I finished converting my animation keyframe tweening over from cosine interpolation to cubic tonight. The cosine interpolation actually looked fairly decent for the simpler animations, but any keyframe that didn’t have the animation curve change direction would get weird hiccups in it. The cubic interpolation takes care of that, and I now get silky smooth transitions regardless of what the derivative of the animation curve is doing sign-wise.

So this is a little outdated, I actually started writing this post in early September but my wordpress installation was so ancient the image uploader wasn’t working anymore. It’s a nice lazy Sunday after a week of vacation, though, so it seemed like a good time to upgrade.

This is a test I wrote up quickly for the collision system in my project. Nothing fancy to look at, you just move the ball around and make it jump and it doesn’t fall out. I toyed with the idea of using a proper physics system like ODE or Bullet to do this, but I don’t actually want a true physics simulation, I’m shooting for something a little tighter since it’s going to be a run n’ gun bullet hell crossover.

I’ve actually got the protagonist’s model in game and replacing the circle, I’ll be uploading a shot of that as soon as I get the running animations done.

Wow, the past two months have been absolutely insane. Between moving and acclimating to my new job, my poor website has been pretty sadly neglected. But no longer! I seem to have picked the absolute worst time to move, being out of work already due to forces completely beyond mine or my company’s control (I wish I could go into more detail, but NDAs prevent me from doing so). So if anybody reading this is looking for some seriously talented game developing mojo with a good smattering of artist included for free, let me know, I’d love to talk to you.

The upside to all of this is that I now have a LOT more time to work on my personal stuff. A fact I’ve been taking thorough advantage of.

Now that the engine is more or less intact, I’ve been going whole hog on asset creation. Below are some screen shots of the latest model, rigged, posed, and rendered in Blender. I made a few changes from the original concept, such as visible feet and the removal of the teardrop scarf around her neck (might put that back in eventually), but overall I wanted to keep the same feel as much as possible. The face and star/moon on the dress will be added as soon as my Wacom pen gets back to me, I left it in Virginia over Christmas break.

435 verts, 834 triangles after loading into the game engine.

For comparison, these are two of the original concept sprites.

New material support is completed and in the engine. The screen shot below is a quick test of a model with multiple materials applied (and outlining turned on because I never get tired of the effect).

Model was created in Blender and exported with my export script, with zero tweaking of the data in the model file. I just load it into the engine and it works. I could get addicted to this kind of toolchain!

Whew, nothing like getting your code working well and then realizing that adding a new feature to it requires large amounts of refactoring to keep the code readable. I’ve started adding material support to Tsunami, but trying to work it into the mesh creation code was proving extremely difficult. Up until now I had been creating the mesh directly as I parsed the file, but adding materials gummed the works up to the point that I realized I had to throw the existing file parser away. Even if I managed to get it work, the whole thing would be so ugly that nobody else would be able to work with it at all, and I wouldn’t have wanted to.

So, I split the mesh file parsing and mesh construction/rendering code into two separate classes. It’s much, much easier to understand what is going on across the board now, much more robust, and it should also make porting the engine to other platforms (Direct3D, XNA) much simpler later on.

So far the output script for Blender has been modified to export mesh data and the file parsing code to get it into memory on the engine side of things have been completed. Those are the hard bits, too, so actually getting the data out of the parser and into the video card should be a snap.

I need to put some more art up too. The gallery just feels so neglected at the moment.