I Tested Canva’s New A.I. Art… Here’s the Results
Is the new A.I. text-to-image function on Canva any good?
--
I’ve been seeing a lot of stories on Medium recently about A.I. art. Seems it’s set to be the next big thing in content creation.
But I didn’t realise until about a week ago that Canva, a platform I use all the time to generate free images for social media posts and other uses, now has the option to use a built-in A.I. image generator.
What is it and how does it work?
I only found out about Canva’s A.I. image generator because I read about it here on Medium. I’d love to credit the writer who brought it to my attention, but I honestly can’t remember now who it was.
Even though I’ve used Canva.com for ages, the text-to-image generator, which uses artificial intelligence to create images based on text that you type in, was easy to miss. I was eager to put it to the test, though, and wanted to share my initial results here.
Once I’d opened Canva’s ‘Text to Image’ app, the first thing I needed to do was type in a description of the kind of image I wanted to generate.
Canva showed me three examples for inspiration. These are:
- A panda riding a bike through a city with depth of field
- A light watercolor painting of koi fish in a pond
- Photo of a magical forest city from the future
I tried each one in turn, and these are the initial results:
A panda riding a bike through a city with depth of field
A light watercolor painting of koi fish in a pond
Photo of a magical forest city from the future
I thought these initial results were pretty decent. I mean, I wouldn’t hang them on the wall or anything but they were OK and would serve well in many contexts where unique, non-copyrighted images were required.
Just as importantly, they were so incredibly easy to generate. I mean, I really didn’t have to do anything apart from type in an image description and click a button, then Canva does the rest.
Next, I decided to try my own text. Given the title of this article, I thought I might as well try generating some images using the term ‘A.I. Art’. At this point, I should point out that when generating an A.I. image in Canva, there are several styles to choose from:
- Photo
- Drawing
- 3D
- Painting
- Pattern
- Concept Art
- ‘Surprise Me’
All the images you’ve seen above were all generated using the ‘Concept Art’ style. For my next test, I wanted to experiment with some of the other styles, starting with ‘Photo’. The app serves up two images at a time, so here are the first two images I got from each category, under the term ‘A.I. Art’:
Photo
Drawing
3D
Painting
Pattern
Concept Art
‘Surprise Me’
As you can see, the platform generated some pretty nice images, many of which would probably make great title images for blog posts or Medium stories.
The main problem I have with it is that the images all seem to be generated in the square format you see above, which is less than ideal. Many of the images generated don’t crop down easily into suitable title image formats and you’ll probably need to generate a few images until you find one which you can suitably crop.
For this purpose, I generated another random image for the title art of this article, one which did crop down quite well and which I’m very pleased with. On the whole, I think Canva’s new A.I. art tool is very useful and I’d definitely recommend you have a play around with it to see what you think.
