Lyrae Shaders

Lyrae • Physically-Based Real-Time Path Traced Shaderpack

📌 NOTICE: This project is currently in very early alpha stages. Expect issues, visual artifacts and varying performance. I have a huge roadmap. I'd appreciate any kind of feedback and your support!

Features

• Real-Time Voxel Path Tracing

  • Monte Carlo global illumination with multiple light bounces
  • Fast voxel traversal via DDA algorithm
  • AABB tracing for detailed sub-voxels

• Sampling & Image Stability

  • Low-discrepancy blue noise sampling
  • Flexible SVGF denoiser pipeline
  • Next Event Estimation (NEE) for explicit sun sampling

• Physically-Based Materials

  • Unified physically-based shading pipeline
  • Support for LabPBR resource packs

• Post-Processing

  • Filmic tonemapping
  • Color grading
  • Auto exposure adjustment
  • Chromatic aberration

Community

You can join the discord server here, share screenshots or provide feedback and report issues.

Attributions

• Kirilpro6284: 0th ray optimization to start in fragment position.

PBR Support

Lyrae supports LabPBR 1.3 standard. So make sure to use a resourcepack with LabPBR material support, here's a list of resourcepacks on shaderlabs site.

Performance Tips

• I tried to write detailed descriptions for all settings, so you can hover over them in-game for more detailed performance tips.
• Samples is one setting with the most impact on the performance, because it increases the path complexity exponentially. Try to keep this at 1 if you want to enjoy real-time.
• You can try changing the Resolution Scale setting, lowering this from 100% will make the lighting quality lower.
• For most scenarios 3 bounces is enough. 2 bounces will not have indirect lighting, so GI will look less accurate.
• If you are in a completely isolated room where no sunlight is visible, you can try disabling NEE. Note that disabling NEE will break sun light.
• You generally shouldn't mess with SVGF settings unless you know what you're doing. If you are okay with noisy image you can disable spatial filtering.