Melanie Kyle
Image generation in Adobe Express
Generate course banner and content images with Adobe Express templates and embed them in Canvas with attribution and alternative text.
How to generate and embed course banner and content images in Canvas using Adobe Express
Course pages carry more meaning when their visuals match the content, but most teaching staff have neither the time nor the design tools to make them. This activity uses pre-sized Adobe Express templates published through TeachAI, so banner and column images fit standard Canvas layouts without manual resizing. Working from a short text prompt, participants generate an image with Adobe Express, judge which of the results is usable, and place it in Canvas with the institutional attribution and the alternative text that accessibility standards require.
00:08Open the image generation templates in TeachAI. In the TeachAI site, go to 'Canvas Course Essentials'. Under 'Other Essentials', select 'Image Generation in Adobe Express (using Adobe Firefly)'.00:53Open and remix a pre-sized template. Under 'Action', choose the template that matches the image you need: the 'Widescreen (16:9) remixable template' for a page banner, or the 'Square (1:1) remixable template' for column content. When the template opens in Adobe Express, select 'Remix' in the top right.01:58Set up the generative image options. In Adobe Express, select '+ Add content' in the left-hand panel, scroll to 'Generative AI', and choose 'Generate image'. Set the following:- Size: 'Widescreen (16:9)' or 'Square (1:1)', matching the template you opened.
- Content type: 'Photo', in the dropdown below the prompt box.
02:50Generate the image from your prompt. Copy the draft image prompt from your Canvas content page or module notes, paste it into the prompt box, and select 'Generate'. Four options are returned. If none of them suit, select 'Generate more'.03:10Add the attribution line to the Canvas page. While Express generates the image, open the target page in Canvas and select 'Edit'. Copy the attribution text from TeachAI, paste it on the line where the image will sit, set the font size to '8pt', and centre it.04:14Scale the image and download it. Select the generated image you want. Drag its corner handles until it fills the template canvas edge to edge. Select 'Download' in the top right, set the format to 'PNG', and select 'Download'.05:04Upload the image to the Canvas page. In the Canvas Rich Content Editor, place the cursor on the line above the attribution text. Select the image icon in the toolbar, choose 'Upload Image', select the PNG you downloaded, and select 'Submit'.05:31Add alternative text and check accessibility. Save the Canvas page. Select the accessibility indicator on the embedded image to open the accessibility panel. Either use the option to generate a description automatically or write your own description of the image, then save it. Confirm the accessibility score reads 100% and close the panel.
Template aspect ratios | Widescreen (16:9) for page banners; Square (1:1) for column content. |
Example prompt — widescreen banner | A photograph of an Australian small business owner sitting in a cafe with a laptop, notebook, and smartphone looking pensively before posting, thinking through ideas, natural daytime lighting, modern interior background |
Example prompt — square content image | A photograph of a smartphone screen displaying a social media feed, slightly blurred, with a person in the background looking pensively at the screen |
Attribution text | Image generated by RMIT using Adobe Express |
Attribution formatting | 8pt, centred, on the line directly below the image. |
Export format | PNG, at the size set by the template. |
Accessibility gauge | If the accessibility indicator does not appear on the image after saving, refresh the Canvas page. |
Related skills
- Create multimodal outputs — Participants produce images with a generative AI tool and publish them as part of a course page.
- Explore AI tool features — Participants work through the Adobe Express generative options — size, content type and regeneration — to get a usable result.
- Apply human judgement to validate AI outputs — Participants choose between generated options and check each one for realism and fit before embedding it in learning content.