UE4_landscape_tiler.py

import cv2
import sys
import numpy as np

#config
outputSize = 1017

image = cv2.imread('in.png', cv2.IMREAD_UNCHANGED)
dimensions = image.shape

# height, width, number of channels in image
height = image.shape[0]
width = image.shape[1]
 
print('heightmap Height       : ',height)
print('heightmap Width        : ',width)

#imS = cv2.resize(image, (512, 512))
#cv2.imshow("original (512 resize)", imS)

x_name = 0
x = 0
while x+outputSize < width:
	y = 0
	y_name = 0
	while y+outputSize < height:
		out_file_name = "IMG_X" + str(x_name) + "_Y" + str(y_name) + ".png"
		print("Currently rendering " + out_file_name)
		crop_image = image[y:y+outputSize, x:x+outputSize]
		cv2.imwrite(out_file_name, crop_image.astype(np.uint16))
		y+=outputSize-1
		y_name+=1
	x+=outputSize-1
	x_name+=1

cv2.waitKey(0)

Great! Thanks for sharing!

Hello, thank you for the tutorial on ue4 forum. I am super new and not sure where to run this code? Can you give me a few more specific instructions to get this up and running? Thanks

Well you just need python 3.6, numpy and opencv4.
Run this where you want, as standalone python script. It expects a file called "in.png" in the current folder and will output in the same folder the tiled heightmap (IMG_Xx_Yy.png), all of it in 16bits.

This is a great find, but i have another question, is there a way to use same script but in reverse?
Meaning i have the tiles in RAW format of 32-bit floating points, and I'd like to merge them back to a single height map and then to slice them for UE using script above since i no longer have the original large single height map.
Is there a way you could help at all?
The tiles i have are 257x257 and must overlap by 1 pixel.
The end result will not be a power of 2 height map.