So you think you’re a photo editing master. You can optimize exposure without batting an eye. You can apply an unsharp mask with the best of them. You even brushed out that clueless tourist from your Washington Monument photo so that no one could notice. Well, here’s a skill that you may be adding to your repertoire one day. The future of photo editing will be inserting and moving around 3D objects within a scene.
Most of us have had experience inserting a person into a photo or combining parts of multiple photos with Photoshop. But these hacked photos typically have one of two problems – they either 1) have to be planned in advance or 2) require extensive work to make the inserted portions of the image look natural (sometimes to no avail). Certainly Hollywood seems to have figured out 3-D CGI rendering – an unfortunate breakthrough for those of us who have had to endure Jar Jar Binks. But Hollywood CGI renderings require costly and complex computer operations which are out of reach for the normal consumer.
Kevin Karsch and his team at the University of Iliinois have been working on these problems and the results are pretty astounding. Using a powerful software algorithm, a user can take a single photo and add synthetic objects within minutes that will obey the physics and take on the lighting of the original scene. Imagine inserting a beach ball into a still image that will roll off a table onto the floor. Or imagine taking a photo of a blank living room and then digitally adding furniture piece by piece – each table leg perfectly rendered with the lighting from the original scene.
The beauty of the algorithm is its simplicity. A user must simply define the coordinates of the room using some simple line drawings and then define light sources. 3D objects can then be added to the scene and moved around seamlessly. One could easily see this algorithm dropping into consumer software soon. But don’t take our word for it, read Kevin’s paper here or watch his mind blowing video below.
Get More Camera Technica Greatness:
Subscribe to this Feed ◊ Follow us on Twitter ◊ Visit Our Google+ Page ◊ Follow Us On Facebook ◊ Email Us
