Figma has unveiled a suite of new AI-powered image editing capabilities, including object removal, isolation, and image expansion.
According to the company, these features eliminate the need to export images to external tools for editing and then reimport them. While generative models like Nano Banana excel at creating visuals, Figma notes that users often require precise editing tools that don’t rely on text prompts.
The platform has enhanced its lasso tool for more accurate selections. Users can now select an object to either remove or isolate it for repositioning—while preserving surrounding elements such as background texture and color. Additionally, selected objects can be fine-tuned for lighting, shadows, color balance, or focus.
Figma has also introduced an image expansion feature within its design suite. This is especially useful when adapting creative assets to specific formats that require extended backgrounds or added details—for instance, transforming a 1×1 image into a web or mobile banner—saving users from repetitive cropping and manual element adjustments.
In addition to these new capabilities, Figma has consolidated all image editing tools into a unified toolbar for easier access. From this toolbar, users can select objects or image regions, modify background colors, or add annotations and text. Background removal—one of the most frequently used actions on the platform—now occupies a prominent spot in the updated interface.
While competitors like Adobe and Canva rolled out object removal features years ago, Figma is now catching up with its own implementation.
The new editing tools are currently available in Figma Design and Draw, with plans for a full rollout across the Figma ecosystem next year.
Figma’s announcement coincides with Adobe’s same-day integration of similar capabilities into ChatGPT, where Figma is among the app’s initial partners—a collaboration first revealed in October. It remains unclear whether these new features will be accessible to users leveraging Figma through OpenAI’s tools.
