All that is incorrect. The thread title is wrong. Shadow Fall is not using any kind of interlacing, vertical or horizontal. Nor is it using horizontal upscaling. This is a new technique, and can't be described using terms applied to previous methods.
Interlacing and upscaling have particular meanings, and refer to methods of using pixels of an undersized video stream to fill in display pixels on the same or subsequent frames. In distinction, the Shadow Fall method cannot be used on a video stream. It requires data only used within 3D rendering engines, and therefore couldn't (for example) make existing Youtube videos look better at higher resolutions.
Because it uses this extra data, it gives much better results than interlacing or upscaling--so much better, that in certain cases pixel counting can't detect the native rendering resolution. (In other cases it can.) The drawback is artifacts--manifested as blurred textures or motion ghosts--when the approximation fails. Despite this, the final image still has much higher IQ than an equivalent upscale or interlace.