So, I have a full time programming internship now, and that’s awesome.
I’ve still done a bit of personal programming in the evenings and on the weekends, but I haven’t updated my blog.
I’ve been floating between various ideas for a while, and there are many details that I’m not certain of yet, but I’m pretty sure it will involve hex tiles and cubes in space.
This weekend (the weekend including June 23rd, 2013), I started work on a cube editor, and now I have it up to a comfortable usable state. It’s still very crude, but it gets the job done.
Available here. Keep in mind that it’s a dev program, and that it’s likely to be unstable.
I’ll be updating my progress every weekend (maybe).
I just recently got the generation speed up to roughly 26 vertices per millisecond on a single core, so I can now generate much larger volumes without as much wait.
I added a test texture and did some adjustment to lighting calculations.
I now use the gradient of the noise function to calculate the surface normals instead of using the generated vertices.
I initially thought that I made an error in my normal-averaging calculations (on tracking linked vertices), but after adding wire-frame rendering, a lot of small triangles become visible. Right now I’m debating on whether or not to cull these out or come up with a weighted averaging system that includes these.
It still needs some work, but I made it super fast.
With normal averaging.
Without normal averaging. Note: both screenshots have the same poly count.
I created a very hacky method to collect and average the normals so that everything doesn’t look so jagged. It’s really slow at the moment, so I’m only generating a small region.
I’m beginning to implement proper 3D terrain.
The first ones use a nearest-pixel filter and the second ones use a linear filter.
The only noticeable difference here is the height-coloring gradient.