CoChat

Efficient Computational Noise in GLSL

Open access
McEwan, Ian, Sheets, David, Richardson, Mark, Gustavson, Stefan · 2012 · Journal of Graphics Tools

The DOI matches its Crossref record; 4 of the 4 references it cites were found in a registry or an index; none of the 12 sources its claims cite is checked for support yet.

Abstract

We present GLSL implementations of Perlin noise and Perlin simplex noise that run fast enough for practical consideration on current generation GPU hardware. The key benefits are that the functions are purely computational (i.e., they use neither textures nor lookup tables) and that they are implemented in GLSL version 1.20, which means they are compatible with all current GLSL-capable platforms, including OpenGL ES 2.0 and WebGL 1.0. Their performance is on par with previously presented GPU implementations of noise, they are very convenient to use, and they scale well with increasing parallelism in present and upcoming GPU architectures.

Overview

CoChat Summary

TL;DR

The paper presents GLSL 1.20 implementations of Perlin noise and Perlin simplex noise that are purely computational, using no textures or lookup tables, and are compatible with OpenGL ES 2.0 and WebGL 1.0. They achieve practical GPU performance, though a previous texture-based implementation is roughly twice as fast on the tested hardware.

Key findings

  • Purely computational noise functions in GLSL 1.20 use neither textures nor lookup tables and are compatible with all current GLSL-capable platforms, including OpenGL ES 2.0 and WebGL 1.0.
  • With reasonably recent GPU hardware, 2-D noise runs at several billion samples per second; 3-D noise attains about half that speed, and 4-D noise is somewhat slower still, with a clear speed advantage for 3-D and 4-D simplex noise compared to classic noise.
  • The old texture-based implementation is roughly twice as fast as the new purely computational version on current GPUs, although the gap appears to be closing with more recent GPU models.
  • 4D classic noise in the new implementation requires 16 pseudo-random gradients, 64 simple quadratic polynomial evaluations, and 16 gradient mappings, versus 48 2-D texture lookups in the previous implementation.

Methods

  • Design: GLSL 1.20 implementations of Perlin noise and Perlin simplex noise; replace permutation tables with computed permutation polynomials (chosen polynomial `(34x^2 + x) mod 289`), use computed points on N-cross-polytopes for gradient selection, and use rank ordering instead of conditionals for simplex selection.
  • Data/sample: performance benchmarks on selected GPUs (Nvidia GTX260, GTX580; AMD HD3650, HD4850) in Msamples/s; source code for 2-D, 3-D, and 4-D simplex and classic noise, plus periodic classic noise, MIT-licensed.
  • Analysis: benchmark comparison of old vs. new implementations; operation count for 2D simplex code: six dot operations, three mod, two floor, one each of step, max, fract, abs, seventeen multiplications, and nineteen additions (counting vector operations as single operations).

Limitations

  • Stated: GLSL 1.20 lacks dynamic array access in fragment shaders, 3D textures, integer texture lookups, integer logic operations, and good conditional optimization; using single-precision floats for integers gives only 24 bits of precision, and the noise domain is limited by fractional-part precision for input coordinates.
  • Stated: The 4-D gradient mapping does not cover the full polytope boundary—it truncates six of eight corners slightly—and cross-polytope gradient vectors differ in length by a factor of sqrt(N), so higher-dimensional noise becomes less isotropic if vectors are not normalized; normalization is approximate or can be skipped.
  • Stated/evident: The purely computational version is not quite as fast as the previous texture-lookup implementation (roughly twice as slow on tested GPUs); benchmarks are limited to the specific GPUs listed and to that previous implementation, with no modern GPU data.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1080/2151237x.2012.649621

Cites

4 of 4 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 Valid · 6 Not valid · 9 Not checked · 32 Own results and interpretations

19 more claims

Create a free account to see all 47 claims and to verify the 3 references still open.

See all claims — free account
  • Probably not a substantial claim. 15
  • It may state two separate claims as one. 17

Create a free account to see these claims.

Citations

Cites

Show all 4 references

Research works printed but not linked: 2

  1. Simon Green, Implementing Improved Perlin Noise. GPU Gems 2, Chapter 26, Addison-Wesley, 2005.
  2. Ken Perlin, Noise Hardware. SIGGRAPH 2001 course notes, Vol. 24 (Real- Time Shading), Chapter 9, 2001.

Cited by

See all 10

Metadata

Authors
McEwan, I., Sheets, D., Richardson, M., & Gustavson, S.
Affiliations
Linköping University (SE)
Corresponding author
  • McEwan, Ian
  • Sheets, David
  • Richardson, Mark
  • Gustavson, Stefan
Venue
Journal of Graphics Tools
Published
2012-06
Volume
16
Issue
2
Pages
85-94
ISSN
  • 1086-7651 (ISSN-L)
  • 2165-347X
  • 2165-3488
OpenAlex
W2042729712
Cited by
10
References
3
Citations by year
2026: 1 · 2023: 2 · 2019: 1 · 2018: 1 · 2017: 1 · 2016: 1 · 2015: 1 · 2013: 1 · 2012: 1
FWCI
1.06
Type
journal article
Language
en
Open access
Locations
  • Journal of Graphics Tools (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • Apollo (University of Cambridge) (submittedVersion)
  • arXiv (Cornell University)
  • Apollo
Topics
  • Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Computer Graphics and Visualization Techniques
  • Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Image Processing Techniques
Keywords
Perlin noise, GLSL, shader programming

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-20, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-20 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.