Interior lighting with Blender Eevee light probes: A practical guide

What will you learn in this guide?

In Blender Eevee, light probes let you reuse scene lighting information for specific areas and surfaces. Volume and sphere probes do not serve the same purpose: a volume probe gathers lighting data at sampling points distributed across an area, and the diffuse component of that data is used. A sphere probe records light from a single position in different directions around it; it is used especially for soft and semi-rough reflections.

This guide explains how to place a volume probe indoors, how to approach resolution and quality options before baking, which resource settings to consider when dealing with GPU memory issues, and how a sphere probe’s influence volume and parallax volume work. The goal is to choose settings based on lighting changes in the scene and practical constraints, rather than raising every value indiscriminately.

Requirements and scene preparation

The source documentation describes light probe features in Blender Eevee; it does not specify a particular Blender version, user account, or additional add-on requirement. No version number or account requirement is therefore provided here. Because volume probe data is static, baking must be started manually. For sphere probes, capture behaves differently in the viewport and during rendering.

Before adding probes, assess the room’s light distribution and geometry. Lighting may vary between an area near a window, a dark corridor, and a more evenly lit region. It is recommended to place the volume grid in a way that matches the room’s shape. Avoid allocating unnecessary samples to empty areas or regions where lighting changes little. If detail is needed only in a specific area, consider adding a smaller grid there instead of increasing the resolution of the entire volume.

You do not have to use a single probe volume across the whole scene. Large scenes can be handled with multiple volumes at different levels of detail. Also, if an object is not inside any Irradiance Volume or indirect lighting has not been baked, the world’s diffuse lighting is used to shade that object. Either of these two conditions is sufficient.

Step by step: indirect light with a volume probe

1. Set up the probe volume to match the room’s shape. The volume probe’s sampling points are distributed across a regular grid within the area you define. The X, Y, and Z resolutions of the grid determine how many cells are used in each direction. Consider the room geometry and lighting changes; focus on the areas that need it rather than sampling the entire volume at the same density.

2. Consider resolution and sampling together. Bake Samples determines the number of ray directions evaluated during baking. As the source states, increasing this setting also increases bake time, depending on the size of the scene representation. Surfel Resolution is the number of surfels generated per local unit of distance. Higher values can improve quality, but can also significantly increase memory usage. The guide recommends using twice the highest Resolution value for Surfel Resolution; do not treat this recommendation as a requirement regardless of the scene’s memory needs.

3. Choose which lighting contributions to include in the bake. Volume probe settings control whether light from the world, indirect contributions from light sources, and emissive surfaces are included in the bake. Baking light from the world may provide more accurate lighting, but the correct blending with surrounding volume probes may be lost in return. Test the options according to the scene’s lighting setup.

4. Run the bake and check the data. Do not think of volume probe lighting data as a live record that updates automatically; it is static and requires a manual bake. During the process, the scene is converted into another representation designed to speed up light transport. This representation can take up a lot of GPU memory, so if the bake does not complete, review the scene size and relevant quality settings.

5. Target problematic sample points. During baking, each capture point is assigned a validity score based on the number of backfaces encountered while capturing lighting. Validity Threshold prevents points below the threshold you set from contributing to lighting interpolation. Dilation Threshold can replace the data of low-validity points with data from valid neighbors; Radius determines the distance within which those neighbors are searched for. Review these settings by checking the results in the affected area.

Sphere probes: reflections and influence volume

Think of a sphere probe as a capture for a different need, not as a replacement for a volume probe. It can provide soft and semi-rough reflections on nearby surfaces. Its influence volume can be set to a sphere or a box; Radius or Size determines the volume’s dimensions, while Falloff sets the range over which the influence decreases linearly.

By default, the parallax volume in which the probe reflects light is the same as its influence volume. The parallax volume is expected to roughly match the shape of the surrounding area. If you need to adjust the reflection volume separately without changing the influence boundaries, Custom Parallax lets you configure the two volumes independently. In the viewport, capture occurs when the probe data or position changes; during rendering, capture takes place at the start of every frame. Keep this difference in mind when testing.

Sphere probes are blended with volume probe lighting for fully diffuse reflections. If ray tracing is enabled, they can also serve as a fallback when a ray fails to reach its target. Sphere probe resolution is set in the Scene data panel, while the resolution of the world’s own sphere probe is set in the World data panel. The source specifies where these values are located; no other menu path is assumed here.

Common mistakes and ways to fix them

• Densifying the grid everywhere: Instead of using high resolution in empty areas or places where lighting changes little, arrange the grid to match the room’s shape. If more samples are needed in a local area, try adding a small grid.

• Ignoring GPU memory: The scene representation prepared for baking may not fit in GPU memory. The options provided in the guide include splitting the scene into smaller sections, lowering Surfel Resolution, and disabling Light Probe Volume visibility for objects that contribute little to the bake.

• Raising bias values too much: Normal Bias and View Bias offset sampling to reduce light leakage. However, if the values are raised too high, they can produce bright-looking results on diffuse surfaces or camera-dependent changes. Facing Bias can smooth interpolation, but it can also increase light leakage.

• Confusing when a sphere probe captures: In the viewport, capture requires a change to the probe data or position. At the start of rendering, a capture is performed for every frame. Treat these behaviors separately.

Next steps for architecture and archviz workflows

For interior visualization, a volume probe helps capture a room’s diffuse indirect light within a defined volume, while a sphere probe addresses reflection needs. This distinction makes it easier to plan probe placement more deliberately, especially in scenes with different surface and room conditions. The option to use multiple volumes at different levels of detail in large projects also makes dividing the scene into sections worth considering.

Before production, test grid placement and bake settings in a small section of an interior. When reviewing the results, focus on light leakage, sample point locations, and GPU memory usage. Consider the memory cost of resolution options that improve quality alongside the sampling settings that affect bake time. If using a sphere probe, also check how viewport updates and capture at the start of rendering affect the scene.

Source and license

This guide has been adapted into Turkish from the “Light Probe Volume” and “Light Probe Sphere” pages in the official Blender manual. Both sources are licensed under CC BY-SA 4.0.