Generating images with AI

Popcorn Editor can make the picture as well as the layout. Ask the Assistant for imagery during a design task, or open Generate image to explore and place pictures yourself. The image panel supports batches, references, and a history of completed results.

Two ways to get AI imagery#

There are two routes, and you'll use both:

• Through the Assistant. When you ask for a design, the Assistant creates images as part of the build. You'll see Generating an image in its action timeline as it works. This is best when the picture is one piece of a larger request.
• Generate image. A dedicated panel for making one to four images from a prompt, comparing them, then inserting one or applying it to a selected frame.

The Generate image dialog#

Select an image or rectangle to target a frame, then open Generate image. You can also open it without a target and insert a result afterward. Enter a prompt, choose how many images to make, then click Generate. Each image is billed separately. The panel stays open while results arrive, so a slow result does not block the others.

When a frame is selected, the first completed image is applied there. Click another result to try it in the same frame. Without a target, click a result to insert it, or drag a result onto the canvas. Results also offer Copy prompt, Download, and Use as reference.

Tip: set the frame's size first. The aspect-ratio label follows the target frame, so a tall frame asks for a tall image.

Brainstorm different directions#

Choose a count from 1 to 4, then click Brainstorm to ask for distinct art directions from the same brief. Each direction gets its own image and label in the results. Use it when you want to compare ideas before committing to one composition. Brainstorm also spends credits for the images it generates.

The results stack includes recent generations, including images made through the Assistant. You can collapse it or change how many tiles appear in each row. A completed generation remains in your account history when you reopen the panel.

Use reference images#

Attach up to three images with the paperclip, or choose Use as reference on a result. Opening Generate image with an existing canvas image can also seed that image as a reference. References help a model follow a subject or look; describe in the prompt what to keep and what to change.

If you attach a reference while using a model that cannot accept images, the panel switches to a reference-capable model. Check the model and its settings before submitting. Remove an unwanted reference chip rather than assuming it will be ignored.

Choosing an image model#

The panel's Model selector lists the available image models. Pick one for the job; models differ in supported settings, speed, and credit use. Model-specific controls may appear beside the composer.

Next to it, a small label shows the target aspect ratio, read from the selected frame when there is one. Adjust the frame before generating if its proportions are wrong.

Transparent background#

For models that support it, the panel shows a Transparent checkbox. Turn it on to request a cutout with no background.

That's exactly what you want for:

• Logos you'll place over different colors.
• Stickers and die-cut labels, where the artwork needs a clean edge.
• Product shots you'll layer over a colored or photographic Canvas.

The checkbox only appears for models capable of producing transparency; if you don't see it, switch to a model that supports it.

Replace with AI#

Already have an image and want a different one? Select it and use Replace with AI. It swaps the current picture for a freshly generated one in the same frame.

If the selected image was itself AI-generated, reopening the tool prefills its stored prompt and model. You can tweak the wording and generate a variation. When the model supports references, the selected image can be used as the starting reference too.

Add to AI chat#

Sometimes you want the Assistant to work with an image. Right-click it and choose Add to AI chat to pin it as context. It stays attached across messages, so you can ask the Assistant to restyle it, place copies of it, or match its look in a broader layout request.

This is the bridge between a single generated image and a full design: generate the hero image once, pin it, then let the Assistant build the rest of the piece around it.

Credits and limits#

AI image generation consumes credits for each image, including images in a batch or brainstorm. If a request fails or credits run out, its result tile shows the failure. See How credits are used for metering and Plans & AI credits for the current plan options.

Tips for better results#

• Describe subject, style, and framing. "A ceramic coffee mug, matte terracotta, soft studio lighting, centered" beats "a mug."
• Reach for Transparent whenever the image needs to sit over other artwork.
• Mind the resolution for print. A generated image has a fixed pixel size; blow it up too far on a large print piece and it softens. Check effective resolution before exporting.
• Pin a brand kit so generated imagery leans on your palette and tone. Marketers can wire this up in Designing on-brand with a brand kit.

Where to go next#

Learn the wider image workflow in Generating & replacing images with AI and Adding, replacing & fitting images.