Toshiba’s twin-lens camera module for smartphones, tablets allows 3D imaging and post-capture focus simulation


posted Friday, September 27, 2013 at 3:59 PM EST

It seems to be the week for mobile 3D modeling! Just two days ago, we reported on a Kinect-like addon device for Apple's iPad tablet. Now, Toshiba has unveiled a twin-lensed camera module for smartphones and tablets which does much the same thing, albeit in a different way. It also allows some neat post-capture effects such as depth-of-field and point-of-focus adjustment.

The Occipital Structure Sensor we mentioned earlier this week uses an infrared laser to project a grid over your subject, and from this can calculate a depth map which corresponds to a still image or video stream captured by the device. Toshiba, by contrast, records the scene using two separate, side-by-side 5-megapixel cameras, then calculates a depth map by comparing differences between the two photos. The end result is the same, though: You should be able to record photos with 3D depth information, and -- with the appropriate software -- create 3D models of your subjects. And from the depth map, Toshiba can simulate a narrow depth-of-field effect, placing the focus point wherever you want it. (The company also suggests that the depth map can be used to identify and remove objects from a photo altogether.)

Unlike the Structure Sensor, which is intended to remain an addon device, Toshiba's TCM9518MD camera module is designed to be built into your next smart gadget. (Of course, there's nothing to stop somebody creating an accessory based around it, either.) It has dimensions of just 18 x 12 x 4.65mm, and twin 1/4-inch type, 5-megapixel sensors with 1.4µm pixels, providing 10-bit raw data. Two four-element f/2.4 lenses have a fixed 61.8 degree horizontal field of view. The lenses have a fixed-focus design, with all focus effects emulated in software. That allows a thinner lens profile. Lens shading correction is built-in.


No resolution was specified for the depth map of the Structure Sensor, but Toshiba does state its resolution, and it's fairly low. You can generate a VGA depth map to accompany your five megapixel images, which means you have one pixel of depth map information for every 16 pixels (4x4 array) of photo information. In other words, your depth map has 1/4 the linear resolution of the still image.

The twin-lens design brings an interesting benefit in another area, though. According to Toshiba, by pairing info from both cameras it can create a 13-megapixel output image from the two 5-megapixel source images. It's likely to work better for distant subjects than those nearer the camera, though, because both cameras will have more strongly-differing viewpoints of the subject the closer it gets.

And as well as shooting single 5-megapixel and twin-lens interpolated 13-megapixel stills, the Toshiba TCM9518MD camera module can also output 5-megapixel stills from both lenses simultaneously, and shoot 1080p30, 720p60, or 480p120 video. It can also output full-res sensor data at 24 frames per second.

(via PMA Newsline)