Renderer 3D Node

Renderer 3D sits at the very end of the 3D chain. It takes the complete scene as input (usually coming from a Merge 3D) and produces a 2D image rendered from the point of view of a chosen camera.

Beyond the plain RGBA render, it can also output auxiliary channels (AOVs: Z-depth, normals, Object ID, …) used in post-production to isolate, relight or composite.

• Effects Library: Tools > 3D > Renderer 3D
• Quick add: right-click in the Node Editor → Add Tool > 3D > Renderer 3D
• Created automatically by the Camera Tracker after a solve, and offered as the output node when you create a 3D composition from the menu.

• SceneInput (orange) - required. Accepts a Merge 3D or any 3D node to render.
• EffectMask (blue) - optional. A 2D image that masks the output (same behavior as on other 2D nodes).

Renderer 3D plugs in at the very end of the 3D pipeline. Minimal workflow:

3D objects ──┐
Camera 3D  ──┼→ Merge 3D → Renderer 3D → (2D output, to Saver/2D Merge)
Lights     ──┘

The Renderer's output is a regular 2D image - it can be composited with other 2D elements, sent to a Color Corrector, saved through a Saver, and so on.

Camera menu

Lists every camera in the input scene. By default, it takes the first camera found, or a default perspective camera if none exists. Classic pitfall: if the menu is empty, the camera isn't connected to the Merge 3D feeding the Renderer.

Eye menu (default: Mono)

For stereoscopic rendering: Mono, Left, Right, Stacked. Outside stereo work, leave it on Mono.

Two fundamentally different engines. The choice depends on your priorities (speed vs consistency) and the kind of output you need.

OpenGL Renderer

• Uses the GPU → much faster.
• Supports supersampling and realistic Depth of Field.
• Ideal for interactive work and previews in the viewer.
• Limitation: it can't produce soft shadows.
• The result depends on the graphics card and driver - it can vary between machines.

Software Renderer

• Uses the CPU only → slower.
• Consistent cross-platform results (essential for network rendering on a farm).
• The only engine that handles soft shadows.
• Supports every Fusion lighting, texture and material feature.
• Produces two extra auxiliary channels (Coverage, BgColor) not available in OpenGL.

Practical rule: OpenGL for day-to-day work and previews, Software for the final production render (soft shadows, reproducibility).

Enable Lighting (default: on)

Enables lighting from the scene's lights. Important: if it is on and there is no light in the scene, everything renders black. Turn it on to see your lights, turn it off for an "emissive" render (pure unlit textures).

Enable Shadows (default: on)

Enables shadow rendering. Significant performance cost - turn it off for quick previews. Soft shadows only work with the Software Renderer.

The Renderer can output extra channels on top of RGBA, accessible downstream through a Channel Boolean or usable for relighting / keying. You find them in the Output tab.

Available on both engines

RGBA (required)

Standard color output. Can't be disabled.

Z

Per-pixel depth value (distance to the camera). For overlapping pixels, the value of the closest pixel is kept. Used for Z-depth compositing or post DOF.

Normal

X/Y/Z orientation of the surface normals, encoded as color ([-1, 1] range per axis). Essential for relighting in post.

TexCoord

UV coordinates mapped to the Red and Green channels. Used for re-texturing after the render.

Object ID (0 - 65534)

Numeric object identifier. Lets you extract a mask that isolates a specific object.

Material ID (0 - 65534)

Numeric material identifier. Same as Object ID but applied to materials (useful when several objects share a shader).

Software Renderer only

• Coverage - Z-buffer overlap information, as a percentage. Used for precise anti-aliasing in compositing.
• BgColor - color of the pixels behind partially covered areas.

Rule for AOV anti-aliasing: enable it only on WorldCoord and Z. On the other channels (Normal, Object ID, …), AA often degrades the data.

Anti-aliasing

Supersampling Rate (LowQ / HiQ)

Internal resolution multiplier. A value of 4 = render at 4× the resolution then downsample. For production-quality AA, 8×8 = 64 samples per pixel is enough in most cases.

Filter Types

Box, Bi-Linear, Bi-Cubic, Bi-Spline, Catmull-Rom, Gaussian, Mitchell, Lanczos, Sinc, Bessel. Catmull-Rom or Mitchell = good default quality / speed trade-offs.

Window Methods (Sinc/Bessel)

Hanning, Hamming, Blackman.

Lighting Mode

• Per-vertex - fast, but gives "blocky" lighting on barely subdivided geometry.
• Per-pixel - higher quality, more demanding. Recommended for the final render.

Transparency Sorting

• Z Buffer (fast) - fast but inaccurate on semi-transparent surfaces.
• Sorted (accurate) - correct sorting, essential for clean transparency rendering.
• Quick Mode - optimized for scenes with lots of particles.

Miscellaneous

• Shading Model - Smooth (standard) or Flat (flat-faced render).
• Wireframe - edge render, with optional AA.
• Texture Depth - texture bit depth (8/16/32 bits).

The Accumulation Effects section of the Renderer. It requires a Camera 3D with a correctly set Plane of Focus.

Enable Accumulation Effects

Master switch for accumulation effects.

Enable Depth of Field

Turns on depth blur.

Quality slider

Higher values → more realistic DOF in out-of-focus areas (but slower).

Amount slider

Low values → more of the image in focus (less blur). High values → stronger blur.

The Plane of Focus is set on the Camera 3D, not on the Renderer. For a "rack focus" effect, animate this value on the camera.

The Motion Blur controls are in the Common Controls tab (the tab shared by most nodes).

Watch out with particle scenes: the Motion Blur settings of pRender nodes must exactly match those of the Renderer 3D. Otherwise, the sub-frame renders conflict and produce wrong results.

• Final 3D scene render - Software Renderer + 8×8 supersampling + shadows on + DOF if needed.
• Quick preview - OpenGL Renderer without DOF or high supersampling.
• Producing AOV passes - enable Z, Normal and Object ID in the Output tab to get channels you can use in post.
• Multiple Renderers on the same scene - possible: plug several Renderers into the final Merge, each with different settings (e.g. one for the beauty, one for a dedicated Z pass).
• Extracting an object mask - enable Object ID, then in 2D use a Channel Boolean to extract the ID value that matches your target object.

• All black output? → first check Enable Lighting and that there is at least one light in the scene. Otherwise uncheck Enable Lighting for an emissive render.
• No output at all? → the camera selected in the Camera menu may not be in the scene connected to the Renderer.
• Square/jagged shadows → the shadow map is too small. Increase the Shadow Map Size on the shadow-casting lights (2048 or 4096).
• Different results between machines → this comes from OpenGL depending on the GPU driver. Switch to the Software Renderer for cross-platform consistency.
• Wrongly sized particles and locators → they are always rendered at their original size, regardless of supersampling. As a result, they look thinner than expected when supersampling is high.
• Several Renderers to save time - a preview Renderer (fast OpenGL) connected in parallel with a final Renderer (Software) lets you work quickly while keeping the quality for the export.
• For renders on a render farm, the Software Renderer is mandatory - farm machines don't necessarily have a GPU available, and OpenGL results vary.