Stefano Corazza is a computer scientist at Adobe, but when I met him, he was pretending to be an ad exec selling a new cherry-flavored soft drink. But before he could create the ad, he needed a photo of the product. “You design something, then make it, then photograph it, then show the photos to an audience,” he says.
Unless you’re using Project Felix, the 3-D rendering tool Adobe unveiled today. In about five minutes, using little more than two stock images, Corazza whipped up a hyper-realistic image of a bottle of Cherry Blast. The bulbous matte red bottle sits on a convincing facsimile of a beachfront boardwalk. A splash of soda and a handful of cherries complete the scene. It looks obviously animated, but no less convincing than anything you’d see on a billboard.
With Felix, Adobe has made photorealistic 3-D rendering available to the creative masses without forcing them to master professional-grade CAD tools. “We are building something that people can use without specific 3-D knowledge,” Corazza says. “You don’t need to know about polygons.”
That’s because Felix relies on machine learning algorithms, not you, to create convincing images. Adobe is making a big push to use AI to remove the tedium of design by automating parts of it. Earlier this year, for instance, Adobe released Smart Tags, a tool that uses image recognition software to generate image keywords.
Felix pulls off a more sophisticated task. The algorithms driving it were trained, by scientists in the Creative Technologies Lab at Adobe Research, to understand spatial relationships between objects, and to replicate how light works. With the Cherry Blast ad, Felix automatically laid a grid over the beach landscape. When Corazza added a plastic bottle to the exposed patch of boardwalk, Felix knew exactly where a bottle that size should sit.
Next, Felix built a panoramic photo of the rest of the scene. The software knew exactly how to render the light, which would in real life diffuse differently as it interacts with plastic or wood or cherries. Felix’s algorithms tell the rendering agent V-Ray—the same one used for CGI in movies like Avatar—where to place pixels of light and dark, creating a 3-D scene that scans as believable.
Felix’s addition to Adobe’s suite of creative tools later this year will precede a host of new machine learning-infused tools. Some of them echo Felix’s functions, like Semantic Aware Sky, a tool that can replace the sky in any given image, and adjust the rest of the light in the photo accordingly. Likewise, Time of Day Hallucination lets photographers recast the direction of the sun in an image. With Felix, Adobe continues on a mission it started a long time ago: to make it easier for creators to create. Letting the machines take over, just a bit, is the logical next step.