▸case-01 I am implementing basic shadow mapping for a directional light in a 3D renderer. Should I render depth to an RGBA8 color texture using custom shader code to pack floats, or format the shadow map render target directly as a 32-bit floating-point depth buffer texture? | pass→pass | 13,938 | 12,961 | -7% | 1 | 1 | 0% | 1,977 | 2,148 | +9% | 0 | 0 | — |
▸case-02 My shadow map renders moiré and striping patterns (shadow acne) across planar terrain. Someone suggested moving the shadow camera far clip plane further out. How should shadow acne actually be mitigated in light-space rendering? | pass→pass | 16,378 | 17,255 | +5% | 1 | 1 | 0% | 2,478 | 2,478 | 0% | 0 | 0 | — |
▸case-03 After increasing depth bias to fix shadow acne, objects appear to float above their shadows without touching the ground. Should I double the shadow map resolution from 2048 to 4096 to fix this detachment? | pass→pass | 11,855 | 13,808 | +16% | 1 | 1 | 0% | 1,927 | 2,032 | +5% | 0 | 0 | — |
▸case-04 I am setting up cascaded shadow maps for an open-world outdoor scene with a camera view distance of 1000 meters. Should I divide the frustum into 4 cascades using equal linear intervals (0-250m, 250-500m, etc.)? | pass→pass | 17,823 | 13,981 | -22% | 1 | 1 | 0% | 2,589 | 2,634 | +2% | 0 | 0 | — |
▸case-05 In my pixel shader for cascaded shadow mapping, I need to choose which cascade texture to sample. Should I compute the distance from the world origin or check the view-space Z depth of the fragment against split plane distances? | pass→pass | 13,219 | 14,931 | +13% | 1 | 1 | 0% | 2,220 | 2,072 | -7% | 0 | 0 | — |
▸case-06 When the player camera rotates or translates, shadow edges alias and shimmer noticeably across cascades. Should I solve this by only updating shadow maps every 5 frames? | pass→pass | 14,958 | 15,377 | +3% | 1 | 1 | 0% | 2,298 | 2,296 | -0% | 0 | 0 | — |
▸case-07 To soften hard shadow map edges, should I average the depth values stored in adjacent shadow map texels first and then compare that single average depth against the receiver depth? | pass→pass | 13,293 | 10,051 | -24% | 1 | 1 | 0% | 2,262 | 1,850 | -18% | 0 | 0 | — |
▸case-08 When building Percentage Closer Soft Shadows (PCSS), can I skip the blocker search phase and just apply a static 5x5 PCF filter to calculate variable penumbra width? | pass→pass | 14,556 | 12,604 | -13% | 1 | 1 | 0% | 2,505 | 2,164 | -14% | 0 | 0 | — |
▸case-09 In PCSS, what is the formula or relationship for calculating penumbra width based on light size w_light, receiver depth d_receiver, and average blocker depth d_blocker? | fail→pass | 10,380 | 10,380 | 0% | 1 | 1 | 0% | 1,646 | 1,600 | -3% | 0 | 0 | — |
▸case-10 My 8x8 uniform grid sampling for soft shadow filtering produces visible square grid patterns and runs slowly. Should I increase the uniform grid to 16x16 or switch to a different sampling strategy? | pass→pass | 17,859 | 14,981 | -16% | 1 | 1 | 0% | 2,532 | 2,355 | -7% | 0 | 0 | — |
▸case-11 Small objects like grass blades and character shoes lack crisp self-shadows where they touch the ground because shadow map resolution is limited. Should I increase standard shadow map resolution to 16k or implement screen-space contact shadows? | pass→fail | 15,852 | 13,406 | -15% | 1 | 1 | 0% | 2,369 | 2,069 | -13% | 0 | 0 | — |
▸case-12 In screen-space contact shadow ray marching, light rays falsely trigger shadow hits when marching behind thin foreground geometry. How should object depth thickness be handled during ray step intersection tests? | pass→pass | 20,841 | 17,913 | -14% | 1 | 1 | 0% | 3,302 | 2,917 | -12% | 0 | 0 | — |
▸case-13 I am writing a ray generation shader for hardware ray-traced shadows. Should shadow rays be cast from the light position toward the visible geometry point, or from the world-space surface hit point toward the light position? | fail→fail | 20,858 | 17,597 | -16% | 1 | 1 | 0% | 2,910 | 2,410 | -17% | 0 | 0 | — |
▸case-14 When tracing shadow rays for hard opaque geometry in DirectX Raytracing (DXR), should I execute full closest-hit shaders or use payload flags like RAY_FLAG_SKIP_CLOSEST_HIT_SHADER and RAY_FLAG_ACCEPT_FIRST_HIT_AND_END_SEARCH? | pass→pass | 19,598 | 16,506 | -16% | 1 | 1 | 0% | 2,696 | 2,738 | +2% | 0 | 0 | — |
▸case-15 How do I produce soft penumbra edges for area lights using hardware ray tracing? Can I cast a single shadow ray directly to the light's center position? | fail→fail | 16,161 | 14,024 | -13% | 1 | 1 | 0% | 2,565 | 2,454 | -4% | 0 | 0 | — |
▸case-16 Casting 1 ray per pixel for soft ray-traced shadows yields noisy results, but 64 rays per pixel drops framerates below interactive levels. How can real-time games achieve smooth soft ray-traced shadows at 1 RPP? | pass→pass | 19,648 | 20,436 | +4% | 1 | 1 | 0% | 2,678 | 2,865 | +7% | 0 | 0 | — |
▸case-17 I am getting severe light leaking artifacts when applying flat depth bias to steep geometry in shadow mapping. How does normal-based bias differ from traditional depth bias? | pass→pass | 29,610 | 13,801 | -53% | 1 | 1 | 0% | 2,404 | 2,489 | +4% | 0 | 0 | — |
▸case-18 I tried blurring a standard depth texture with a Gaussian filter to get soft shadows, but shadow comparisons broke completely. What shadow map representation allows direct filtering/blurring of shadow maps? | pass→pass | 14,999 | 18,545 | +24% | 1 | 1 | 0% | 2,557 | 2,803 | +10% | 0 | 0 | — |
▸case-19 When allocating shadow map texture memory across 4 CSM cascades covering 0m to 500m view distance, should I allocate higher texture resolution to far cascades since they cover a larger area? | pass→pass | 18,206 | 16,902 | -7% | 1 | 1 | 0% | 2,562 | 2,690 | +5% | 0 | 0 | — |
▸case-20 When light passes through a colored stained-glass window, standard shadow maps only render opaque black shadows. How do shadow techniques handle translucent colored light transmission? | pass→pass | 17,806 | 24,714 | +39% | 1 | 1 | 0% | 2,731 | 3,464 | +27% | 0 | 0 | — |
▸case-21 I want to calculate dark crevices and ambient light occlusion in corners where geometry comes close together without direct light sources. What algorithm should I use to calculate screen-space ambient occlusion (SSAO) sampling offsets in depth buffers? | fail→fail | 21,711 | 18,273 | -16% | 1 | 1 | 0% | 3,133 | 2,757 | -12% | 0 | 0 | — |
▸case-22 I am building a wet floor rendering pass using screen-space ray tracing to reflect nearby objects visible on screen. How do I setup screen-space reflection (SSR) ray marching for glossy floor reflections? | pass→pass | 30,785 | 20,522 | -33% | 1 | 1 | 0% | 4,539 | 3,460 | -24% | 0 | 0 | — |
▸case-23 I want to render god rays and volumetric light shafts in foggy air using frustum volume grids or screen-space radial blur. How do I construct a 3D volumetric light grid for atmospheric fog scattering? | pass→pass | 30,959 | 27,670 | -11% | 1 | 1 | 0% | 4,762 | 4,287 | -10% | 0 | 0 | — |