AI image generator Midjourney has unveiled a new “Vary” feature which offers users the capability to edit specific areas of an image. This feature emerges as a new addition to the platform's existing tools, Vary (Strong) and Vary (Subtle). Once an image is upscaled, the Vary (Region) option becomes accessible. Users are advised to engage the /remix mode to optimize inpainting outcomes. This mode can be activated by typing /remix and deactivated by retyping the same command.
Functionality and Usage
Upon selecting the Vary (Region) option, an editing interface appears, enabling users to pinpoint the image areas they wish to inpaint. Selection can be made using the Rectangle and Lasso tools. For precision, users can employ both tools, starting with the Rectangle tool for broader areas and refining with the Lasso tool1. The editor currently lacks a “subtract” function, but an Undo button is available for deselecting areas.
AI expert Matt Wolfe discusses the new feature in a YouTube video, emphasizing its potential and showcasing its application. The feature is particularly effective for larger image sections, ranging from 20% to 50%. However, it's not a universal solution for all inpainting challenges. Sometimes, the Vary (Subtle) option might be more suitable. The feature's efficacy is also influenced by the nature of the changes. For instance, adding a hat to a character is more feasible than placing a dolphin in a forest setting.
Midjourney's new Vary (Region), offers users an innovative way to edit and enhance images. While it holds promise, its effectiveness varies based on the nature and extent of the desired changes. As with all tools, understanding its strengths and limitations is key to maximizing its potential.