Adobe unveils AI tool that can intelligently merge objects from images

Adobe has demonstrated an AI tool that can automatically merge different objects from images into a realistic-looking whole. The tool can estimate the scale of objects, match colors and lighting conditions, and generate shadows.

The AI ​​tool has been given the name ‘Project Clever Composites’ and has been demonstrated during the Adobe Max event. There, Adobe researcher Zhifei Zhang showed how he could create a realistic-looking image composition from two photos without having to do extensive manual image editing.

Zhang first showed a photo of an empty road with a mountain landscape in the background. The man then searched for a suitable photo of a car via the AI ​​tool. To find it quickly, he marked the desired target location on the first image and so the tool knew what perspective to take into account. Zhang pasted the photo of the car on top of the photo with the road and landscape, but that did not produce the desired realistic result. Afterwards, the man pressed the ‘Auto Composite’ button, which enabled the artificial intelligence to match the scale, colors and shade of the chosen car to the parameters of the first image in seconds. In another example, the researcher showed how he merged objects from images and how the AI ​​tool also took the lighting conditions into account.

It is not clear if and when the AI ​​tool will make its way into Adobe’s products. Zhang suggests to TechCrunch however, an improved version is already in the works that also supports three-dimensional objects.