Fusion 3D

A classic Fusion 3D branch contains objects in a coordinate space. Geometry, 3D text, image planes, cameras, lights and materials are assembled into a scene rather than layered directly as 2D pixels.

That scene can be transformed and viewed from different cameras without re-rasterizing every object at each step. The conversion to a normal image happens only when Renderer3D evaluates the scene from a camera.

Merge3D combines scene elements that coexist in 3D space. It does not place one flat image over another the way the regular Merge does. Several objects, cameras and lights can feed a Merge3D and remain distinct 3D entities.

A normal Merge expects images. Therefore a Merge3D output normally goes to Renderer3D first. The Renderer3D output is then a regular image that can be connected to a 2D Merge.

Geometry defines form: Text3D, Shape3D, ImagePlane3D and imported meshes are common examples. Transform3D positions scene elements. Camera3D determines the viewpoint. Lights and materials control how surfaces respond visually.

Renderer3D resolves those scene relationships into pixels. Its settings affect quality, lighting, shadows, transparency and other render behavior, so it is both a technical boundary and a creative part of the scene.

Use 3D when perspective, parallax, camera movement, volumetric positioning or lighting relationships matter. 3D titles, screen walls, product cards, layered image planes and camera-driven motion are common production cases.

If the visual can be built as two flat layers with a simple scale or position animation, classic 3D may only add complexity. Choose it when the spatial relationship is part of the design.

A clean workflow often ends the 3D branch at Renderer3D. From there the rendered image is merged over live footage and matched with ordinary Blur, Glow, color, matte and grain tools.

Keeping that boundary clear prevents the common mistake of trying to feed a live 3D scene into nodes that only understand pixels.

A typical title setup starts with Text3D or another 3D object, combines scene elements through Merge3D, adds Camera3D and one or more lights, and then uses Renderer3D to turn the scene into an image.

Only after Renderer3D should the title be treated like a normal layer. A regular Merge can place it over MediaIn, and ordinary 2D tools such as Glow, Blur, Color Corrector or masks can finish the integration.

This example shows the essential boundary: Merge3D assembles objects that still live in a scene, while Merge assembles rendered images. Keeping those jobs separate prevents one of the most common architectural mistakes in Fusion.

A texture feeds a compatible material input; the material feeds the geometry’s Material input or ReplaceMaterial3D. The shaded object, Camera3D and SpotLight enter Merge3D through independent branches, and the resulting scene feeds Renderer3D. Materials, lights and cameras are not sequential image effects after Merge3D.

For footage on a plane inside the scene, use MediaIn → ImagePlane3D → Merge3D. For final compositing, keep footage in 2D: MediaIn → Merge Background and Renderer3D → Merge Foreground. Choose according to whether the plane must participate in camera/object relationships or simply form the final background.

First check the camera selected in Renderer3D: the Viewer’s free perspective is not necessarily the final render view. Then check object position, near/far clipping, scale and normal orientation. Adjust material only after those checks; increasing specular cannot reveal an object outside the camera frame.

Lighting and shadows depend on the selected renderer, its settings, the light and the object’s properties. Check Lighting/Shadows where those options are available, and confirm material support in Software/OpenGL. Judge antialiasing and motion blur in the final Renderer3D output rather than solely in the interactive 3D Viewer.