How do you test and debug shadow mapping and shadow volumes in CG?

1 Shadow mapping basics

Shadow mapping is a technique that uses a depth map of the scene from the light source's point of view to determine which pixels are in shadow. The depth map is stored in a texture, called a shadow map, that is compared with the depth values of the pixels in the final rendering. If the pixel's depth is greater than the corresponding value in the shadow map, it means that the pixel is occluded by another object and should be shaded darker. Shadow mapping is fast and easy to implement, but it can suffer from aliasing, resolution, and bias issues.

2 Shadow volumes basics

Shadow volumes are a technique that uses geometry to define the regions of the scene that are in shadow. The shadow volume is created by extruding the edges of the occluders along the direction of the light source, forming a closed volume that encloses the shadowed areas. The shadow volume is then rendered using a stencil buffer, which counts how many times a pixel enters and exits the volume. If the pixel is inside the volume, it means that it is in shadow and should be shaded darker. Shadow volumes are accurate and robust, but they can be expensive and complex to compute, especially for dynamic scenes.

3 Testing shadow mapping

To test and debug shadow mapping, you need to inspect the shadow map and the final rendering separately. You can use tools like RenderDoc or PIX to capture and analyze the shadow map texture, and check if it has enough resolution, coverage, and precision for your scene. You can also use visual aids like false colors or wireframes to highlight the shadow edges and see if they match the occluders. You can adjust the shadow map parameters, such as size, bias, filtering, or projection, to improve the quality and reduce the artifacts.

4 Testing shadow volumes

To test and debug shadow volumes, you need to verify the geometry and the stencil buffer of the shadow volume. You can use tools like RenderDoc or PIX to capture and visualize the shadow volume mesh, and check if it has any holes, cracks, or degenerate faces that can cause incorrect shadows. You can also use visual aids like wireframes or solid colors to see the shape and size of the shadow volume. You can adjust the shadow volume parameters, such as extrusion distance, capping, or clipping, to optimize the performance and avoid unnecessary overdraw.

5 Debugging shadow mapping

One of the most common problems with shadow mapping is shadow acne, which is caused by the discrete sampling of the shadow map and the numerical errors in the depth comparison. Shadow acne appears as jagged or noisy patterns on the shadowed surfaces, especially near the occluders. To fix shadow acne, you can try increasing the shadow map resolution, applying a filtering technique like PCF or VSM, or using a bias value to offset the depth values in the shadow map. However, too much bias can cause another problem, called Peter panning, which is when the shadows detach from the occluders and create gaps or floating effects. To fix Peter panning, you can try using a slope-scaled bias, a normal offset, or a cascaded shadow map.

6 Debugging shadow volumes

One of the most common problems with shadow volumes is stencil buffer leakage, which is caused by the inaccurate or incomplete rendering of the shadow volume. Stencil buffer leakage appears as dark patches or holes in the shadows, or as light bleeding through the occluders. To fix stencil buffer leakage, you can try using a depth pass or a depth fail algorithm, depending on the orientation of your camera and light source. You can also try using a scissor rectangle, a depth clamp, or a z-fail fallback to reduce the stencil buffer operations and avoid edge cases.