Facebook is rolling out a new technology that creates 3D models of human beings from their photographs. As per the report by Slashdot, the new technology named PIFuHD works on multi-level architecture based on a deep neural network that allows us to create highly-detailed 3D models via a photograph.
This technology can be used in automated image digitization in fields like virtual reality and medical imaging. Facebook 3D image conversion demo version is also available on Google Colab for the users to try.
What is PIFuHD?
PIFuHD stands for Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization. It involves an end-to-end trainable coarse-to-fine framework for implicit surface learning for high-resolution 3D clothed human reconstruction at 1K image resolution.
The technology first translates the image into the front and back perspective and then delivers the 3D figure in high resolution.
This technology is going to be one of the most revolutionary achievements of 2020 as it can change the way we look at virtual reality and do wonders in the fields of medical imaging as well.
Limitations of Facebook 3D conversion
As the technology is in its initial phase, it still has a few limitations. Currently, the technology cannot accept images with resolutions greater than 1K.
However, a major reason behind this limitation is that accurate predictions also demand more context. The memory limitations in current hardware is also a challenge as of now.
As the PIFuHD technology will get to a more advanced level, changes will be made in the hardware accordingly.
You can find more details about this technology by Facebook at this link.