QORPO Blog / Industry

Unreal shadow performance: isolate moving foliage before changing quality

Q
QORPO Studio18 November 2026 · 4 min read
A wind-bent sapling casting detailed leaf shadows across a graphite tiled floor
Concept illustration created for this article, not a gameplay screenshot.

When an Unreal scene with moving foliage has expensive Virtual Shadow Maps, compare a fixed camera and light with foliage motion enabled and disabled before lowering overall shadow quality. The purpose is to isolate invalidation from the rest of the scene and preserve the visual detail that players actually need.

Epic’s current documentation explains that cached shadow pages can be reused until relevant movement invalidates them. A useful diagnosis therefore changes one source of movement at a time. This is a proposed production method, not a benchmark or a claim about measured QORPO performance.

Why an attractive forest can be a difficult comparison

A forest scene contains several moving systems. The camera travels, leaves deform, characters cross the frame and lighting may advance through a day cycle. A single before-and-after frame cannot reveal which change caused a shadow-cost difference.

Epic’s Fortnite shadow engineering article, published on 26 January 2023, discusses animated trees and a moving sun as challenges for caching. That is a useful engineering example rather than a promise that another project will produce the same results.

Choose a small representative section of your own scene. It should include the foliage and lighting behaviour that matter to the game, with enough surrounding geometry to preserve the original visual context.

Establish a repeatable baseline

Record the engine version, build configuration, device, resolution and relevant rendering settings. Save a camera position that shows the problem clearly. Allow initial loading and shader work to settle before capturing the comparison.

Then record a short baseline with the intended foliage motion and light behaviour. Keep the same capture duration in subsequent runs. If the game uses a changing sun, note whether it moves during that interval.

Our effective console-variable guide helps avoid comparing an intended setting with a different active value. A settings note is useful only when it describes the state actually used for the capture.

Separate camera, light and foliage movement

Start with a stationary camera and a fixed light. Compare the normal foliage animation with the same scene under a temporary diagnostic setup that stops that animation. Preserve the rest of the scene.

Next, restore foliage motion and compare a fixed light with the intended light movement. Finally, add the normal camera route. These are separate experiments, each answering one question.

Stopping animation for diagnosis does not automatically make it a suitable shipping solution. A still forest may establish a cause while damaging the scene’s intended feeling. Keep the diagnostic result separate from the final content decision.

Read cache evidence alongside frame timing

Use the Virtual Shadow Map visualization and profiling tools documented for the engine version under test. Compare the visible invalidation pattern with the relevant GPU work, rather than using total frame rate as the only observation.

The Virtual Shadow Maps documentation describes caching, invalidation and debugging views. Exact labels and controls can differ across versions, so record the view used in the report.

If shadow work changes while total frame time barely moves, another workload may dominate the frame. Conversely, a faster overall run with no corresponding shadow change does not establish that the shadow adjustment caused the improvement.

A model example with an honest conclusion

Imagine a hypothetical grove where the camera and sun remain fixed. With normal leaf motion, the cache view changes repeatedly around the trees. Stopping leaf motion reduces that activity and the measured shadow work in repeated captures.

That observation supports investigating those moving casters. It does not establish a percentage improvement for the whole game, prove that every tree is expensive or justify disabling all foliage shadows.

The next candidate might adjust one representative asset or its distant presentation. Compare that candidate against the baseline, including the visual consequences. Keep the acceptable contact detail and movement cues in view when making the decision.

Review the picture after the numbers

Walk the normal player route and inspect the places where shadows convey shape, depth and movement. Look at leaves near the camera, character contact with the ground and transitions between distances.

A setting that hides an artifact from one saved camera may reveal it elsewhere. Our exposure comparison guide explains why lighting and viewing conditions should remain controlled when assessing visual changes.

Do not force a cache policy that leaves moving shadows visibly stale simply to improve one capture. Any engine override needs a correctness check for the actual animated content, followed by a performance comparison.

What belongs in the final report?

Save the baseline, the single variable changed, the capture conditions, repeated measurements and matched images. State which devices and routes remain untested. Use repeatable Unreal Insights bookmarks when a route needs consistent markers.

The useful outcome is a defensible next change, not a universal shadow-quality preset. Follow QORPO’s studio work for more practical development coverage grounded in explicit assumptions and reviewable evidence.

Sources