To improve the image quality of composition layers, enable composition layer filtering. Layer filtering methods support supersampling tuned to the Quest display-optics system.

Enabling supersampling (left) reduces flicker for high contrast edges (right) for layers that are undersampled by the compositor to match the display resolution.

Sharpening improves clarity of high contrast edges and can counteract blur when upsampling to meet display resolution.

By enabling auto filtering, the layer is rendered at peak visual quality while preventing adverse artifacts, like screen tears. In this mode, the runtime automatically applies the chosen supersampling and sharpening filters only when beneficial. Examples include:

If the layer does not require any filtering then opting into auto filtering will result in no operation with no performace overhead.

Auto filtering accounts for dynamic and complex interaction with the layers, like layer resizing and player movement in 3D space. Additionally, it considers the following parameters before filtering the layers:

The XR_FB_composition_layer_settings OpenXR extension provides settings to enable composition layer supersampling and sharpening. The XrCompositionLayerSettingsFB struct contains additional XrCompositionLayerSettingsFlagsFB flags that control the layer filtering features. There are currently four bits available that enable two forms of supersampling and sharpening.

For complete API reference as part of the OpenXR spec, read XR_FB_composition_layer_settings.

If both bits are set, the compositor defaults to the normal supersampling method.

If both bits are set, the compositor defaults to the normal sharpening method.

Note: In OpenXR, all these settings are suggestions and the compositor may enable or disable them, as it sees fit.

For details, see the XR_FB_composition_layer_settings specification and Super Resolution Blog.

Important: Composition layer filtering requires more GPU resources. This is especially true for the sharpening and supersampling algorithms as these utilize larger kernel footprints. If used directly for composition layers, using these algorithms increases the GPU resources that are essential affecting composition timing. This may lead to frame drops and negatively impact the VR experience. The bottom line is that when applying composition layer filtering, you should weigh the increased visual fidelity against the additional GPU resources required to offer the best VR user experience. Alternatively, you can use auto filtering mode to mitigate the risk of unnecessary filtering. Auto filtering will gracefully degrade the panel quality during instances of constrained GPU resources.