Project Turntable: Adobe built a good feature

This is a five-minute demonstration of a new feature in Adobe Illustrator that derives a 3D shape from a 2D vector. You kind of have to see it to believe it.

#ProjectTurntable | Adobe MAX Sneaks 2024 | Adobe by Adobe (YouTube)

Demonstrator Zhiqin Chen selected a vector, “generated views” for it (which took a few seconds), then was able to rotate it along both the horizontal and vertical axis to reveal that the tool had extrapolated a complete 3D shape from the vector. Wherever he left the shape, the tool continued to treat it as a 2D vector that the artist could continue to manipulate. Finally, he showed that, even after a shape had been cloned several times, manipulations of the original could be applied to the copies—in their respective orientations—by “updating the views”. All of the 2D vectors continue to be just that, no different from vectors that had been drawn manually rather than having been generated by the tool.

Turning a dragon in 3d spaceThe tool uses voodoo to “pull” a 2D vector up into 3-dimensional space, then lets you choose how to map it back into 2D space. The model remains in the background, allowing the user to continue to choose a different extrusion at will—until, presumably, the link to the 3D space is broken by changing the 2D view on it manually, in which case it becomes an untethered copy. From there, the user can generate a 3D view from the new 2D shape.

He showed how the tool was even able to derive four legs for a horse that had been drawn with only two legs. This suggests that the tool has a map to indicate to which part of the “3D-shape space” a particular 2D shape should be mapped. You already saw it with the dragon and the warrior, where the effect was subtler but essentially no different. The tool has to know that the large oval on the dragon’s belly should be belled out in 3D space. I really wonder how generally applicable this is, especially when using shapes for which the tool has less training material.

Still, the demo was very convincing—this seems like a powerful feature akin to being able to copy/paste text out of images and videos in MacOS and iOS.