I didn't even think this was news. I heard about it months ago.
Lucian Armasu :
the hardware requirements, which included not just specialized computer vision processors, such as Movidius’ Myriad VPU, as well as multiple wide-angle and depth cameras, seem to have been too high for regular smartphones. Google’s initial Tango-enabled tablet cost $1,024.
Few points about this:
■ That tablet was from 2014 and they quickly dropped the price to $512. But it was a developer kit, so still not worth reading too much into that pricing. It had not only a Myriad 1, but also used Nvidia's Tegra K1.
■ Both the Lenovo and ASUS phones had only a Qualcomm Snapdragon SoC, with no additional vision processor or GPU. The only special hardware they had was a second camera (not uncommon, in high-end phones) and a depth sensor.
As more smartphones gain wide-angle and telephoto cameras, as well as powerful computer vision/machine learning chips (such as the latest Pixel 2), it may be possible to upgrade ARCore to be at least as advanced as Tango was in a few years.
No, depth sensors and stereo cameras are complementary. If you need accurate scene geometry, then the depth sensor is required for cases with low texture, high scene complexity, or poor lighting. Conversely, stereo is required for longer ranges and brightly lit environments.
As Tango didn’t seem to have any future anyway neither with consumers nor with smartphone manufacturers, killing it seems to make sense.
It seemed to be paving the way for Hololens-style AR glasses. Given that we still don't have that (except for Microsoft's), then it's hard to justify.
If Google can show some usefulness for its ARCore AR platform as well as high adoption, it can just continue to improve that platform until smartphones become more capable of seeing the world as well or better than we we do in 3D.
I hope so. Last I checked, ARCore was functionally inferior to Tango, even on the same hardware. I developed some non-public apps for my employer that you cannot make with ARCore.