16bit heightmap

[SirArtturi]

One question I wonder is whether a 16-bit mode is really useful if you begin with OTA maps, that are of small dimensions and then enlarge them. I know that you can't regain lost information by upscaling the image, but maybe something happens to the 8-but image when it's being processed.

So what is more important? Colour depth or dimension?

I think it matters! (Actually I think it matters whaever the case might be) Though this is just based on my own experience and deductions, so I might be wrong.

By this I mean, that you basically want to go 16-bits after exporting the map from OTA. Why? Because when you start to enlarge the image, you need to use somekinda method for it. Depending on the filter you use, the number of color depths really matter, cos' the more the filter has them in use to fill the 'holes' left by enlargement, better the final result will be, and match better with the original proportions. For instance: nearest neighbour filter has much more depths in range to choose the best nearest neighbour. The final result is much more smoother and sharper!

[wolas]

Code: Select all

convert heightmap_good.bmp -depth 16 -selective-blur 50x100+3%  heightmap.tiff

Imagemahick smooths 8bit heightmap quite good to 16bits still you can adjust the percentage, or other values to your needs something like 0.5% = 255/100x0.5=~1,* shouldnt do any harm to heightmap only smoothing.

[KaiserJ]

not too relevant but i get my 16 bit heightmaps by creating a mesh for my level (rather than a bitmap) and then exporting it to xnormal and rendering a heightmap from there