Displace

Purpose. Displace neither moves 3D vertices nor composites one image over another. It resamples a 2D image according to a separate control map, making it useful for heat haze, water, glass, waves, and procedural warps.

Three inputs, three responsibilities. Orange Input carries the image to distort; green Foreground Image provides the controlling map; blue Effect Mask optionally restricts where the effect contributes. The mask and displacement map are not interchangeable.

A real map is required. Input and Foreground Image both need appropriate image streams. Feeding Fast Noise only to Effect Mask does not provide a displacement field. Connect the map first, then mask the processed result if needed.

Choosing a map. Fast Noise, soft gradients, rendered channels, and prepared Red/Green maps can all drive displacement. Broad, smooth transitions usually produce fluid refraction, while hard detail causes sharp kinks and visible image jumps.

Typical flow. MediaIn → Displace.Input; FastNoise → Blur → Displace.Foreground Image; Displace.Output → MediaOut. Add Ellipse → Effect Mask when you need a localized heat haze or pane of glass.

Type: Radial. The warp moves samples inward or outward relative to Center using a single Refraction Channel / Offset / Refraction Strength set. This suits lenses, bubbles, shockwaves, and spherical-looking localized refraction.

Center in Radial. X/Y values place the center of the radial effect in the image. A mismatch between the map feature and Center can make an otherwise correct warp appear offset. Center is not a universal pivot for X/Y mode.

Radial Refraction Channel. Choose Red, Green, Blue, Alpha, or Luma according to where the useful map data lives. The control routes information from Foreground Image; it does not color-correct the original Input.

Radial Offset. Offset appears beside Refraction Strength in the captured Inspector. In practice it changes the bias or neutral response of the map, but the manual does not specify the internal normalization equation; verify it with a test grid in your target Resolve version.

Radial Refraction Strength. This sets the intensity of radial refraction. Increase it gradually after Center, Channel, and Offset have been verified; high values may reveal out-of-frame samples and transparency artifacts.

Type: X/Y. Instead of one radial response, there are two independent axes. X Refraction Channel, X Offset, and X Refraction control horizontal movement; the corresponding Y controls drive vertical movement.

Independent channels. One image can store the horizontal field in Red and the vertical field in Green. Selecting the same Luma channel for both axes is also valid for simple heat haze, but then the two directions no longer have independent source information.

Debugging X/Y. Set Y Refraction to zero and validate X, then reverse the process. If the wrong detail moves when testing X, fix X Refraction Channel or the map instead of hiding the problem with a larger displacement amount.

Per-axis Offset. X Offset and Y Offset are adjusted separately. A value such as −0.5 is sometimes used to center neutral mid-gray maps, but it is not universal; inspect the map range and actual response in the target build.

Light Power. This controls simulated map-contour highlights and shadows, not a second pixel-displacement engine. Build the warp at Light Power = 0 before introducing shading.

Light Angle. This changes the direction of the simulated light along map contours. It does not rotate the X/Y displacement field; compare two angles with an otherwise fixed warp to see the distinction.

Spread. This broadens and softens the lighting/refraction response around map ridges. A high value can make abrupt contours less pronounced, but will not fix incorrect X/Y channels or excessive displacement.

Light Channel. Independently selects Color, Red, Green, Blue, Alpha, or Luma as the source for simulated shading. For example, you can warp using Luma and shade using Alpha from a prepared control image.

Effect Mask acts after processing. It gates the contribution of the finished warp and blends the distorted image with the original. It does not replace Foreground Image or define X/Y direction; a soft mask edge can hide the boundary of a localized effect.

Preparing the map. Adjust its scale, contrast, and blur in a separate upstream branch. Reducing the tiny details of a Fast Noise map often improves natural heat haze more effectively than simply lowering Refraction Strength.

Alpha and frame bounds. Displace resamples RGB together with alpha, so check transparency around text and keyed elements. Large warps may request pixels beyond the source frame; leave an image margin upstream and crop deliberately afterward.

Animation. For organic refraction, animate Seethe/Evolution or a Transform of the control map. Large Strength oscillations without an evolving map can feel mechanical, while high-frequency fast-moving noise often boils or flickers.

Validating the effect. Compare source, map, and output in separate Viewers and perform bypass A/B. Review at 100%, inspect RGB and alpha, check frame edges, and play several frames: a still does not expose temporal instability.

Evidence boundary. The captured Inspector shows Radial, the Refraction Channel menu, and the Light Channel menu; conditional X/Y settings are supported by the manual and existing control signatures. Exact Offset math and extreme edge sampling still require a target-build Resolve test; vertex-based deformation belongs to Displace3D.