Images
Generating & replacing images with AI
Type a few words, get a picture on your Canvas. Popcorn Editor can generate a fresh image from a text prompt or replace the artwork inside an image you already placed — keeping its frame, position and crop exactly where they were.
Both flows use the same image models the Assistant uses, and both spend AI credits.
Two ways in
There are two entry points, and they open the same floating prompt bar.
- Replace with AI — select an image, then click the sparkles button in the Properties panel (next to Replace). This regenerates that image, dropping the new artwork into its existing frame.
- Generate image — with a plain rectangle selected (or from the menu), choose Generate image to fill a shape with a brand-new AI image.
Because a replace reuses the current frame, the size, crop window and corner radius all carry over — only the pixels change.
The prompt composer
Whichever way you start, a slim composer floats at the bottom-center of the Canvas so you can still see your design behind it. It has four things:
| Control | What it does |
|---|---|
| Prompt box | Describe what you want. The placeholder reads "What do you want to create?" |
| Model | A ghost pill that appears when more than one image model is available. |
| Aspect-ratio chip | A small mono label showing the ratio the result will be generated at. |
| Transparent | A checkbox — shown only for models that support it — for a cutout with no background. |
Press Enter or click the up-arrow button to generate. The composer closes if you click away and deselect the target, but it never closes mid-generation.
Writing a good prompt
The result is only as good as the description. Name the subject, the style, the background and the lighting, and keep it specific:
"A single ripe avocado on a plain cream background, soft studio lighting, top-down, photorealistic."
If you want a cutout with no backdrop, say so — "on a transparent background" — and tick Transparent. For the full craft of prompting, see Writing prompts that work.
Choosing a model
Different models produce different looks, run at different speeds, and cost different amounts of credits. Some support transparency; others don't — the Transparent checkbox only appears when the selected model can honor it.
When you reopen Replace with AI on an image you already generated, the composer pre-fills its original prompt and model, so you can nudge the wording and try again without starting over.
Transparent backgrounds
Tick Transparent and the model returns a cutout with no backdrop — ideal for product shots, logos and stickers you'll layer over other artwork. Because the background is truly empty, the frame's own Background fill (if you set one) shows through cleanly. This checkbox is hidden whenever the chosen model can't produce transparency.
How the aspect ratio is chosen
You don't pick a ratio by hand. The composer reads the shape of the target frame and shows that ratio in the mono chip; the image is generated to match, so it drops in without awkward letterboxing or stretching. Want a different shape? Resize the frame first, then generate.
Credits and limits
Every generation spends AI credits. If you run out, the request stops and an upgrade prompt appears — credits come only with a subscription, so see Plans & AI credits for what each plan includes.
Iterating on a result
Rarely perfect on the first try — and that's fine. Reopen Replace with AI on the generated image, adjust the prompt (it's already filled in), and regenerate straight into the same frame. Once you're happy with the picture, use the Image fit controls and crop to place it precisely — the same tools you'd use on any placed photo.
Tip: keep a brand kit pinned and steer the Assistant to generate images that match your palette and style automatically. See Generating images with AI.
What's next
- Generating images with AI — the Assistant's own image workflow, including on-brand generation.
- Adding, replacing & fitting images — fit modes, cropping and framing whatever you generate.