|
On September 16 2013 18:16 Xiphias wrote:@LaLush's Eco concern. I think I know why the stupid worker AI acts differently in the SC2 engine than it did in BW. This is just a thought, I have done no tests here. This is what I think happens: In BW when a worker gets to a patch and there is someone there it goes to find another IF there are others which are open! In SC2 I think they move regardless if others are not open. I suspect this is why adding a ton more workers in one mineral line is more effective in BW than in Starbow atm. Anyone up for testing this?  @ Lings Vs Marines. Let's test this more and see how it feels. If the engine does this or that it is a difficult thing to prove, much more easy to test and see. What I mean by saying that is simply this: X lings should beat Y marines in a certain setting. If not, then maybe we need to change some values. I have no need to discuss this matter further.
You are correct in that SC2 workers will move around regardless. Nice explanation of the BW queuing condition, hadn't thought of it that way. This "queue when all patches are occupied" behavior mostly just affects income rate at 3worker/patch saturation though (as it seems there are pretty much always some mineral patches free at lower saturations).
However, the only variable that affects max income rate is harvest time. And the implementation of BW style mining in SC2 seems to have a higher harvest time than in BW. The best reason I can think of why is because the stupidity/inefficiency of workers on low saturations in BW is hard to recreate. So they try to match the lower saturation income rates by an increased harvest time.
Reasons why BW worker behaviors that can cause lower than expected income rate on low saturations:
* Turn delay
* Return delay (not sure if this exists in BW, but might).
* Inefficient pathfinding
* Inefficient acceleration/deceleration.
* Slower/inefficient AI when directing workers to unoccupied patches (explains why workers don't autosplit when you send 4 to the same mineral patch at the start, workers queue when they arrive at the mineral patch at the same time).
These things can explain why our implementation in SC2 has a lower max income rate. Because we are trying to compensate for all this stupidity at lower saturations by setting harvest time at a too high value. Not sure if this can be fixed at all to be honest.
But 800-1000 less minerals per 5 minutes (depending on how many patches the base has) obviously will affect the race that tends to be on fewer bases and turtle. For example Terran in TvP and TvZ or Protoss in PvZ. Those races tend to oversaturate their bases more frequently. It can add up over teh course of a game and affect balance.
|
Do you think it would help to add a trigger which makes workers wait by the patch if all others are being harvested the moment it tries to harvest from an occupied patch? (If that made any sense...)
I don't even know if such a trigger is possible to make.
More stuff might be needed as well as you pointed out.
|
Such a trigger would make income rate max out at 3 saturation, instead of being a curve that slowly levels off. Right now, for example, income at 3 workers per patch in Starbow is 4370 per minute. If you go up to 55 workers it's 4830 per minute. If you push it to 100+ workers I think you are almost at max, which is around 4900 per minute.
With a queue trigger you could reach the 4900 per minute at 3 saturation instead. Which would be better I think.
But the max income rate is decided only by harvest time (you can't go up in income any other way when the mineral patch is already being mined from 100% of the time). This is the reason why SC2BW and Starbow have almost the indentical max income rate (their harvest times are 5.45 and 5.412 respectively I think).
The income rates between the mods may vary a bit in lower saturations, but the max is only decided by harvest time.
|
Well, it used to be maxed by 2 workers per patch, so as long as there is a difference in 2 per patch and 3 per patch (in the sense that spreading out on more bases is always better than over-saturation) then I think we can live with that. For a long time we wanted a 1/0.65/0.35 ish relation between the first, second and third worker on a single patch in terms of mineral effectiveness and that might be achieved this way.
In BW too it seems that adding more workers than 3 per patch does very little according to the graph you posted.
Also I'll be racing Braid (speedrun) tonight at around 10 or 11 pm at www.twitch.tv/speedathon vs the current WR holder if anyone is interested  (I am KanBan85)
|
The next step of Starbow?
I upload a patch with the last BW stuff in it. (Lurker dmg, SCV life etc) Now all basics in the game should resemble the BW balance decently well - all stats, build time, unit cost, damage, speed and so on should be quite equal. (It will ofc not be 100% accurate due to pathing, AI, unlimited selection, maybe misstakes in the numbers etc) But decent enough so we finally have a solid foundation. The last weeks have I and December mostly modified everything to be BW. And I looked through a lot of it today and I think it should be ok now.
The masterplan:
1.) Establish BW balance.
2.) Tweak it so the new units & spells & features of Starbow fit in the balance.
Maybe the economy needs some more fine tuning. Maybe some units movement speed are slightly too slow compared to BW. But overall, number 1 should be ok enough.
Ok. So what now?
>>>+ Show Spoiler +Three things: 1.) Before I upload any further patches, it would be cool if we could update the Starbow wiki: http://starbow.wikia.com/All potential changes that happens in the future shall immediately be edited in there. I will also make detailed patch notes from now on. (And publish them in the opening post.) I updated all Barrack units on the Wiki. It would be nice if people could help with this, register an account, go into the Starbow-file in the editor, check values and update stuff on the wiki. I think this can be done within a few days if many of us do it. 2.) I will create another "Unit Tester map" for Starbow. There we can try different solutions for units. (So more people that I can play with it.) Compare stats, see what works. Try this value. Try that. Makes it easier to try suggestions from you all. 3.) I will create a Starbow test map, which is basically just a copy of a normal Starbow melee map. In it we can try alternative balance changes in a "real game." Hopefully can those three methods help us to balance the game better.
What are potential problems in Starbow?
+ Show Spoiler + - Is the economy accurate enough? - Are the macro mechanics (Inject, Chrono boost, Calldown SCV, Reactor) balanced enough? - Should Marines, Zerglings, Zealots have different stats due to the pathing/unlimited selection? - How can we solve so Probes are able to scout vs Zerg with Queen? (It was crucial in BW to scout for Hydra busts etc) - Reapers vs Zerg! - Vikings vs Mutalisks! - Banshees! - Dragoon/Stalker/Immortal?!! <-- Buagaahashahhahahah - Storm! - Sentinel at Stargate or Robotic facility? What to do with it? - Breed? - Dark Swarm! - Nerve Jammer - Planetary Fortress? - Sensor Tower? - Warp in? - Guardian & Devourer? - Is SCV all-ins with fast access to Marauders too strong? These are just some potential questions. It will ofc be hard to get the balance 100% super good with new stuff in the game. But we can try as good as we can. As long as the game is fun, then we are at least doing something right.  If we assume the BW balance is ok enough, then we can make sure so all new things for the races are evened out. For example. Bio has access to Medics with Matrix. But on the other hand, Zerg has access to Banelings. Both of those things changes Bio combat vs Zerg a lot. And both things are very strong for respective race.
Known work I have to do in the editor:
+ Show Spoiler + - Fix so all units deal full dmg vs Protoss shields. Especially to get splash attacks to work.
- Fix some icons shown in the production tab for observers.
- Fix so all players can use the Arbiter spells. (Some players can not see them T_T)
- Fix so hotkeys work. (People report problems with many buttons)
Any thoughts on these subjects and the future work?
Ps. I have published Fighting Spirit on EU now.
|
Nice eco work.
Mmmm, big problem with reducing movement speed. Any increase to melee units decreases how microable they are.
Another big problem is we would need to take every single one of our maps and increase its size by 10% as unit speed is currently based on 168 x 168 rush distances to match BW rush distances. We're at a nice point right now where moving your army around the map doesn't feel so instant like in sc2.
Also how DPS and relative movement speed matches BW. Range extremely close to BW. Buffing movement speed would make that zealot able to hit that spine crawler THAT much quicker. All melee units would be buffed in relation to ranged units (as they already seem to be due to smarter pathing).
Also keep in mind the matter of perspective. With a zoomed out camera (compared to BW) and being used to HOTS's crazy fast unit speed will obviously make SBOW unit movement seem too slow.
I think everyone here is completely used to the movement speed, having played and obs'ed enough games to adjust.
If that is out of the picture is is more a matter of what is more beneficial to the game.
|
I would say, it would be alright, if you could get a casting time on the marauders weapon switch. I do know that it have been tried before.
It doesn't make sence with a Zero-cast time transformation it might aswell have both weapons enabled. Or maybe we should look into another solution. I don't know. The hybrid marauder have worked fine, but i do think we can still polish on the design.
|
Ps. I have published Fighting Spirit on EU now.
Its bugged. Players don't start with any bases or workers.
|
Do we really want 60 hp SCV's if marauder all-ins seems strong already? That's one thing I really think we can scrap from BW before implementing it.
45 hp seems just fine imo.
|
On September 17 2013 04:15 Xiphias wrote:
Do we really want 60 hp SCV's if marauder all-ins seems strong already? That's one thing I really think we can scrap from BW before implementing it.
45 hp seems just fine imo.
I think if you tweaked worker pathing it would make it better. Plus, if you reworked bio pathing a bit, marauders wouldn't be as good. Come to think of it, marauders and medics jsut sound scary....
|
Also, any news on when siege pick up will be back in the game (as an upgrade)?
December removed it due to some bug. Not sure what will happen with it. If it can be fixed, maybe it can come back.
Its bugged. Players don't start with any bases or workers.
I fix that now. Let me know if there is still any problem.
Do we really want 60 hp SCV's if marauder all-ins seems strong already? That's one thing I really think we can scrap from BW before implementing it.
45 hp seems just fine imo.
It is probably not needed. But I do not want to be arbitrary with the BW values I add. Cause if I only add BW values I think will work or not, then we get some kind of mixed balance. And I want the BW balance to be properly set at first. Then we can playtest it, evaluate it, then change it if needed. And ponder if Terran need some kind of compensation?
Also, all basic stats should be BW now. So that is complete, unless there is something huge I have missed.
|
Starting position at right upper hand corner at FS is misplaced. My nexus was too far away from mineral pathes
|
Warp prism speed has been reduced to BW values (2.5). IMO that's a mistakeas it doesn't create BW balance as terran has speed boost on dropships. If we want to have consistency here we might as well remove dropship speedboost. But we haven't done that as it creates fun games and I believe the similar concept can be applied to warp prism speed, so I believe the change should be reverted.
|
Starting position at right upper hand corner at FS is misplaced. My nexus was too far away from mineral pathes
Fixed
|
I've updated the basic stats for all the zerg units on the wiki. That means I haven't looked at the abilities, spells or upgrades yet. I'll keep going with the zerg tomorrow so if someone else want to edit something you might want to start with terran or protoss.
|
Keep in mind that BW SCV is not only 60hp worker. SCV also has shorter range, faster attack speed and is only worker without passive regeneration.
Also, i really dislike that its impossible to deny scouting as zerg. Zerglings are unable to damage fleeing worker, and its much easier to avoid lings than catch worker. In BW you actually could stop opponent from scouting you. Good lings control did the job, and you could even hold position with lings on ramp Right now, in starbow, you just rightclick on mineral patch and you have all intel you want (and after queen is out its still much harder than in sc2 to stop worker scouting due to much lower queen dps). I think mineral walk was not possible without vision.
|
This testing has made me think that the reason Blizzard went to 5 minerals per trip and decreased harvest time in the first place was that the 5-per-trip-solution is actually closer to Brood War income rate than anything you can achieve with 8 minerals per trip in the SC2 engine.
After trying lots of configurations (like 20 different ones) I no longer think it's possible to replicate BW max income rate in SC2 without seriously super-boosting worker speed (and I mean making them way too fast).
When testing BW income rate I noticed the breadth of gameplay thread has grossly incorrect values for BW income.
Blizzard is actually really close to BW income rate. It's just that the backside to the solution Blizzard found leaves players no reason to want to expand beyond 3 bases.
But in regards to being as close as possible to BW income on a single 8-mineral patch base -- their solution is actually surprisingly accurate (for as much shit as they were given in that thread).
In my older analysis of macro thread I was actually comparing a 9 mineral patch main to a 8 mineral patch one. It's not wholly surprising that there would be some differences there. Still, SC2 on a single base (even 8 vs 9 patches) is closer to BW income than either SC2BW or Starbow can claim to be.
Man. I'm tripping out from this. I still think there exists a configuration for 9 or 10 minerals returned per trip that can give more BW-like mining rates. I'll try more of those tomorrow.
Right now though... I came to the realization that Blizzard did their homework for BW income rates. And the irony of it all is... if they had not been so insistent on keeping mining rates as similar as possible to Brood War, I believe we would have had wandering workers from 1+ worker per patch saturations. Blizzard kept TOO true to Brood War.
Just something that suddenly hit me when I realized there was no way I could recreate the BW mining rates with 8 minerals per trip. Hahaha.
|
@Lalush
Oh god the irony!!!!
Even after all of that, SC2 style eco is entirely not ideal.
Main flaws being eco startup time for 1 per patch, and encouraging expansions.
If you come up with an eco system using our current worker speed (I really don't think we could afford to make such a huge change to unit speed right now) but replicates BW's ability to quickly get really nice eco with just 1 per patch (quick startup time), no bounce, and that encourages taking expansions that would be amazing.
SC2 engine is such hell. Moving shots, micro, eco, air dancing, pathing >.<.
|
On September 17 2013 05:30 Hider wrote:
Warp prism speed has been reduced to BW values (2.5). IMO that's a mistakeas it doesn't create BW balance as terran has speed boost on dropships. If we want to have consistency here we might as well remove dropship speedboost. But we haven't done that as it creates fun games and I believe the similar concept can be applied to warp prism speed, so I believe the change should be reverted.
Just some quick follow up thoughts on whether strong dropships are desireable or not.
I believe that dropships makes the game better/more entertaining when dropplay is strong in the phase where nothing else is really supposed to happen/turtling is easy. For instance early game/early midgame TvP in BW was often times quite boring. Yes once in a while we saw a timing push, maybe a reaver drop if we were lucky, dragoons shotting at a bunker or vultures planting mines over the map - but genererally stronger harassoptions in this phase of the game would be desireable.
While reaver drops could do a lot of damage, it also suffered from being tech heavy, thus it first came into play in the midgame (not in the early game), and it was required that you did a decent amount of damage with it. Stronger warp prism with Stalker + Immortal combo makes early game harass more viable as the investment is less, and since the damage output of these units are relatively limited, the risk/reward is relatively low, which IMO is good. This means that we have some early game action + multitasking without the potential downside of a very volatile gameplay.
However, dropplay should not be so strong in phases where it potentially replaces real battles. For instance, if one player can't move out of his base out of fear of being counter-dropped, this isn't benefical for the gameplay. Generally, static defenses have an important role here. Further, one could also question whether the idea of warp prism having a slow standard speed with an upgrade is good for the gameplay. Why not just give warp prism a default speed of 3 and remove the upgrade? This means that warp prism doesn't become unnecesarily good late game, but instead, it becomes strong when it is benefical for the gameplay.
|
When you dropped stalkers/immortals, robotic costed 150/100, stalker range 100/100 with x3 faster BT
Warpprism moved around 0.5 faster.
Warpprism around 0.2 faster buildtime
No wonder you could do so much with those drops. Even when i did a push against you, there was not hard for you to get units up and a warpprism and start dropping right after.
Protoss still have the warpprism upgrade.
Looking at dropships/warpprisms for action is the wrong way to go, when someone learns to defend he will deflect drop.
Its what that guy do after, if he still just camps and expansions we still get a slowpaced game
There must be other ways for a more consistent action game.
@ecobooster
I have some concerns here
The ecobooster + fast expansion makes it double effect = the opponent must either go economy himself or allin attack.
I believe even some one base harass will be very hard now, because either do alot of damage or be behind tremendously.
In bw, it already were economy based + tight defence = long uptime before the "game started".
Hider wants faster actionpacked, dont we all?
One of the reasons for the long uptime is as i said the economy gameplay.
BUGS
The nervejammer can not be killed = immune to damage
hallucination units are immune to damage = they should be taken x3 damage
The spidermine bug completed(?) = If many spidermines hit a unit, not all did damage
Something offtopic
I wanna mention this, i gave some thoughts to how blizzard did sc2. If we look at terran bio versus protoss = protoss needs aoe or he die horrendous in bw.
Bio were not availabe for terran at all.
In sc2, it is the same thing here = no aoe for protoss, he dies. But bio is available here.
So he needs colossous or hightemplar or he have no chance. Pvt in sc2 is kinda very bad but its not that blizzard failed that hard if you look at the surface.
Though, they could have had skipped the ghost vs templar duel, what would have happened if they did?
Would ghost and templar be even more dominant? Maybe they would.
What would have happened if they skip colossous = cant take damage from air? Would colossous be even more dominant?
Look at the reaver, if the reaver is well protected in a big armee, its very hard to reach him if you do not have a long range unit like the siegetank for example.
|
|
|
|
|
|
EPT
