Rendering and Compositing Wireframes in Houdini 16

This Houdini 16 lesson combines a clean viewport wireframe with a shaded Mantra render, then animates a wipe between the shaded image and its wireframe overlay. The wireframe comes from a flipbook capture, so matching the camera crop and removing viewport decorations are essential parts of the workflow.

The example contains a sphere and a ground plane. It renders single frames for the two source images; the final movement is a compositing transition between those stills. The older Image Network and compositing nodes are described in their recorded context, without assuming the same interface in a newer release.

Choose a wireframe display that communicates the surface

From 00:30, create a camera from the current view and a Mantra output node. Select that camera when preparing the viewport capture. Moving away from the camera changes the viewport to a free view, so the existence of a camera alone does not guarantee that the flipbook is using it.

Around 02:02, compare the viewport display modes. A wireframe bounding box shows only an object’s enclosing box, which does not explain the sphere’s topology. Ordinary wireframe shows front and back edges together and becomes visually crowded. Wireframe Ghost reduces the prominence of the rear structure.

The instructor chooses Hidden Line Invisible so that edges on the far side do not show through the front surface. That is an intentional presentation choice: the overlay should describe the visible surface without the extra clutter of the rear edges. Use the display-mode comparison to decide what information the topology presentation needs to convey.

Capture a clean, camera-aligned wireframe image

At 03:04, open Flipbook with New Settings. Set the range to frame 1 through frame 1 for this still-image example, and save the output as PNG so the wireframe image can retain an alpha channel.

The first capture includes the grid and other viewport decorations. Return to the flipbook settings and enable Beauty Pass Only to exclude that interface information. This option cleans up the viewport capture; it does not turn the wireframe into the separate shaded Mantra beauty render made later.

The cleaned image appears black when viewed against a black background because its lines are black. Inspect the alpha or place it over another image before concluding that the capture failed. The useful result is the isolated wireframe, not an opaque black rectangle covering the eventual render.

From 05:01, set the camera resolution to HDTV 1080, or 1920 by 1080 in the demonstration. Match the flipbook output size to the camera, then enable Crop Out View Mask Overlay. Without that crop, the flipbook can include the surrounding viewport region and scale the framing differently from the camera render.

This is the main alignment check: both images must represent the same camera view and frame at the same output dimensions. Matching pixel dimensions without matching the crop is insufficient. The instructor also suggests increasing antialiasing quality for cleaner wireframe edges. Motion blur and depth effects are not needed for this still capture.

Once these settings are saved, the normal flipbook button reuses them. If the capture unexpectedly includes decorations or mismatched framing later, reopen the settings rather than assuming they remained at their defaults.

Render the shaded image from the same camera

Around 08:01, return to Mantra and confirm the camera selection. Add an environment light and render to MPlay. The initial light tint is rejected; the instructor desaturates it and chooses a neutral midtone so the shaded image is easier to read.

Save the finished frame as a JPEG, using the example name beauty.jpg and quality 100. The wireframe remains the PNG with alpha. The tutorial uses these two files as inputs to the compositor; it does not establish a complete image-sequence naming or output pipeline.

Check the saved images themselves before compositing. The shaded image should contain the intended camera composition, and the wireframe should have the matching crop. A compositor cannot correct a camera mismatch merely by placing one image over the other.

Layer the wireframe over the beauty pass

From 12:05, create an Image Network with two File nodes. Replace the default sample image in one with the beauty render and load the wireframe PNG in the other. Use the Composite View to inspect the resulting images.

Add an Atop node. Connect the beauty image to its background input and the wireframe to its foreground input, using the input labels to confirm the order. The demonstrated result is the black topology over the shaded sphere and ground.

Inspect registration at recognizable contours. Lines that sit off the model usually point back to a capture or camera mismatch, whereas a correctly aligned but unreadable overlay may be a line contrast or display-mode choice. Keep those two problems separate when troubleshooting.

The Atop weights can fade the contributions. The lesson mentions that as one possible transition, then moves to a spatial wipe so the change travels across the image.

Animate a wipe between the two presentations

At 14:01, add a Wipe node with the plain beauty image and the completed wireframe composite as its two image sources. This distinction is important: one input already contains both beauty and wireframe, allowing the transition to reveal topology without discarding the underlying shaded appearance.

Change the operation to a vertical wipe and move its amount to inspect the direction. The instructor tests the divider’s line width and color, then sets the width to zero to remove the visible border.

Keyframe the wipe amount so the wireframe comes on over roughly one second, holds for roughly another second, and then comes off. The transcript gives the timing intention but not a complete list of frame numbers or a verified scene frame rate. Use the scene’s timeline to express those durations rather than assuming a particular frames-per-second value.

Play the compositor result from the beginning. The shaded image should remain registered while the wireframe is revealed and removed. This completes the demonstrated presentation effect. Extending it to moving rendered footage also requires matching the two source sequences frame by frame; that sequence workflow is mentioned as an application, but is not built in this still-image lesson.