Supersampling renders the image at a higher resolution and then down-samples to fit the screen you're viewing (in this case, a VR headset). It blurs the original high-res image a little, but you end up with a better visual result than if you simply rendered once - especially for edges and high-contrast blocks of colour butting up against dark backgrounds (or thereverse) where jaggies etc would be highly noticeable.
TorTorden - you're partly right. Supersampling an image
does result in 'more' information being displayed - even when downsampling aferwards, as seen in VR. A higher initial resolution image is rendered - it has intrinsically more pixels and thus contains more visual detail. When you downsample that image to the Rift/Vive resolution, the resulting pixels are fewer, yes, but each retains some of the characteristics of the original image - each pixel is blended from the source. Its not simply knocking lines of pixels out to drop the resolution (an early technique).
At higher rates of supersampling, (1.5x+) you quickly run out of additional detail to render (ED game assets are not that highly detailed), and the ever-smaller details captured by high resolutions would be lost in the down-sample - even though they're still contributing in some small way, you won't notice the added effect. hence many don't see much impact of SS past 1.6... ED's art assets and scenes don't contain much information that can be improved past that point - it just adds a bit of smoothing, is all.
Supersampling really helps in ED because Elite generates a lot of very high-contrast scenes. Like black backgrounds with white stars and bold uniform colours, harsh highlights etc. Its easier to see all the artifacts, especially at the Rift/Vive comparatively low resolution. Its of less use in moderately- to well - lit games with more visual detail ie many FPS games where there is more detail for your eyes to take in, and less opportunity for artifacts to be noticed.
The AA isn't poor exactly - its just ED is a very-close-to-the-extreme end-member scenario for rendering. Anti-aliasing was/is mostly intended to do the same job as brute-force supersampling, but pick the bits of the image that need it and discard the rest etc to save performance. SS and AA achieve amny of the same things, but are different techniques. But for SS or AA, ED's very high contrast scenes, fast movement of small-ish objects at distance, and those objects being lit brightly against dark backgrounds etc mean its a worst-case scenario for the renderer - Frontier obviously have some render code/shaders in there to try to help, which is why the in-game supersampling looks a bit different to the same setting used in HMD Quality, or any of the AA settings, which struggle even though they're very effective in other games).
You're absolutely right in that learning to 'sweep' your view across a scene really helps form a more 'solid' and believable image in your mind. In a way, you're viewing more 'pixels across a wider range of view and bringing in more detail - your own eyes do this by constantly moving (saccades). In VR, I find you do need to move your head around a bit
The trade-off is the extra graphics horsepower needed to do the supersampling (in a short enough time to keep the VR frame rate as high as possible).
The HMD-Quality setting and the in-game ED Supersampling setting achieve similar output through different pathways. In my opinion, the HMD setting is superior, and costs less performance. The Debug Tool/HMD setting are the same thing - a low-level supersampling pass that has no additional shader or code to get in the way/slow things down etc.
Edit: Whoops sorry for the essay... oh well that's lunchtime gone!
![[HJ]-RedRaven](https://forums-cdn.zaonce.net/data/avatars/m/67/67389.jpg?1553527112){kind=avatar}
