Sunday, November 9, 2014
75 LTBucks awarded
Why 75? Well, obviously I didn't hit the devlog time. But that's understandable...I'm going to blame it on the cold that I've come down with thanks to the frigid Tennessee winter. If I had been able to stay awake until 12PM, I would have gotten 175 Oh well. But why 75 instead of 100? Well, warp rails work!! But the splitting/merging of lanes isn't fully implemented. Luckily, the most intense parts of the logic are all done and working! That's right, today I got to take my first ride on the rails It's already exciting without the VFX & SFX...but I can't wait to see what it'll be like when I add the graphics and sound
Today was intense, to say the least. There was a lot of math involved in getting lane allocation nailed down. When you think about it, what I've really written is like malloc (the C memory allocator), but with wedges in a cylinder, rather than linear blocks of memory. And it works!! Cool
Watching a rail test with 100 AI ships of varying sizes was quite rewarding Since splitting/merging of lanes is not complete, some ships are simply unable to enter until a lane frees up. But with another hour or two, that'll be changed. It's still so cool to watch a bunch of ships zip along those rails, even without effects.
In addition, I didn't quite get around to locking ships inside their lane, so the trajectories of ships are still a bit...dangerous But locking is easy and should require maybe another 15 minutes or so. Lanes are already safely computed, so it's just a matter of locking a ship into that lane rather than just 'pulling' it toward the next waypoint along the rail.
Just need another day of solid work on these rails and things should be in great shape!
Now, tomorrow? Let's lay down some rewards. 100 LTBucks for impressive VFX AND SFX of the rails. It's gotta be compelling. Partial rewards are available for good but not great effects, but I want to push myself to achieve excellence! No reward for completion of all logic -- that should be a given. In fact, I might deduct some bucks if, for some crazy reason, rail logic isn't 100% done after tomorrow. Finally, I'd like to offer 200 LTBucks for a major stretch goal of AI that is fully-capable of using rails. That means AI tasks must use pathfinding, handle requests to enter a rail, etc. I don't anticipate that this can be completed in a day, but it's worth posting a bounty 100 LTBucks for 12PM EST, although it seems I have a pretty bad track record of delivering on that one
So excited to see this work come to completion!!
PS ~ In theory, rails don't have to be perfectly linear. Every rail segment (e.g., the rail that joins to intermediary nodes) must be linear, but the overall path can have some curvature to it. We need to be careful here, because the math to compute non-colliding lanes with arbitrary curvature is something into which I'm not going to delve. But if we apply a bit of 'padding' to each lane (e.g., allocate a lane that is slightly too large for the requested object's size), we can safely add 'a bit' of curvature to the overall lane. I know, it's scary throwing around those words without mathematical precision, but I'll be doing plenty of testing to figure out exactly how much I can safely bend the lanes.
PPS ~ I took an hour-long break today from the warp rail logic to see how far I could push the double-precision engine. I've been curious to do this for a while, since I haven't actually established where it breaks down. The systems I've shown are large (large enough that a single-precision engine couldn't handle them), but make no mistake, they're absolutely tiny compared to what the engine can actually handle. Well, it turns out that I start to lose noticeable precision at coordinate values of around 10 trillion meters (10 billion km). When I say 'start to lose noticeable precision,' I mean that the motion of objects is visibly jerky. So, knowing the scales of ships and camera motion, I would guess that this means we are getting centimeter accuracy out to roughly 1 trillion meters. That's pretty absurd (the current systems are in the realm of 10 million meters or so). So for those of you who want to make a 'realistic scale mod,' or at least a 'close-to-realistic-scales mod,' just know that the LT engine is very friendly and very much ready to take on some absurdly-large game worlds We can't quite do a full-sized solar system with cm accuracy, but we can certainly create spaces that are bafflingly-large enough to provide the illusion of real space
Post
Mon Nov 10, 2014 9:29 pm
#1
Week of November 9, 2014
“Whether you think you can, or you think you can't--you're right.” ~ Henry Ford