Google is partnering with HTC and Lenovo to produce standalone (no smartphone or tether) virtual reality headsets [theverge.com]. The headsets could cost around $500-$700, comparable to the Oculus Rift or HTC Vive. As they will have less computational/graphics power than flagship smartphones or desktops, Google has developed a rendering system [theverge.com] that they claim can compensate by decreasing the amount of polygons needed to render a scene (related video [youtube.com]):
Meanwhile, a rendering system called Seurat — named after the pointillist painter Georges Seurat — is supposed to offer image quality that rivals what you'd get on a high-end PC. Andrey Doronichev, Google's director of product management, describes Seurat as "computational magic." It takes a rendered three-dimensional scene and samples shots of it from many different angles. As seen [here [vox-cdn.com]], Seurat uses these images to assemble a facade that drastically reduces the number of polygons the headset needs to render, without a visible loss of quality.
Google can also use the same Daydream user interface it's been fine-tuning for the past year on phones. A software update [theverge.com] codenamed Euphrates will add the features you need for devices that users can't just pop apart and use as a phone, like a full-featured web browser and a dashboard for accessing settings and other non-VR parts of Android.
Google envisions VR and AR converging into mixed reality headsets [medium.com], building on the augmented reality technologies developed under Project Tango [wikipedia.org] as well as Daydream VR [wikipedia.org]:
To make VR more transporting, and AR more convincing and useful, everything behind these experiences must improve: displays, optics, tracking, input, GPUs, sensors, and more. As one benchmark, to achieve "retina" resolution in VR — that is, to give a person 20/20 vision across their full field of view — we'll need roughly 30 times more pixels than we have in today's displays. To make more refined forms of AR possible, smartphones will need more advanced sensing capabilities. Our devices will need to understand motion, space, and very precise location. We'll need precision not in meters, but in centimeters or even millimeters.
Both the Rift and Vive have 2160×1200 displays. Roughly 30 times more pixels would mean a resolution of around 11880×6600, or 16704×4698 (32:9 aspect ratio [roadtovr.com]).