Skip to main content
Account

Table 4 Various presented methods (2012–2019) compared for their modeling, rendering and animation times (\(T_{m}\), \(T_{r}\) and \(T_{a}\), respectively)

From: A survey of modeling, rendering and animation of clouds in computer graphics

Method

Model size (g/p)

\(T_{m}\)

\(T_r\)

\(T_a\)

Feature

[12]-2012

\(200\times 200\times 200\)/–

30 s

Image-based (GPU)

[107]-2014

\(100\times 100\times 200\)/80 K

5 min

Image-based (CPU)

[109]-2014

–/10 K

36 ms

Precomputed lighting (GPU)

[2]-2015

–/840 K

1219 ms

Position-based fluids (GPU)

[28]-2016

–/49K

1173 ms

Lennard–James potential (CPU)

[34]-2017

–/82

42 ms/fr

4.3 ms/fr

Macro-physics (CPU), micro-rendering (GPU)

[24]-2017

–/250K

44 ms/fr

SkewT/LogP diagram (CPU)

[49]-2017

\(1200\times 1200\times 1200\)/–

9 min

Deep scattering, RPNN (GPU)

[99]-2018

10 s/fr

30 min

Key-frame interpolation, morphing (GPU)

[32]-2019

–/12

3.3 ms/fr

1.6 ms/fr

Cloud map, macro-physics

  1. Some of the times are listed per frame (/fr). Model size is given in the sampling grid resolution and/or the particle count (g/p) used