In Hyper-reality Offline – Creating Videos from Photos we saw how a 3D parallax style effect could be used to generate a 3D style effect from a single, static photograph and in Even if the Camera Never Lies, the Retouched Photo Might… we saw how the puppet warp tool could be used to manipulate photographic meshes to reshape photographed items within a digital photograph. But can we go further than that, and generate an actual 3D model from a photo?
In 2007, a Carnegie Mellon research project released an application called Fotowoosh that allowed users to generate a thee dimensional model from a single photograph:
Another approach to generating three-dimensional perspectives that appeared in 2007 built up a three dimensional model by stitching together multiple photographs of the same scene from multiple perspectives. The original Photo-Synth application from Microsoft allowed users to navigate through a series of overlaid but distinct photographs in order to navigate the compound scene.
The current version, which can be found at photosynth.net, generates a photorealistic 3D model that can be smoothly navigated through. Here’s an example:
Photosynth is very impressive, but is more concerned with generating navigable 3D panoramas from multiple photographs than constructing digital models from photographs, models that can then be manipulated as an animated digital object.
In 2013, researchers from Tsinghua University and Tel Aviv University demonstrated a more comprehensive tool modeling for generating 3D models from a single photo.
The Fotowoosh website has long since disappeared, and the 3-Sweep software is no longer available, but other applications in a similar vein have come along to replace them. For example, Smoothie-3D allows you to upload a photo and apply a mesh to it that can then be morphed into a 3D model.
So why not grab a coffee, find an appropriate photo, and see if you can create your own 3D model from it.