<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5" -->

---
title: Split Radiance Cascades | daily.dev
description: Radiance cascades can now be extended to 3D scenes using a technique called ray splitting, avoiding the massive memory cost of a full probe grid that scales...
canonical: https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Split Radiance Cascades | daily.dev
og:description: Radiance cascades can now be extended to 3D scenes using a technique called ray splitting, avoiding the massive memory cost of a full probe grid that scales...
og:url: https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5
og:image: https://api.daily.dev/og/posts/6sD2RS8m5.png
og:image:alt: Split Radiance Cascades
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Split Radiance Cascades

**[Hacker News](https://daily.dev/sources/hn)** · 16 min read · 0 upvotes · 0 comments

## Summary

Radiance cascades can now be extended to 3D scenes using a technique called ray splitting, avoiding the massive memory cost of a full probe grid that scales cubically. Probes are allocated sparsely only on visible surfaces using a hashtable, with four cascades of halving resolution built bottom-up. Rays are traced from screen-reconstructed positions rather than probes to avoid self-occlusion, and LODs plus temporal accumulation reduce flicker and memory further. In an Unreal implementation on a 5070 laptop GPU, rendering the San Miguel scene, radiance cascades cost about 2ms, with roughly 13ms for 2spp raytracing and another 2ms for depositing rays.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.jacktollenaar.top/articles/split-rc.html>

## Questions this post answers

### How do you avoid the massive memory cost of radiance cascades in 3D scenes?

Allocate probes sparsely rather than filling a uniform 3D grid, since probe counts scale cubically with resolution. World positions are reconstructed from the screen, rounded to grid cells, and inserted into a hashtable so probes only exist on visible geometry. Higher cascades are built by halving the positions of existing probes bottom-up rather than re-running the gather pass, using four cascades in the reference implementation.

_daily.dev surfaces real-time GI techniques like this for engineers optimizing 3D rendering memory budgets._

### Why do radiance cascade probes cause light leaking or dark splotches when rays are shot from the probe position instead of the surface?

Grid-aligned probes are not aligned with surfaces, so rays shot from a probe can travel below the surface and get unfairly blocked, causing overocclusion and dark splotches, or leak light through walls. The fix is tracing rays from reconstructed world positions on real geometry instead, then placing the resulting radiance into the nearest probe.

_developers debugging GI artifacts can track rendering technique writeups like this on daily.dev._

### What is the rendering performance cost of split radiance cascades for real-time global illumination?

In an Unreal Engine implementation on an NVIDIA 5070 laptop GPU rendering the San Miguel scene, radiance cascade merging and application costs about 2ms depending on probe sizing, on top of roughly 13ms for raytracing at 2 samples per pixel (1 primary, 1 bounce ray), plus another 2ms for depositing rays.

_teams benchmarking GI techniques can follow real-world performance numbers like these on daily.dev._

## Similar posts on daily.dev

- [Ray tracing massive amounts of animated geometry using tetrahedral cages](https://daily.dev/posts/ray-tracing-massive-amounts-of-animated-geometry-using-tetrahedral-cages-fepvmjmjr) · AMD GPUOpen · 0 upvotes · 0 comments
- [New RADV Code Can Deliver 10x Faster Ray-Tracing Pipeline Compilation For Some Games](https://daily.dev/posts/new-radv-code-can-deliver-10x-faster-ray-tracing-pipeline-compilation-for-some-games-s7orkar07) · Phoronix · 0 upvotes · 0 comments
- [Lighting That Feels Just Right: An Exploration of Real-Time Global Illumination](https://daily.dev/posts/lighting-that-feels-just-right-an-exploration-of-real-time-global-illumination-vutl6ptzt) · Maxime Heckel · 0 upvotes · 0 comments
- [Gaussian Splats part 3: There's always a bigger scene](https://daily.dev/posts/gaussian-splats-part-3-there-s-always-a-bigger-scene-uepjmbrpw) · Hacker News · 0 upvotes · 0 comments

---

Tags: [#graphics-programming](https://daily.dev/tags/graphics-programming)

[View this post on daily.dev](https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Split Radiance Cascades","url":"https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5"},"datePublished":"2026-11-22T09:50:12.035Z","dateModified":"2026-11-22T09:50:42.865Z","description":"Radiance cascades can now be extended to 3D scenes using a technique called ray splitting, avoiding the massive memory cost of a full probe grid that scales...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c519c92b8f6f9641c25254171d3bc7b6?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c519c92b8f6f9641c25254171d3bc7b6?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"graphics-programming","timeRequired":"PT16M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"Split Radiance Cascades"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/split-radiance-cascades-6sd2rs8m5#faq","mainEntity":[{"@type":"Question","name":"How do you avoid the massive memory cost of radiance cascades in 3D scenes?","acceptedAnswer":{"@type":"Answer","text":"Allocate probes sparsely rather than filling a uniform 3D grid, since probe counts scale cubically with resolution. World positions are reconstructed from the screen, rounded to grid cells, and inserted into a hashtable so probes only exist on visible geometry. Higher cascades are built by halving the positions of existing probes bottom-up rather than re-running the gather pass, using four cascades in the reference implementation. daily.dev surfaces real-time GI techniques like this for engineers optimizing 3D rendering memory budgets."}},{"@type":"Question","name":"Why do radiance cascade probes cause light leaking or dark splotches when rays are shot from the probe position instead of the surface?","acceptedAnswer":{"@type":"Answer","text":"Grid-aligned probes are not aligned with surfaces, so rays shot from a probe can travel below the surface and get unfairly blocked, causing overocclusion and dark splotches, or leak light through walls. The fix is tracing rays from reconstructed world positions on real geometry instead, then placing the resulting radiance into the nearest probe. developers debugging GI artifacts can track rendering technique writeups like this on daily.dev."}},{"@type":"Question","name":"What is the rendering performance cost of split radiance cascades for real-time global illumination?","acceptedAnswer":{"@type":"Answer","text":"In an Unreal Engine implementation on an NVIDIA 5070 laptop GPU rendering the San Miguel scene, radiance cascade merging and application costs about 2ms depending on probe sizing, on top of roughly 13ms for raytracing at 2 samples per pixel (1 primary, 1 bounce ray), plus another 2ms for depositing rays. teams benchmarking GI techniques can follow real-world performance numbers like these on daily.dev."}}]}
```

