This repository contains Digital Terrain Models collected by the authors of the state-of-the-art or obtained by the re-implementation of methods. It is meant to be public and maintained throughout time.

300 steps have been applied of Cordonnier 2017 method

16 bits greyscale Tiff Cell size: 10m/pixel Alt min: 1057.71 Alt max: 2061.54 Latitude (°): 45 Compass (°); 90

256x256

Provided by authors

[Cordonnier-2017.tiff] : main DEM

[Cordonnier-2017-0iteration.tiff] : initial terrain before simulation

8 iterations of diamond square algorithm applied on an initial 8x8 map.

PNG 16 bits unsigned int greyscale.

1793x1793

Re-implemented

[Fournier-1982.png] : final DEM (8 subdivisions)

[Fournier-1982-0.png] : initial 8x8 DEM (no subdivision)

[Fournier-1982-1.png] : first subdivision

[Fournier-1982-2.png] : second subdivision

[Fournier-1982-3.png] : third subdivision

[Fournier-1982-4.png] : fourth subdivision

[Fournier-1982-5.png] : fifth subdivision

[Fournier-1982-6.png] : sixth subdivision

[Fournier-1982-7.png] : seventh subdivision

Faulting algorithm (also known as random faults).

PNG 16 bits unsigned int greyscale.

1024x1024

Re-implemented

[Faulting.png] : 400 iterations of the faulting algorithm (i.e. 400 faults are present)

[Faulting-50.png] : first 50 iterations

[Faulting-100.png] : first 100 iterations

[Faulting-200.png] : first 200 iterations

Digital elevation models from Tasse et al 2014. These are real DEMs modified using the first-person technique described in the paper.

Tiff 16 bits greyscale unsigned int. No special information on the spatial range.

512x512

Provided by authors

[Tasse-2014.tif] : DEM

[Tasse-2014-2.tif] : another example

[Tasse-2014-3.tif] : another example

Hydraulic erosion performed on a noise

16 bits greyscale png

512x512

Re-implemented

[Hydraulic.png] : eroded DEM

[Hydraulic-original.png] : initial DEM

[Hydraulic-sand.png] : amount of sand

Digital elevation model from Guerin et al 2017

PNG 16 bits greyscale unsigned int. 30m per pixel

2048x2048

Provided by authors

[Guerin-2017.png] : final DEM amplified x4 with a sparse method

[Guerin-2017-input.png] : the input sketch at 512x512 resolution

[Guerin-2017-river.png] : the river footprint

[Guerin-2017-riverbed.png] : the riverbed footprint

Digital elevation model from Stava et al 2008

grayscale PNG 16 bits

1024x1024

Re-implemented

[Stava-2008.png] : main DEM

Terrain fully generated by combining Stream Power erosion law and procedural tectonic uplift

16 bits greyscale Tiff Cell size: 100m/pixel Alt min: 0 Alt max: 5749.82

1000x1000

Provided by authors

[Cordonnier-2016.png] : 512x512 crop of the dem in a 16 bits greyscale format

[Cordonnier-2016-full.tiff] : main DEM

[Cordonnier-2016-uplift.tiff] : uplift rate (0 - 3e-3 m/year)

Digital elevation model from Belhadj et al 2005

color PNG 16 bits

513x513

Provided by authors

[Belhadj-2005.png] : main DEM

[Belhadj-2005-skeleton.png] : the input skeleton that gives the constraints

Sum of 8 octaves of ridged noises, with a lacunarity of 1/0.467 and a persistance of 0.5. The amplitude of the next octave is attenuated with respect to the previously computed octaves by using a cubic smoothstep, in a way similar to Musgrave 1989.

color PNG 16 bits

1024x1024

Re-implemented

[Multifractal-noise.png] : main DEM

Digital elevation model from Gain et al 2015

PNG 16 bits greyscale unsigned int. 10m per pixel.

1024x1024

Provided by authors

[Gain-2015.png] : DEM

[Gain-2015-2.png] : Another example DEM (canyon)

[Gain-2015-3.png] : Another example DEM

[Gain-2015-4.png] : Another example DEM

digital elevation model from De Carli et al 2014.

Tiff 8 bits greyscale unsigned int. No special information on the spatial range.

1025x1025

Provided by authors

[DeCarli-2014.tiff] : main DEM

DEMs from Cordonnier et al. 2018.

32 bits greyscale Tiff Cell size: 50m/pixel 16 bits png greyscale

2000x2000

Provided by authors

[Cordonnier-2018.tiff] : DEM with actual altitudes coded in float32

[Cordonnier-2018.png] : same DEM coded in 16 bits greyscale, with the max greayscale corresponding to 4000 meters

[Cordonnier-2018-1.tiff] : second DEM

Digital elevation model from Guerin et al 2016

PNG 16 bits greyscale unsigned int.

4096x8192

Provided by authors

[Guerin-2016.png] : final DEM cropped at 4096x4096 for rendering harmonization

[Guerin-2016-full.png] : final DEM

[Guerin-2016-sketch.png] : the input sketch consisting of a 1024x2048 DEM

[Guerin-2016-rivermap.png] : the alpha mask for the footprint of the river

[Guerin-2016-riverbedmap.png] : the alpha mask for the footprint of the riverbed

[Guerin-2016-withoutriver.png] : the raw result without applying any river carving post-processing

Sum of 8 octaves of ridged noises, with a lacunarity of 1/0.467 and a persistance of 0.5.

color PNG 16 bits

1024x1024

Re-implemented

[Ridged-noise.png] : main DEM

Sum of 8 octaves of Simplex noises, with a lacunarity of 1/0.467 and a persistance of 0.5.

color PNG 16 bits

1024x1024

Re-implemented

[Simplex-noise.png] : main DEM

Digital elevation model from Genevaux 2015, Figure 22 Winding Road.

PNG 16 bits grayscale

1024x1024

Provided by authors

[Genevaux-2015.png] : main DEM

Digital elevation model from Belhadj et al 2007

grayscale PNG 16 bits

513x513

Provided by authors

[Belhadj-2007.png] : main DEM

[Belhadj-2007-skeleton.png] : the input skeleton that gives the constraints

Digital elevation models from Gain et al 2009

PNG 16 bits greyscale unsigned int. 1m per pixel.

1024x1024

Provided by authors

Digital elevation model from Zhou et al 2007, Figure 10c. Terrain generated from an initial sketch that represents a lambda letter.

PNG 16 bits greyscale unsigned int. No special information on the spatial range.

800x800

Provided by authors

[Zhou-2007.png] : DEM

500 iterations of the faulting algorithm described by Mandelbrot 1982.

PNG 16 bits unsigned int greyscale.

1024x1024

Re-implemented

[Mandelbrot-1982.png] : DEM

Thermal erosion performed on a real DEM

16 bits greyscale png

1024x1024

Re-implemented

[Thermal.png] : eroded DEM

[Thermal-original.png] : initial DEM

[Thermal-sand.tif] : amount of sand

Sum of 8 octaves of ridged noises, with a lacunarity of 1/0.467 and a persistance of 0.5. The amplitude of the next octave is attenuated with respect to the previously computed octaves by using a cubic smoothstep, in a way similar to Musgrave 1989. In addition, the domain is warped using a low frequency simplex noise.

color PNG 16 bits

1024x1024

Re-implemented

[Warped-noise.png] : main DEM