Driving through Istanbul Maslak in Red-Cyan Anaglyph Stereoscopic 3D

This is a 640x480 pixel 2D PAL video shot in Istanbul, Turkey with a 150 dollar Canon pocket digital-camera. The footage was then sharpness optimized and 2D-to-3D converted with a prototype Stereo 3D conversion algorithm called "DeepCam 3D".

DeepCam 3D is a work-in-progress, V 0.6, 2D-to-3D conversion algorithm written in C-Sharp NET. It currently converts 2D Standard Definition video to Stereoscopic 3D at a speed of 1 frame every 2 seconds (0.5 FPS) using one CPU core. DeepCam 3D performs two operations on 2D footage during that time period. 1) Sharpness optimization using a custom digital footage refocusing technique. 2) Conversion to Red-Cyan Anaglyph Left-Eye and Right-eye Stereoscopic 3D.

Future versions of DeepCam will likely be multi-threaded (so as to take advantage of multi-core CPUs) and also GPU accellerated. We expect a GPU accellerated version of DeepCam 3D to reach realtime Stereo 3D conversion speeds in Standard Def (20 - 30 FPS or more). This is a realistic goal as DeepCam 3D is written to be as short, simple and fast as possible. HD content in 720P/1080P will likely take longer than realtime to 3D convert, as the pixel count per frame in HD is much more than in Standard Definition PAL/NTSC.

Please leave us a comment if you like this footage. If comments are largely positive/useful/constructive, we will post more Stereo 3D videos of Istanbul and its surroundings.

Category:

Science & Technology

Tags:

• 3D
• Anaglyph
• Stereo3D
• Stereoscopic
• Red-Cyan
• 2D-to-3D
• conversion
• algorithm
• CPU
• GPU
• Dimensional
• VR
• Virtual
• Reality
• car
• driving
• Istanbul
• Maslak
• digital
• camera
• Canon
• FPS
• POV
• deep
• depth
• stereo-eye
• left
• eye
• right
• red
• and
• blue