If you’re like me, generating graphics with AI can be a double-edged sword. You love the imagery, but the text leaves a little to be desired.
Whenever I create something using Dall-e, I hold my breath that the text will be correct. However, nine times out of ten it isn’t.
I’m here to tell you there’s a fix.
Canva!
Here’s how to do it. Watch the step-by-step tutorial here.
- Create the image using Dall-e.
- Download the Dall-e image.
- Upload it into Canva.
- Select the image. Click Edit photo.
- Select Grab Text from the Magic Studio options.
- Canva scans the image and identifies text for editing.
- Edit the text.
- Save the file.
- Download the revised image.
In Action
In ChatGPT, I asked it to create an image using Dall-e for a LinkedIn post image 1920px x 1080px. The image should include:
- a group of people on a marketing team
- the headline “Best AI Marketing Tools for Small Teams in 2025”
- using a comic book style
Below is the original image. You can see that the text is not uniform and LinkedIn doesn’t have a capital “I.”

Once I’ve used the grab text feature in Canva, I can modify the text elements one by one to create a new image, which you see below.
What I’ve done.
- Removed the LinkedIn misspelled name in the top left corner.
- Removed the smudged AJI text in the middle of the table.
- Removed the image size 1920x text in the top.
- Replaced the headline text to be uniform. I also applied a shadow effect and increased the font size.
- Download the revised image for use.

Your Turn!
Now that you have this Canva hack, keep using those Dall-e images in your content design.
Want more Canva Hacks?
Download our Canva Activity Workbook and the Canva Activity Booklet 2024.