https://risa2000.github.io/vrdocs/docs/hmd_fov_calculation.html Zitat: Since it does not seem to look like a good idea, the question is, why would anyone want to do that? The answer is simple. They have no choice. Unfortunately, there is still a lot of VR applications, which were developed with an (unsupported) assumption that the stereo rendering always happened into parallel views and went a long way to optimize their code around this assumption. Even Nvidia’s first hardware support for the stereo view rendering only worked on parallel views. Before Pimax, the world was only parallel. Now, the demand for the non-parallel rendering already reached the developers and the hardware designers, but there is still a lot of work to do. And some applications are simply stuck with their implementation for ever. So to accommodate all those unfortunate cases, the parallel projection mode was introduced. In this mode the headset behaves as if it has the parallel view geometry and pretends to the VR subsystem that all is right and shiny, while doing some nasty tricks afterwards, to put those parallel images onto its non-parallel screens.
Interessanter Artikel, danke dafür Dort geht es aber um die Möglichkeit, dass man von Anfang an trapezförmige Viewports für geneigte Displays verwenden könnte, um Ressourcen zu sparen. Stattdessen werden, weil in Engines nicht anders möglich, sehr große rechteckige Viewports verwendet, von denen bestimmte Bereiche wieder geclippt werden müssen bzw. andere Ecken fehlen (das bezeichnen die mit "parallel projection mode"). Nicht erwähnt wird allerdings die Möglichkeit ein wide FOV aus 4 Viewports zu konstruieren (so wie es teilweise für die StarVR One geschieht), aber ist eh ne andere Angelegnheit...
Sorry, aber du verstehst das komplett falsch. Das aber hier auseinander zu klamüsern ist mir zu viel Aufwand, nix für ungut.
