Spherical Camera Node

A regular camera (Camera 3D) only sees a limited cone of the scene (the frustum). The Spherical Camera sees everything around it - 360° horizontally and 180° vertically - and flattens this view into a 2D image in one of 5 available formats (equirectangular, vertical cube cross, horizontal cube cross, vertical strip, horizontal strip).

The output size depends on the Image Width parameter of the Renderer 3D - each cube face is computed at that size, then the faces are assembled according to the chosen layout.

• Effects Library: Tools > 3D > Spherical Camera
• Quick add: right-click in the Node Editor → Add Tool > 3D > Spherical Camera

• Image Input (orange) - optional. Accepts images in a spherical layout (LatLong 2:1, Horizontal/Vertical Cross, Horizontal/Vertical Strip) if you also want to use the camera as a skybox.
• Stereo Input (green) - optional. To connect a second camera (right eye) for VR stereo rendering.

No input is required - the Spherical Camera works as a pure camera with no image source.

Typical pipeline for a 360° render:

3D objects ──┐
Spherical Camera ──┼→ Merge 3D → Renderer 3D → Saver (360 video)
Lights     ──┘

In the Renderer 3D, select the Spherical Camera in the Camera menu. The render then produces an image in the chosen layout.

Recommended resolution for VR: 4096×2048 (4K equirectangular) minimum. 8K (7680×3840) if you are targeting high-resolution headsets.

LatLong

Equirectangular 2:1 (e.g. 4096×2048). A single image that "unwraps" the full sphere. The standard format for YouTube 360, Facebook 360 and most VR headsets. The most common.

VCross (Vertical Cross)

The six cube faces arranged in a vertical cross (3:4 ratio). Front in the center, top above, bottom below, left/right/back along the column. The classic format for a video game skybox.

HCross (Horizontal Cross)

Same principle but as a horizontal cross (4:3 ratio). Compatible with more external tools (some 3D engines).

VStrip (Vertical Strip)

Six faces stacked in a column (1:6 ratio). Order: Left, Right, Up, Down, Back, Front. Useful for archiving or specific workflows.

HStrip (Horizontal Strip)

Six faces in a row (6:1 ratio). Same order as VStrip.

Rule: LatLong for a standard VR export, VCross/HCross for use as a cubemap texture in an external engine.

Most parameters are identical to those of the standard Camera 3D:

• Near / Far Clip - culling of geometry that is too close or too far.
• Adaptively Adjust Near/Far Clip - automatic adjustment based on the extent of the scene.
• Viewing Volume Size - orthographic mode only.
• Plane of Focus - for DOF.
• Control Visibility - shows or hides the indicators in the viewer (frustum, view vector, clipping planes).

The Spherical Camera supports stereo rendering for VR. Main parameters:

Stereo Method

Three modes:

• Toe In - converging cameras. Can create uncomfortable vertical parallax in VR.
• Off Axis (default) - the recommended method for viewing comfort in VR, with no vertical parallax.
• Parallel - strictly parallel cameras.

Eye Separation

Distance between the two cameras (left/right eye). Human standard: ≈ 6.5 cm, to adjust to the scale of your Fusion scene.

Convergence Distance

Convergence distance of the two cameras. Used in Toe In mode.

For VR, always keep Off Axis - people wearing a headset will notice vertical parallax problems (headaches, motion sickness).

The setup is almost identical to the standard Camera 3D, but with two fundamental differences:

Camera 3D

Limited conical view (angle of view). Produces a flat rectangular image. The standard for any regular render.

Spherical Camera

Omnidirectional view (360°). Produces a spherical image flattened into the chosen layout. Specialized for VR / skybox / reflection map.

• 360° video for YouTube/Facebook - LatLong 4K minimum, encoded with 360 metadata (through the Spatial Media Metadata Injector) so the player recognizes the format.
• VR content for headsets - Off Axis stereo + 8K LatLong for an Oculus / Quest / Vive render.
• Skybox for a 3D engine - VCross or HCross, rendered at 1024×1024 per face (4096×3072 total for a 4:3 vcross). Importable into Unity / Unreal / Godot.
• Reflection / environment map - a 360° capture of the scene from a precise point, then used as a reflective texture on another object.
• 360° scene preview - see a Fusion scene from every angle without moving the camera.

• Resolution = multiple of 2 - for a LatLong, use a strict 2:1 ratio (e.g. 4096×2048). Otherwise 360 players may misinterpret the image.
• Off Axis by default in stereo - the other modes create vertical parallax = guaranteed motion sickness for viewers.
• Render times are much longer than with a regular camera - the scene is rendered 6 times (once per cube face) then assembled. Plan some margin.
• Exporting in stereo: for stereo VR, the final file is usually top-bottom (left eye on top, right eye at the bottom) or side-by-side. You often need to composite these two views in 2D after the Fusion render.
• 360 metadata - a LatLong render isn't automatically recognized as 360° by YouTube. Inject the metadata with the Spatial Media Metadata Injector (a free Google tool) before uploading.
• For use as a texture (skybox, env map), switch to Image Input (if you already have an existing LatLong) instead of re-rendering.