(A 2.7GB, 77252px x 61628px png render)
By using the same FOV=1 → Orthographic view trick that most people are aware by now, you can create an automated software to take bunch of screenshot around the entire map, and then snitch them together later on.
And if you turn the lighting’s exposure down to pitch black, then apply a highlight to segments you want to highlight, you can easily create an alpha mask for whatever you want (multiple group, all in one pass):
This technique allow you to:
- Lighting (well, some of it anyway)
- Texture
- SurfaceGui, Decal, and other textures
while being much higher quality and render way faster too.
However, this technique currently suffer from:
- Lack of depth map (unlike RoRender). I tried using DLL injection to hijack and try to get the depth map, so far no success on that end
- Manual adjusting still needed since I can’t figure out the right math for it (lol). To be fair, some existing pattern matching algo that satellite imaging uses is being applied here, so the tweaking is minimal at worst.
If you’ve tried similar stuff in the past and have some ideas on how to get aronnd these limitations, let me know!