Unreal Engine 5 Documentation has been moved to the Epic Developer Community

Contributors: Oculus

Monoscopic Far Field Rendering (or, more accurately, Hybrid Monoscopic Rendering) is an advanced rendering feature for VR Mobile applications being built in Unreal Engine 4 (UE4). Generally, Monoscopic Far Field Rendering improves the performance of levels containing a significant amount of distant content. The reason that Monoscopic Far Field Rendering improves the performance of levels having a lot of distant content is that distant objects are rendered only once for both eyes. At Epic, we've observed that rendering distant objects this way yields significant performance improvements for VR Mobile applications.

With the release of UE4, Version 4.16, Monoscopic Far Field Rendering is enabled with Multiview on GearVR devices, thereby ensuring that the stereoscopic near field rendering portion of your scene is optimized.

Enabling Far Field Monoscopic Rendering

When you create a new project, Monoscopic Far Field Rendering is disabled by default. To access the Monoscopic Far Field Rendering setting, open the Project Settings menu from Unreal Editor, and select Rendering from the side bar. You should be able to see the VR dropdown menu from inside of the Project-Description menu.

To enable Monoscopic Far Field Rendering, select the checkbox next to the Monoscopic Far Field rendering label.

After enabling the Monoscopic Far Field setting, you'll be prompted to restart UE4. When you're ready, go ahead and restart UE4 to apply the Monoscopic Far Field rendering settings to your project.

Adding a Monoscopic Camera

When Monoscopic Far Field Rendering is disabled, your project has two existing stereoscopic cameras, one for each eye.

When you enable Monoscopic Far Field Rendering, UE4 adds a third camera between the existing stereoscopic cameras. The new camera is a Monoscopic camera, which enables the hybrid monoscopic (and stereoscopic) rendering pipeline.

Stereoscopic Camera Frusta Symmetries

Depending on your device's platform, plane geometry of the stereoscopic camera frusta may vary. For GearVR devices, the stereoscopic and monoscopic camera frusta are symmetric; however, for Rift devices, the stereoscopic and monoscopic camera frusta are asymmetric. Please refer to the following diagrams to see how the camera frusta differ between GearVR and Rift devices.

As shown on the left, GearVR stereoscopic cameras have symmetric frusta, which means that the monoscopic camera shares the same projection matrix as the stereoscopic cameras. On Rift devices, however, the stereoscopic cameras are asymmetric (the inward angle towards the nose is less than the outward angle), which means that the monoscopic camera's frustum is the union of both the left and right frusta, which makes the monoscopic render target marginally bigger than the stereoscopic render targets.

Split Plane Properties

When UE4 constructs a monoscopic camera, the renderer adds a split (culling) plane.

The following properties apply to the split plane being added to your project:

• The split plane is parallel to the project's z-axis.

• The split plane's default Mono Culling Distance is set to 750.0 centimeters (7.5 meters).

• Objects on the far side of the plane are rendered with the monoscopic camera.

• Objects straddling the culling plane are rendered with both camera types (monoscopic and stereoscopic).

In the following image, the shaded (green) area represents the far side of the split plane, objects in this area are rendered in the monoscopic camera's depth buffer. Objects falling on the near side of the split plane are rendered in the stereoscopic camera's depth buffer.

The following image comparison demonstrates the split plane in action:

Performance Consideration

Thus far, you've learned that UE4 utilizes a depth buffer rendering technique to separate content being drawn by either stereoscopic or monoscopic cameras. From a high level, the UE4 mobile forward renderer was modified to use depth buffers, determining if stereoscopic pixels have less depth than monoscopic pixels. When compared to an object-based rendering technique, utilizing a pixel-based split (culling) plane can yield performance advantages, however, when objects go through the split plane, they have to be drawn monoscopically and stereoscopically (even if the object's pixels are far enough to be drawn monoscopically). These extra draw calls typically generate unnecessary bandwidth, increasing GPU and CPU vertex costs.

A good example of this performance hit can be seen in the following image: where the cliff's bounding sphere collides with the split plane, thereby causing the stereoscopic near field cameras to render the cliff (even though the cliff's pixels should only be rendered by the monoscopic far-field camera).

Optimizing Performance

To assess your project's performance, refer to the statistical output being generated by the Rendering Hardware Interface (RHI) . To view statistics being generated by RHI, perform the following steps:

1. From the Editor, click the Viewport Options button.

   Click for full image.

2. Hover your mouse cursor over the Stat option to open the Statistics menu.

   Click for full image.

3. Navigate to Stat > Advanced .

   Click for full image.

4. Finally, select the RHI option to display your project's RHI Statistics .

   Click for full image.

End Result

At this point, you should see your project's RHI Statistics .

If you determine that your project's RHI profile is unacceptable, there are a few strategies you can take to optimize your project's performance, including:

• Updating the Split Plane Distance . There may be scenes where you have to modify the split plane distance to accommodate large (or small) rendering areas.

• Updating Monoscopic Rendering Modes . Changing Monoscopic Rendering Modes allows you to determine if certain objects need to be forced to render in mono.

• Forcing Far Objects to Render in Mono . Inevitably, there will be use cases to force certain objects (such as skyboxes or background meshes) to render in mono.

Updating the Split Plane Distance

If you'd like to update the split plane's distance, open your project's World Settings in the Editor, locate the VR menu, and update the Mono Culling Distance to your desired distance.

Updating Monoscopic Rendering Modes

After enabling monoscopic rendering, you may want to update your project's monoscopic rendering mode. Rendering a scene in different modes can help you debug potential performance issues with your project. To update your application's monoscopic rendering mode, go ahead and enter vr.MonoscopicFarFieldMode [0-4] into the Editor's Command Console.

By default, vr.MonoscopicFarFieldMode is set to 1 , however, if you want to change the default value, please refer to the following table.

Forcing Far Objects to Render in Mono

Let's say that you're working on the following scene.

Now, you input the vr.MonoscopicFarFieldMode 3 console command, enabling stereoscopic far field rendering while also disabling the split plane.

If (after observing your scene) you decide that want to force the cliff's mesh to render in Monoscopic Far Field mode, open your mesh using the Blueprint Editor and select the Render in Mono option under the Rendering menu in the Details panel.

Stereopsis

The inspiration behind Hybrid Monoscopic Rendering is the ocular process of stereopsis. According to the American Academy of Ophthalmology , stereopsis is synonymous with depth perception. When you look at an object, each eye looks at the object from a slightly different angle, with each eye sending images back to the brain, where the images are "composited" into a single image. The different images being sent back from each eye form a binocular disparity, which the brain uses to extract depth information.

Leveraging Binocular Disparity

Leveraging the fact that the brain has a hard time extracting depth information from far away objects, the UE4 mobile forward renderer was modified to perform the following (in order):

1. Render opaque objects with stereoscopic cameras.

2. Shift and combine the output to create a monoscopic occlusion mask, pre-populating the monoscopic depth buffer.

3. Render opaque objects with the monoscopic camera.

4. Composite the monoscopic camera's result into the stereo depth buffers.

5. Render all transparent content and perform all post-processing with stereoscopic cameras.