(08-30-2018, 02:31 AM)epikus29 Wrote: Awesome! Can you tell us how you did it? 
If you still want to open the uexp file you can do it with Gildor's umodel (set override game detection to unreal 4.18)
Thank you.
I took a good hard look at the swizzled image in Photoshop and made some observations:
• Things appear to be on a 4×4 grid
• Four adjacent tiles can be assembled to a bigger 8×8 tile that looks right, if your order is [1, 2, 3, 4] => [top left, bottom left, top right, bottom right]
I wrote a script to assemble these 4×4 tiles to my 8×8 tiles and exported them to individual PNGs. I then started to assemble them in order and made some observations:
• Things only look right when you assemble the tiles from top to bottom in four tiles per column
So, in essence:
1
2
3
4
5
6
7
8
9
10
...
Looked kinda right, but:
1 5
2 6
3 7
4 8
9 13
...
Looked
absolutely right.
So I wrote a script to spit out these 16×256 strips, and they all looked right.
I then started to assemble these strips and made some observations:
• The strips look right when assembled from left to right, but a new row must be started when you reach a width of 1024 (64 strips)
So here we are. Much of the process can be automated, but I only worked with the 1024×1024 image. I'm now trying to write a program that is dimension-agnostic and works on any size image.
you, probably Wrote:man, how did you figure out this shit
Just looking at the image up very close and making observations is enough! I used Python and PIL to program the unswizzler. I highly recommend the Python course in codecademy to get started with Python - it's never too late to learn a programming language.