buckinha / diamondsquare Goto Github PK

Hello. I use the plotly module to easily create 3d visualizations of the resulting images. And when using your code, I noticed such a feature that virtually every created landscape necessarily has very unrealistic maximum and minimum points.
You can check this for yourself by running the provided code snippet.

from hkb_diamondsquare import DiamondSquare
import numpy as np
import plotly.graph_objects as go

def show_terrain(terrain_array):
    fig = go.Figure(data=[go.Surface(z=terrain_array)])
    fig.update_layout(
        title="Terrain Visualization",
        autosize=True,
        scene=dict(
            zaxis=dict(range=[np.min(terrain_array), np.max(terrain_array)]), aspectratio=dict(x=1, y=1, z=0.5)
        ),
    )
    fig.show()

map = DiamondSquare.diamond_square(shape=(250, 250), min_height=0, max_height=5, roughness=0.35)
show_terrain(map)

It feels like there is some kind of flaw in the algorithm. I'm pretty sure this problem is not related to rendering issues. But this code in its current form is very poorly applicable for real use.
I understand that the repository is already quite old, but I poorly understand the operation (or rather the implementation) of the presented algorithm, so I cannot say what exactly the problem is.
I would be grateful for any comments.

The "square step" is not being done as efficiently as possible. Namely, A grid of 2n points are being checked for each n points that need to be changed, so half of those that are iterated over are just being skipped.

Thanks for the implementation, this is very useful. If you would change some of the '/' to '//' to make them integer division, the code would also work for Python 3.

Aiming at making this easier to install, so clearly it needs to have some setup.py goodness and so on.