
The wall can be part of the lighting, not all of it
An LED wall can give an actor contextual color, broad ambience, moving reflections, and a background visible to camera. Those are valuable functions. They are not automatically the same as a controllable key, edge, eye light, or motivating practical. When the crew expects one display to do every job, the shot becomes difficult to expose, difficult to shape, and vulnerable to a change in background content.
Epic’s ICVFX best-practices page says that real-time ray tracing can be expensive and points to baked lighting with volumetric light maps and reflection probes as ways to obtain stable image quality in some cases. That is an engine-side performance observation, not a prescription for on-set lighting. It does reinforce the broader caution: a good reflection or lighting result has both a scene/render cost and a photographed-world cost.
Split the lighting conversation into three passes
First test the wall as background only: exposure, color, texture, frame rate behavior, and whether it creates unwanted spill. Second add the planned key, fill, negative fill, and practicals; judge the face and foreground object, not only the wall. Third introduce the reflective or transmissive objects—chrome, glass, polished floor, wet fabric, eyes, lacquer—and examine the recorded camera result during movement.
| Element | What it should do | What to test |
|---|---|---|
| LED wall | World context and intended reflections | Exposure, spill, pattern, and brightness changes |
| Key/fill | Model the subject | Face shape with wall content changing |
| Practical | Motivate a local source | Whether its source is visible and controllable |
| Reflective prop | Carry believable world information | What it reveals at each camera angle |
Make reflection a framing issue
Epic’s rendering-settings reference distinguishes Lumen, screen-space, and other reflection approaches, and notes memory/performance consequences for very high reflection-capture resolution. In production terms, do not call a virtual reflection “correct” until the physical object has been seen through the intended lens and move. A screen-space limitation and a chrome prop showing an unwanted crew reflection are different defects with different owners.
Atlas proposed test card
For each risky prop or surface, attach a card to the shot: camera position, lens, wall content state, physical light state, desired reflection, unwanted reflection, and allowed mitigation. The mitigation may be re-angle, flag, change content, use a practical reflection card, alter surface treatment, or defer a plate. Write “post option” only when someone has defined what will be captured for post.
Illustrative: a glossy black helmet must reflect a passing city. The wall supplies a moving city pass, a separate soft key shapes the performer, and a controlled practical adds the required eye catchlight. The test finds that the camera body appears in the helmet on one move. Re-angling the helmet solves it; increasing wall brightness would not.
- Test the wall alone, then conventional lighting, then reflective foreground objects.
- Judge recorded camera output during the actual move.
- Separate virtual-render issues from physical unwanted reflections.
- Document what the wall is responsible for and what another light must provide.
Sources & evidence
Primary documentation establishes tool behavior or a reported production. Checklists and illustrative examples are Atlas editorial guidance, not results of independent testing. Match documentation to your installed version.
Discusses the performance cost of real-time ray tracing and alternatives including baked lighting and reflection probes.
Source published: Not established · Retrieved 19 September 2026
Documents distinct reflection methods and cautions that very high reflection-capture resolution can affect memory and performance.
Source published: Not established · Retrieved 19 September 2026