Exposure & the Post-Process Look

A Post Process Volume is an actor you place in the level; inside its bounds (or set to Unbound / Infinite Extent to affect the whole level) it applies a stack of image adjustments to whatever the camera sees. Think of it as two things at once. As a camera, it controls exposure and lens effects - how bright the image is, bloom, depth of field, chromatic aberration, vignette - the physical behaviour of a real lens and sensor. As a darkroom / grade, it controls the look after the fact - white balance, contrast, saturation, shadow and highlight tone, colour tints - the equivalent of developing and grading a photograph. Crucially, post-process does not change your lighting or your model at all; it changes how the final image is interpreted and finished. You can have several volumes with different priorities and blend between them, which is how you give, say, an interior and an exterior their own look in one level.

Because it is non-destructive and live, the Post Process Volume is where you do the final polish once the lighting is right - and it is genuinely where a competent render becomes a convincing one. But it is also where beginners accidentally sabotage good work: a wrong exposure setting can bleach or crush a perfectly lit scene, leading people to fight their lights when the camera was the problem. So the order of operations matters: light the scene to sane, real-ish levels (Lesson 4.3), then use the Post Process Volume to expose and finish it. Get that sequence right and post-process becomes the reliable last step it should be, not a mysterious source of frustration.

The real world contains an enormous range of brightness - the sky outside a window can be thousands of times brighter than a shadowed corner indoors - and no screen can show that whole range at once. So the camera must choose an exposure: which slice of that brightness range becomes the visible image. Expose for the dim interior and the window blows out to white; expose for the bright window and the interior falls to black. This is not a flaw, it is exactly what a real camera (and your own eye) does, and choosing it well is a big part of a good image.

Unreal offers two modes. Auto exposure (eye adaptation) mimics the human eye adjusting as you move from bright to dark - walk from a sunlit terrace into a dim hall and the image gradually brightens to compensate, just as your eyes would. It is great for interactive walkthroughs and VR, where the viewer moves between wildly different light levels and you want it to feel natural. But it is the number-one cause of beginner confusion: the scene looks different depending on where the camera was a moment ago, so your carefully lit room appears too dark or too bright for reasons that seem to have nothing to do with the lights. Manual exposure fixes the exposure to a set value (in EV - exposure value / stops), so the image is stable and predictable - which is what you usually want for a controlled hero still or a consistent set of renders. A common professional workflow: use manual exposure while lighting and grading so you are judging a stable image, use auto exposure (tuned, with sensible min/max brightness limits) for the final interactive experience where natural adaptation adds realism. Either way, when a lit scene looks wrong, check exposure first.

The controls that govern this live on the Post Process Volume under Exposure. The Metering Mode is usually Auto Exposure Histogram, which analyses the frame's brightness distribution to pick an exposure; Exposure Compensation then nudges that result up or down in stops (EV) to taste. The two settings that tame the wandering-brightness problem are Min EV100 and Max EV100: clamp them to a narrow band and auto exposure can only drift within limits, so a dark corner cannot bloom to daylight brightness. Speed Up and Speed Down control how fast the adaptation happens - slow it down and the eye-adaptation feels natural rather than a jarring flash as you enter a room. Switching Metering Mode to Manual (or setting Min EV100 equal to Max EV100) pins the exposure outright, which is the stable footing you want while judging lighting and grading a hero still.

Even after you choose an exposure, the scene is computed in high dynamic range - brightness values that far exceed what a display can output - and those values must be mapped down to the 0-to-white range a screen shows. That mapping is tone mapping, and it is doing quiet, essential work in every frame. A naive mapping would simply clip everything above a threshold to pure white, giving harsh, digital-looking highlights. A good tone mapper (Unreal uses a filmic tone-mapping model by default) rolls the highlights off gently the way photographic film does, preserving detail and colour in bright areas and giving the image its pleasant, non-clinical contrast. It is a large part of why modern Unreal renders look filmic rather than flat straight out of the box.

You mostly leave tone mapping to do its job, but it helps to understand it for two reasons. First, it explains why you cannot just read raw brightness off the screen - what you see is the tone-mapped result, not the underlying value, which is why exposure and units interact the way they do (Lesson 4.3). Second, the Post Process Volume exposes tone-mapper controls (and film-like slope, toe and shoulder settings) that let you tune the overall contrast curve of the image - deepen the shadows, soften the highlights - as part of the grade. For most archviz you accept the sensible defaults and adjust lightly; knowing that a filmic curve sits between your HDR scene and the final pixels is enough to stop tone mapping from being a mystery, and to appreciate why real-time archviz finally looks cinematic.

With exposure and tone mapping handling the fundamentals, the rest of the Post Process Volume is where you apply taste. Bloom is the soft glow that bright sources bleed into surrounding pixels - the halo around a lamp, the shimmer off a bright window - which real lenses and eyes produce and which, used gently, adds life and a sense of real brightness. Overdone, it is the hallmark of an amateur render, so keep it subtle. Color grading is the heart of the look: white balance (warm or cool the whole image), overall contrast and saturation, and separate control of shadows, midtones and highlights so you can, for instance, cool the shadows and warm the highlights for a filmic split-tone. A gentle vignette (darkening the frame edges) draws the eye inward; a touch of chromatic aberration and film grain can add photographic authenticity - all in tiny doses.

The golden rule of the architectural grade is restraint. The goal is an image that looks like a very good photograph of the space, not an Instagram filter - the best grades are the ones you do not notice. Nail the lighting first; post-process should refine, not rescue. A reliable finishing sequence: set a stable manual exposure, confirm the tone-mapped contrast reads well, dial bloom back until it is barely there, then grade lightly - a whisper of white-balance warmth, a gentle contrast curve, a soft vignette - and stop early. This is the same darkroom craft a photographer applies, and it is the final 10% that separates a flat render from an image a client believes. From here, Module 6 puts a real camera and composition around this look, and the Movie Render Queue outputs it at full quality; you now have the complete lighting-to-image chain.

The concrete controls are worth knowing so you can keep a light hand. Bloom has two methods - the cheap Standard (a good default for real-time and VR) and the higher-quality Convolution for hero stills; keep Bloom Intensity low, roughly 0.2 to 0.6, so sources glow without haloing. Under Color Grading you set Temperature (white balance in kelvin) and Tint, then push Global, Shadows, Midtones and Highlights wheels for saturation, contrast and gain separately - the way you can cool the shadows and warm the highlights for a filmic split-tone. For a repeatable house look across a set of renders, bake your grade into a LUT (a colour lookup texture) and apply it via the volume, so every image in the deliverable shares one consistent, restrained treatment rather than being eyeballed one at a time.