LinkedIn and X/Twitter publications.
Landscape: great for YouTube thumbnails and videos, X/Twitter, Facebook and LinkedIn headers and banners.
Portrait: ideal for mobile screens and vertical contents, like Instagram and Facebook stories, TikTok, Snapchat, and Pinterest.
Let’s keep the setting at Square, since we are designing a square Instagram post.
Below the Generate Image button, we find the credits information. These credits are used to generate images and graphics in Magic Media. Free users have a limit of 50 text-to-image uses per account. Pro users have 500 credits to spend, renewed each month.
Let’s click the Generate Image button and move on to the next step.
I recommend the Pro subscription if you are going to work with AI content.
Magic Media generates four images from our prompt (only 1 credit is spent per batch). Don’t forget to read the rules for creating and using AI generated images to have a clear understanding of how this new technology impacts your job as a graphic designer. Do this often to make sure you stay up-to-date.
Click one of the thumbnails to place the image inside the page. You can pick your favorite or bring them all into the page, depending on what you are designing and the quality of the generated image. AI is a learning technology, it’s very common to find imperfections and glitches across these results, so closely examine the details and make sure that the result is accurate enough to be used professionally. In the near future, these will surely look a lot better and more consistent in quality and real-life representation.
Can you spot what’s wrong with this AI generated dragonfly?
Let’s turn up the complexity level and try a more detailed prompt, following the model I shared with you earlier in this post. I’ll try this: “A vibrant, realistic dragonfly with detailed wings, resting on a green lift in a natural setting, in front of a full lit moon.”
A properly structured prompt gives you additional control over the images Magic Media generates.
They look pretty nice. As you can see, there’s a lot more context and depth in these new thumbnails. Can you tell the difference in texture, color, and depth from the first set of images?
One important note about AI and generated content: it’s a trial-and-error process. It’s not easy to hit the mark with a first prompt, or a second, or a third. AI is good, but it doesn’t guess or read minds. Given a prompt, AI builds its own interpretation and versions of the description we’re typing, so there’s a lot of additional information sent back and forth until we come up with the closest representation of our initial vision. Don’t get frustrated if you are looking for a turtle in cowboy hat and boots and Canva gives you a real turtle wandering across a desert plain – no hat or boots. AI just needs some more instructions to understand exactly what you are imagining! Add something like “standing on two feet” or “vintage cartoon style” to your prompt description and you will be surprised!
Timeout #1
Feel like there is so much more to know about Generative AI? Once you complete this step-by-step guide, check out this MIT News article on what it is and how it works.
Let’s say that we like the second generated image and the other doesn't match our design at all. Can we ask Canva to generate more images like the second one?
Yes, we can! Click on the three dots that pop up on the corner of the thumbnail when we hover the mouse cursor over the thumbnail and pick Generate more like this.
I like the second image and that detailed moon. Canva can generate new content based on this sample.
As promised, let’s look at Styles. This is one additional layer of control that we have over a new generative AI image. Each Style holds a set of instructions that will complement the description we type in the Prompt box and generate a result that is closer to our initial idea.
There are many different styles to choose from, each of them suited for a specific kind of effect and response from our users. We can pick:
Inside Photography styles: Filmic, Photo, Long Exposure, Minimalist, Neon, etc.
Digital Art styles: Anime, Dreamy, 3D, Midcentury, Psychedelic, Etc.
Fine Art: Watercolor, Oil painting, Stained Glass, etc.
Can you imagine all the variations we can bring to our prompt description with a single click on a Styles button? This is incredible!
Above the Generate again button we find some default Style thumbnails. Let’s click on See all to further explore the available options.
Let’s have a look at all of the Styles that Canva has to offer.
Scroll up and down the list to see all the available options and pick the one that best fits your design needs. I’ll take Minimalist. You can go ahead and choose the one you prefer, that is perfectly fine!
Each Style gives a new generated image a different visual identity. Pick the one that best fits your goals.
Let’s see how applying the Minimalist Style changes the image Canva will create from our prompt. We click the Generate again button, give it some time to process the updated instructions, and once we have the results, all that’s left is clicking on the thumbnail that shows the result that is closest to the one we need. This will place the new generated image on the page.
Magic Media generated a new image from the same initial prompt, including the Style’s new directions.
How cool is that? Wow! This creates a completely different mood and emotional response to the image we are sharing with our users. I love AI features so much!
We’ve covered lovely and complex pixel images, but there may be a small doubt crossing your mind, right now. What about vector-based graphics? Can AI generate these as well? I guess I spoiled that earlier, when I showed you the Graphics tab at the top of the Magic Media panel. Yes, Canva can create graphics from text prompts and, if you have a Pro account, export them as vector SVG files ready for full resizing without any loss in quality.
Let 's find out how this works!
Scrolling back up to the top of the Magic Media panel, let’s click on Graphics. We’re going to keep the basic prompt we typed to create the first image set with Canva AI image generator: “dragonfly, resting on a leaf”.
Keep Aspect ratio set to Square, don’t add any Style for now, and hit the Enter key on your keyboard or click the Generate image button to start generating the new graphic.
Can we go to Vector-mode in Magic Media? Yes, but we will need a paid subscription to export SVG files.
Once we have the four sample thumbnails, we can click on the one that’s more appropriate for our kindergarten design, or any other project you decided to work on. The graphic will be placed on the page, in this case with a transparent background, which is extra awesome!
This graphic looks good! And Canva added those leaves that frame the dragonfly right in the center. Nice touch, AI!
Timeout #2
You’re almost there! Once you complete this post (and your awesome Canva design), dedicate some time to this interesting and comprehensive read about the future of Generative AI.
Let’s finish with testing AI generated graphics and Styles. You will notice that there are different Styles available in the Styles section when we create Graphics with Magic Media. This is one of Canva’s huge advantages: dynamically adjusting its User Interface (UI) features to the tool that is currently active. That's a great workflow.
Let’s choose one of the default Styles. I’ll go for Gold foil. You can choose any other Style that you prefer. All we need to do next is click the Style’s thumbnail and hit Enter or click the Generate again button.
We can set some really eye-catching Styles for vector-based graphics as well.
Give it a few seconds and there we are! Click the thumbnail of the generated graphic you prefer to place it on the page we’ve been working on.
We can always adjust graphic size or crop any unwanted parts to add a finishing touch.
That’s it, let’s add a welcoming message for our children and post our Magic Media dragonfly graphic on social media! I’m sure everyone will love it!
Congratulations if you made it this far! You are becoming an AI wizard, now!
Not bad for a first try, huh? Nice job, everyone!
AI technology will never stop amazing us and Canva is already delivering some really powerful results! Magic Media is an incredible productivity tool, even for those who only choose the Free version. I’m so excited with all the potential that I may have spent most of my monthly credits already. Time to focus on something else, I guess!
Join BYOL and have fun exploring Canva in my Canva Design Essentials course as well as my 30+ additional courses on Illustrator, Photoshop, Lightroom, InDesign, Figma, and more. As a BYOL member you will also enjoy personalized support, earn certificates, and tackle exciting community challenges. Get started here.
See you in class! – Dan