Spring RTS Engine

Pressure Line wrote:units do sometimes need help (ie micro) selecting targets... because a targets value in a given situation is not always directly related to its metal cost (which iirc is what Spring uses to determine targetting priority)

AFAIK it attacks units with the highest power, which is: ud.metalCost + (ud.energyCost / 60.0f).
But there is a power unitdef tag to finetune it on modlevel.

indeed. but often a unit's relative value* can increase or decrease depending on the situation it is in, this is where user micro overrides the units automatic target selection.

*ie a reaper that is shooting at a fusion is 'worth' more than a reaper thats shooting at a solar, due to the economic impact of their respective targets. the target selection routines do not take this into account, and a HLT that is closer to the reaper shooting the solar will just shoot at it, even though killing the reaper shooting the fusion benefits the player more. the player notices this and tells the HLT to kill the reaper near the fusion.

all im saying is that the target selection routine isnt perfect, and will not always choose the best target in every situation. taking away the player's ability to control their units directly seems questionable to me...

but as i said, it's Argh's game, and he can do whatever the hell he wants :)

Argh wrote:

and from memory units in a transport with the isfireplatform tag active keep their los

Yes they do, but they're selectable as well, which is not a desired outcome.

Spring.GetUnitIsTransporting
Spring.SetUnitNoSelect

Shouldn't take more than a few lines of code to make an effective monitor. Or you could write some code which detected them being selected, and passed the selection on to the building instead. That might have to be a widget though, I forget if the selection stuff works in the unsynced portion of gadgets.

Argh wrote:I should have just ignored Lurker's question, apparently

Then don't throw around the phrase "fundamentally awesome"

@Gnome: I already use SetUnitNoSelect, combined with some other stuff, for my fake transport behaviors, for correct behavior.

As for the rest of it... meh, whatever. Moving on now.

I have a large list of content to plow through next week, then it's time to build another demo map, showing a few of the other new toys.

The new MortarTank. Model and paint by me.

are you sure its blocky enough?

srsly tho. looks good.

I really want to see how you get that metal right, my metal always looks eh... non-metalish

Metal's actually pretty easy. Here, lemme show you my process.

1. Dump a metal texture.


2. Preshade with a combination of pattern-stamp and Multiply for dark zones. Pattern-stamp with a lighter version of the metal (I have about 8 different "steels" I use) or just use the Dodge tool, or a combination of both, to make the metal highlights.

This is the part that looks hard. It's not. I just redid this from scratch, in about 2 minutes. The real key is to keep it reasonably subtle, unless you want a 'toon look. Keep in mind that the engine will do a fairly good job with lighting, so there isn't any reason to exaggerate it beyond a certain point.

3. Add detail, and then preshade a bit more, when you have a better feel for how light is interacting with the surface. I don't always do two preshading steps, but pretty often. Note the extra "grease stains" at the barrel joints (the barrel's in two pieces, for recoil) and around the muzzle, and the darker muzzle.


4. If you're viewing it flat-shaded, which is how I do it while painting, then you might want to weld it at the end, just to make sure it will look good when rounded, etc. Usually, I don't bother, since I know what Spring will do with it when I do this step in UpSpring.

There ya go... easy metal. Nothing to it, really. I painted that in about 4 minutes, it took longer to type this and make the illustrations than to paint it. Compared to all of the custom work on the armor, it's trivial really.

Lastly, for surfaces like the metal bits on the upper legs, I put the greebles on first, before beginning any shading steps, just to save some time, since everything there is metallic. That may not be appropriate in some cases- for example, a painted greeble needs a different approach.

Anyhow, that's how I do it, because it's simple, fast, and really quite easy. The only times it can be difficult is when you're working with certain metals like copper and brass, which have a very specific verdigris and patterns of wear. For those metals, I strongly recommend looking at reference photos to get them right- and they usually require quite a lot of airbrush.

Oh yeah... and one last hint. If you want to be able to do this kind of thing quickly and easily, how the piece is unwrapped is entirely the key. For example, the seam is on the bottom, and depending on the piece, I either unroll it with Wings, in a nice relaxed way, or distort it a bit but get it true-flat with UVMapper Pro.

Unless it's something where I think players will see it, I just leave the seam, and don't worry about hiding it. However, if I need to hide it, I generally take the texture from about 16-32 pixels in on one side, make a new layer, flip it, then put it on the other side, using the eraser a bit to hide the transition with the other side, and little touchups. It doesn't hide the seam perfectly, but it's mirrored and therefore "close enough" for all but the biggest objects.

The above might sound like, "duh", but many of the uvmaps of people's stuff I've inspected over the years don't have this simple, time-saving thing done. 30 seconds more work on a uvmap can save you many minutes of painting time.

Well, I just got done playing through with 3 RAI with the new version, to get a feel for where things stand in terms of feel and balance.

For one thing, it sure doesn't feel slow, at least on my 8X8 competition-sized map. If anything, it was very fast and really emphasized rush play, and I found myself scrambling a lot. Combat started within 2 minutes. Massive numbers of units were produced by all sides, and destroyed rapidly.

Movement didn't feel terribly slow, except during the early parts. I'm going to have to speed up Engineering Teams and MegaConstructors in particular, I think, they're a bit draggy on slopes and just aren't fast enough to get to the Ancient Facilities during early play, which are spread out a bit more than they were.

Since I didn't get any feedback about the balance patch, I think I'll go ahead and nerf all of the weapons a bit, and see how that works, and if I like it, great, if not, oh well. I think that giving people more time to react to combat might be a good idea, though.

I'm also going to have to fix up the balance and early curves for Resistance before release. Overmind has the advantage now that the economy's fixed up, simply put. It's not a resource problem, it's a spending problem. They simply can't spend their early resources without resorting to turret spam, which, I might add, doesn't work for cost at all, because of successive nerfs requested by people who don't like fixed defenses.

I think that the build-times for factories and units in general needs to get revisted- for example, to counter early Resistance rush, I made it so cheap for Overmind to spam factories that the AI happily spammed a half-dozen, and that's doom for a Resistance player, they get behind the curve too early, and have no way to effectively counter the horde on maps that don't have serious chokepoints. I think that the key may be to revisit the MegaConstructor's gimmicky use-once functionality, or maybe it's just down to slowing the Overmind by raising the initial cost of a LandFactory a bit. Dunno yet. People hate the early part of a RTS, where you're just sitting there, waiting- maybe I should give both sides a factory at the start.

The good news for everybody who hates the Resistance Pillbox is that turrets are mainly irrelevant for rush play for Resistance now. Given their cost and time, it makes far more sense to get a building and make it a bunker instead. It has more hitpoints and costs nothing except for some micro time. So I'm pretty tempted to give Resistance a Bunker instead, so that they have a role for their infantry pretty much forever, and that gives me an excuse to build more infantry stuff- mortars, a heavy-weapons unit, etc. Not a bad thing, imo.

The buildings are really quite fun. They made the game really interesting. Being able to finally interact with the environment is a really big deal, and blowing up lots full of cars, etc., was neat.

Lastly, the new MortarTank rocks, and definitely solves the problems it was intended to solve. It's still a siege unit, it still outranges everything but the longest ranged guns, but it's range has been reduced, accuracy reduced considerably, and rate of fire has been increased, so that now it's not outranging ArtilleryCannons unless you accept the penalties for using long-range fire. That said, a couple of them together can now effectively saturate an area, and you simply cannot ignore them for very long. Even as slow as they are, and they're certainly slow, they're very dangerous, if guarded.

all i can say is that its probably a good thing EE doesnt have a 6 legged mortar mech (because Fang would have beaten you by quite some time if there was ) because that model is looking more and more EE like :/

way bigger turret/gun or way smaller legs

I would like to announce that I've finally figured out a way to fix the worst problems with Resistance and AIs.

[EDIT]I've revamped it, and now it's a lot more cool. All AI players start with a base and enough stuff to give players a pretty rough time. Expect to get rushed, heh.[/EDIT]

this stuff is looking pretty awesome, can't wait to try next version

so AI's can play PURE at the moment? which ones work?

I hope AI do work because my net is so shocking at my new house

I would like to announce that I have finally figured out how to store complete unit automation information and play it back correctly for World Builder, for interactive objects within a WorldBuilder script.

IOW, maps can have mobile elements that actually "remember" their commands now, such as cars driving around on roads, a heliport where helicopters occasionally land, worlds can have animals, robots and people who wander around seemingly aimlessly, etc., and unlike RC3, it's easy to set up- make a unit, give it some orders, save the script, voila.

Hopefully I can extend this to making simple single-player missions fairly quickly. First I want to get a good demo made, though. The whole idea of a "living city" that has actual stuff going on is really pretty thrilling.

If somebody's bored, I'd like a model of some sort of horse-like critter to animate running around a gameworld, and maybe a robotic taxi or something for the demo movie. Preferably skinned, if you can find the time, as I have some IRL stuff going on this week.

I'll make the cars and trucks actual vehicles, also for people wanting to try and extend all of this to Sim City or Godzilla sims or whatever, but in P.U.R.E.'s background, it obviously wouldn't be appropriate, as everybody's dead. Anything still on the road is a robot, so a cute robot taxicab would be nice. I'm thinking that it should echo the design from Syndicate myself

BTW, if anybody's looking for a reference, or has no idea what Syndicate was, here's a helpful link.

This is the entire opening movie, and shows many shots of the car in question. I can't take an exact replica, obviously, but the whole "robot car of the future" or "car on demand" concept (which is a very plausible future-tech thing to have for World Builder settings, imo) is what I'm looking for.

Here are a bunch of other helpful visual references.

That was easy. Need to make a "police" and "ambulance" version for the Syndicate reference to be complete

cute, reminds me of the total recall taxi:
http://farm4.static.flickr.com/3080/270 ... 25.jpg?v=0