The most expensive CG element in a comp can fail because the lighting doesn’t quite match the plate. The most modest CG element can succeed because the lighting integration is invisible. The difference isn’t usually in the rendering quality — it’s in what reference the 3D team had to work from when they lit the asset.
Lighting reference is captured on set, in the small windows of time between takes, by crew members who often don’t realize how consequential what they’re capturing is. A 30-second HDRI grab is the difference between a render that integrates instantly and a render that takes three days of back-and-forth to feel right. A chrome ball held in the scene for ten seconds preserves information that the post team would otherwise have to reverse-engineer from analysis of the plate.
This post is the working 3D artist’s guide to lighting reference: what each capture provides, when it’s necessary, and what the fallback looks like when each level wasn’t shot.
The Reference Hierarchy
Lighting reference has a hierarchy. From most useful to least useful, the levels are:
Level 1: Spherical HDRI captured on set. A panoramic, high dynamic range image captured from the position where the CG element will be placed. The image preserves the exact light directions and intensities of the scene, in the format that 3D rendering software uses for image-based lighting (IBL).
Level 2: Chrome and gray ball reference. Two spheres held in the scene during a brief capture: chrome reflects the surrounding environment, gray provides neutral exposure and color reference. Together they preserve enough information to reconstruct lighting approximations.
Level 3: Plate analysis. The 3D team studies the plate itself for lighting cues — shadow directions, highlight positions, color temperature, fill characteristics. The analysis is approximate and time-consuming, but produces usable lighting models when nothing else exists.
Level 4: Educated guess. The 3D team estimates lighting based on the scene description, the time of day, the location, and general experience. This is the floor — what you fall back on when no on-set reference was captured at all.
Each level down the hierarchy increases the post-production time, decreases the integration accuracy, and increases the risk of the final shot looking subtly wrong. The cost of capturing at higher levels is small. The cost of not capturing them is large.
Level 1: HDRI Capture
A spherical HDRI is the gold standard for image-based lighting. The capture is a panoramic image of the scene’s lighting environment, with the full range from deepest shadow to brightest highlight preserved across the file.
How HDRI is used: in the 3D scene, the HDRI is mapped onto a sphere surrounding the CG element. The CG renderer treats the sphere as a light source, with the brightness and color of each point on the sphere acting as a directional light hitting the CG element from that direction. The result is CG that’s lit by the actual on-set environment — not an approximation of it.
What HDRI captures that nothing else does: subtle ambient lighting from many directions at once. A scene with a warm key light, a cool fill, a soft top light, and bounced light from a nearby colored wall has all four contributing to the CG integration when an HDRI is used. Without the HDRI, the 3D artist might capture two or three of those, but the subtle bounce from the wall is the kind of detail that’s easy to miss in plate analysis and easy to capture in HDRI.
Capture methods:
- Dedicated HDRI rig. A panoramic camera setup, often motorized, that captures the spherical environment in 30–60 seconds. Most expensive but most thorough.
- Bracketed chrome ball. A still camera photographs a chrome ball from multiple angles with multiple exposures; the images are combined into a spherical HDRI in post. Cheaper and faster than a dedicated rig.
- Smartphone spherical panorama. Modern smartphones can capture spherical panoramas in 30 seconds. The quality is approximate (limited dynamic range, sometimes stitching errors) but usable when nothing else is available.
When HDRI is essential: any shot with a hero CG element that has to integrate convincingly. Any shot where the lighting is unusual or hard to recreate procedurally (mixed-color sources, specific architectural light environments, golden-hour exteriors with specific atmospheric conditions).
When HDRI can be skipped: shots where the CG element is small in frame, doesn’t catch significant light, or is in motion fast enough that the lighting integration is forgiving. A flying object passing through a shot doesn’t need the same lighting precision as a hero character standing still.
Level 2: Chrome and Gray Balls
When HDRI isn’t possible, chrome and gray balls are the fallback. A chrome ball reflects the surrounding scene; photographing it captures, in compressed form, what the lighting environment looks like from that point. A gray ball lit by the scene’s lights gives a neutral reference for the scene’s color temperature and exposure.
How chrome ball reference is used: the 3D team examines the chrome ball image to identify the major light sources in the scene — where they are, what color they are, how bright they are relative to each other. They then reconstruct the lighting setup in 3D using virtual lights, calibrated against the chrome ball’s reflection.
How gray ball reference is used: the gray ball, photographed in the same lighting, provides a neutral reference for color and exposure. The 3D team uses the gray ball’s color values to white-balance the CG render and match the plate’s exposure level.
The combination of chrome and gray gives most of what HDRI gives, in a much faster capture. The difference is in the subtleties — bounced light from off-camera surfaces, complex light interactions, atmospheric scatter — that the chrome ball captures less precisely than HDRI.
Capture practice: hold both balls in the scene at the talent’s position, at roughly the talent’s height, lit by the same lights. Photograph for 5–10 seconds. Repeat at multiple positions in the scene if the lighting varies (close to camera, far from camera). The total capture takes 30–60 seconds.
When chrome and gray are sufficient: shots where the lighting is moderate complexity and the CG integration tolerates approximation. Most production VFX work fits this category.
When chrome and gray fall short: hero shots with subtle lighting that needs to feel exactly right. Shots with complex bounced light. Shots where the audience’s eye is going to study the integration.
Level 3: Plate Analysis
When neither HDRI nor chrome/gray reference exists, the 3D artist analyzes the plate itself for lighting cues. The analysis looks for:
Shadow directions. Where does the talent’s shadow fall? In what direction? How sharp or soft? This identifies the key light’s position and softness.
Highlight positions. Where are the brightest areas on talent and props? This confirms the key light’s direction and gives clues about its color.
Fill characteristics. How dark are the shadow areas? What color is the fill? This identifies the fill light’s role and color.
Reflections in eyes or shiny surfaces. Talents’ catchlights reveal the position of light sources directly. Shiny props can act as makeshift chrome balls.
Color temperature across the frame. The plate’s neutral midtones reveal the overall color temperature of the scene.
The analysis takes 30–60 minutes per shot for an experienced artist. The result is a lighting model that’s approximately right — usually good enough for non-hero CG integration, sometimes not good enough for hero work.
When plate analysis is the right approach: when on-set reference wasn’t captured but the work has to proceed. Acceptable for moderate-complexity CG; risky for hero work.
When plate analysis fails: when the plate has no useful lighting cues — clean greenscreen plates, talent in flat lighting with no visible shadows or highlights, plates where reflections aren’t visible. In these cases, the analysis can’t anchor against anything, and the lighting model is essentially a guess.
Level 4: Educated Guess
The bottom of the hierarchy is the educated guess. The 3D artist knows the scene description, the time of day, the location, and reads the script for context. They build a lighting model based on what’s typical for those conditions.
A scene described as “interior, evening, restaurant” gets warm low-key lighting from above, with practical sources. A scene described as “exterior, midday, beach” gets hard top-down sun with cool sky fill. The model is generic but plausible.
Educated guesses produce CG that looks “VFX-y” — competent but not specifically integrated with the plate. The audience may not be able to articulate what’s wrong, but the integration doesn’t quite read as believable.
This is what producers are paying for when they skip lighting reference and ask the post team to “match the plate.” The post team will make a generic match. The match will be approximately right. It will not be specifically right, and the difference shows.
What This Costs in Post
Each level of the hierarchy has a different cost in post-production time and quality:
HDRI: 1–2 hours to set up the scene’s lighting in 3D. Result: integration that feels correct on the first pass.
Chrome and gray ball: 4–8 hours to set up the scene’s lighting, with iteration to match. Result: integration that feels correct after one or two review cycles.
Plate analysis: 8–16 hours of analysis and lighting setup, with multiple iterations to refine. Result: integration that’s acceptable for standard work, sometimes inadequate for hero work.
Educated guess: 4–8 hours of setup based on generic models, with significant iteration in review to refine. Result: integration that looks competent but not specifically integrated. Often requires several review cycles to land.
Across a project of dozens of CG shots, the cumulative post-production time difference between full HDRI capture and no reference at all can be weeks of work. The on-set time difference is minutes per setup.
What “Lighting References” Actually Costs On Set
The full lighting reference protocol — HDRI plus chrome and gray balls — takes 1–3 minutes per lighting setup. On a production with 10 lighting setups across a day, this is 10–30 minutes of crew time. The crew member doing the capture doesn’t have to be the DP or the gaffer; a designated VFX assistant can do it during the camera reset between takes.
The capture protocol becomes faster with practice. Crews that capture lighting reference routinely do it as part of the standard reset; crews that capture it occasionally remember every time how to do it and slow down. Productions doing significant VFX work over multiple shoot days benefit from making the protocol part of the AD’s checklist.
How FXiation Digitals Uses Reference
When productions send us full HDRI plus chrome and gray, our 3D team sets up the lighting for the first shot in the sequence within an hour or two, and the rest of the sequence inherits the setup. Reviews land on the first or second cycle.
When productions send us partial reference, we’ll work with what’s there and flag what’s missing. Sometimes we’ll request additional capture (still photos of the location, video of the lighting setup, anything that adds information) before committing to the look-development phase.
When productions send us no reference, we’ll analyze the plates and build the model. The cost shows up in the bid as additional look-development time, and the producer sees what the gap is paying for.
If you’re scoping a project with CG content and want to confirm your on-set capture protocol covers what post will need, send us the brief. The protocol takes minutes per setup and saves hours per shot. We’ll mark up your existing capture checklist or share ours, depending on what helps your team. Producers tell us this is the cheapest piece of pre-production prep they’ve done.
Common Questions