|
On other topics, I finished the first quest in DemonCrawl after a few attempts. It has the same problem that Immortal Redneck has with scrolls: chests are your main source of upgrades, but sometimes they gently caress you over! That said, none of the ones I found hosed me over as dramatically as some of the worst scrolls in IR; on my winning run I limped into the final level with half inventory space, poison, and something that reset the entire board every 150 turns and still managed to squeeze out a victory. The UX continues to be completely awful, and looks like it was lazily ported from PC, too (there are references to right-clicking). It also chews through battery alarmingly fast. Despite that, I want to play more, and will probably pick up the PC version.
|
#
?
Nov 20, 2019 02:38
|
|
|
|
| # ? May 23, 2024 17:32 |
|
|
Alright, alright, look, I said I'm sorry, okay? Let's not get snippy about it. Talking about this kind of thing is always a balancing act because people will either complain that you're talking over their head or that you're not taking them seriously enough. Programming and computer science are different from each other in the same way mechanical engineering and theoretical physics are, so I try not to make any assumptions, but you always end up stepping on somebody's toes anyway. I don't exactly have any clairvoyant knowledge of your CV, so let's just leave it at that. ToxicFrog posted:If it were entirely about pathfinding for dwarves, running clean and cleanowned wouldn't have such a dramatic effect. But, as it happens, it does, because while all the other stuff is a lot less expensive individually, there's also a lot more of it. It's still likely nowhere near the majority of DF's CPU load -- I say "likely" because I don't have detailed profiling information for DF, and neither, I suspect, do you -- but it's not trivial either. What I was talking about was a situation where we have an island map that's, say, 50k by 50k squares in size, filled with creatures and obstacles and being wandered by 200-300 NPCs who are actively pursuing each other, fighting, figuring out places to set ambushes or traps, having to deal with environmental interactions, managing inventories, etc. etc. etc. No unloading, no sneaky off-screen shortcuts. Even if you say that Dwarf Fortress is still playable, I do not see a turn-based game with a latency of 0.5 seconds per keypress and therefore player turn as something playable enough to call it a feasible project... and honestly, I consider Dwarf Fortress factually unplayable once it drops into the sub-10 FPS range. Yes, it's technically still doable and no, I do not know that number for a fact, but I think we can guess at it well enough to assume that this is probably enough to start causing problems for the average home user with a middlingly old desktop PC or outdated notebook. So if it really matters that much to you, I suppose you win that argument because it's technically not impossible, just really impractical. Cardiovorax fucked around with this message at 03:03 on Nov 20, 2019 |
|
#
?
Nov 20, 2019 02:59
|
|
|
Nah, I call bullshit. Sure, if you try to just run an APSP algorithm on a 2.5 billion node grid graph once per turn then it's not going to go super-well for you. But how often are your NPCs on that island going to need to pathfind to something outside their immediate vision range? And how often will that be to something that's not a major landmark where you could compute the paths in advance? A battle royale island is a far more open space than a dwarf fortress - why not go for an approximate solution, e.g. by splitting the map up into large chunks and considering a graph based on which chunks are connected to each other? Again unlike a dwarf fortress, a battle royale island is mostly static terrain, so most NPCs pathing data will stay very similar from turn to turn - why not try and exploit that? Why not go for a solution where all (or almost all) the NPCs are initially ignorant of the island's layout and learn it as the game progresses, so your problem sizes only get large for NPCs that survive a long time? There's got to be some way of doing it. (I'm a CS lecturer, but the sort who normally stops caring after "polynomial time" so likely all the devs in the thread know more than me about pathfinding algorithms.)
|
|
#
?
Nov 20, 2019 03:41
|
|
|
pumpinglemma posted:Nah, I call bullshit. Sure, if you try to just run an APSP algorithm on a 2.5 billion node grid graph once per turn then it's not going to go super-well for you. But how often are your NPCs on that island going to need to pathfind to something outside their immediate vision range? And how often will that be to something that's not a major landmark where you could compute the paths in advance? A battle royale island is a far more open space than a dwarf fortress - why not go for an approximate solution, e.g. by splitting the map up into large chunks and considering a graph based on which chunks are connected to each other? Again unlike a dwarf fortress, a battle royale island is mostly static terrain, so most NPCs pathing data will stay very similar from turn to turn - why not try and exploit that? Why not go for a solution where all (or almost all) the NPCs are initially ignorant of the island's layout and learn it as the game progresses, so your problem sizes only get large for NPCs that survive a long time? There's got to be some way of doing it. Agreed. "Maybe... you don't have to forego all approximations and optimizations?" was my initial reaction and I don't even have any particular CS cred worth naming, just an engineer who has been using the cheat code "3.14159=close enough" since middleschool.
|
|
#
?
Nov 20, 2019 04:30
|
|
|
Honest question, why a 50k by 50k grid? Having never developed a roguelike, I assume that would be way too loving big to have fun, but I don’t have a good point of comparison . A 2x2 embark is 96 tiles across. A 1000x1000 map would be over a hundred times as large before you factor the important value of depth. What is the size of a map in UnReal World? Id feel confident that would be too large for a competitive battle royale. Cataclysm: DDA might be a good point of inspiration for map sizes.
|
|
#
?
Nov 20, 2019 04:35
|
|
|
madjackmcmad posted:Hunger Clock
|
|
#
?
Nov 20, 2019 05:09
|
|
|
Well, as a point of comparison, if you do nothing but tap left, and you manage four taps per second, it will take roughly three and a half hours of tapping to cross a 50k-tile-wide map. Desert Bus RL when
|
|
#
?
Nov 20, 2019 05:11
|
|
|
Put another way, on a "standard" 80x24 column display that 50kx50k world would be 1.3 million screens. That gets into Minecraft territory of "sure it's there, but why would you bother exploring it?" for me. My usual rule of thumb for "entirely big enough" is the maximum size of a Vanilla Angband dungeon, which is 16 80x24 screens, or just over 30k tiles. That's big enough to have a good bit of exploration and avoidance, without making it completely trivial to run away from every problem. It might not be enough to represent a Battle Royale island; I'm not familiar with the genre enough to say. But I'd wager it could do a pretty good approximation. But in fairness to Cardiovorax, 30k tiles is also more than enough that a naive A*-based pathfinding design (without navmeshes, heatmaps, or any other clever tricks) will have noticeable slowdown with even a small number of actors. EDIT: vvv because your training has covered big-O analysis but nothing else about performance. TooMuchAbstraction fucked around with this message at 05:53 on Nov 20, 2019 |
|
#
?
Nov 20, 2019 05:28
|
|
|
Why would you ever assume no optimizations?
|
|
#
?
Nov 20, 2019 05:35
|
|
|
Oh, for god's sake, it's called "a loving example." I've played even completely conventional roguelikes like ToME 2.3.4 where I could start feel a noticeable delay between turns on just a regular-sized overland map when holding down a directional button to travel, if it was too filled with enemies.
Cardiovorax fucked around with this message at 06:11 on Nov 20, 2019 |
|
#
?
Nov 20, 2019 06:09
|
|
|
And it must be pathfinding? Not loading an object per enemy from god knows where in memory and thrashing any cache locality? Credentials: I’ve played over 4 roguelikes
|
|
#
?
Nov 20, 2019 07:25
|
|
|
You could have a seamless map, or you could have regions like Qud, and due to murder collars you’re informed when someone is within 5 regions and everything outside that radius is abstracted
|
|
#
?
Nov 20, 2019 07:30
|
|
|
I am not a programmer or a computer scientist but fwiw I once played a roguelike that did track all the NPCs in the game fighting each other and leveling up even while the player was on other levels. I don't know how it did it or even if it fully tracked movement for all of them. I want to say it did but it was a long time ago so I can't completely trust my memory. Unfortunately I can't remember what it was called, and I have no idea if it got past alpha or beta stages of development, but someone has actually made something along those lines that was playable in some form.
|
|
#
?
Nov 20, 2019 09:57
|
|
|
Cardiovorax posted:Oh, for god's sake, it's called "a loving example." I've played even completely conventional roguelikes like ToME 2.3.4 where I could start feel a noticeable delay between turns on just a regular-sized overland map when holding down a directional button to travel, if it was too filled with enemies. I am not a compsci grad or a games programmer but I have made a lot of bad posts on the internet, so allow me to say this isn't the kind of thread where you'll do well by trying to be superior or by yelling at people about stuff. It's a nice place full of lovely and smart people talking about a deeply niche games genre. Drop the aggro, it's not going to help.
|
|
#
?
Nov 20, 2019 09:57
|
|
|
LordSloth posted:Honest question, why a 50k by 50k grid? I've never played URW, but I was struck by the proposed 50k×50k so I did some math. A game of Nethack -- assuming it generates the largest possible dungeon and you obsessively dig out all the walls so that each level is a single giant room -- has a total playable area of about 18k tiles. (In practice of course it's much less because most area of most levels is solid rock, but let's call it 18k as an optimistic approximation.) This means that the notional 50k×50k map would be roughly as large as 19 thousand games of Nethack laminated over a flat surface. Assuming 200 contestants, each one has area equivalent to 94 complete copies of the Mazes of Menace to frolic in. You could wander for a very long time without encountering anyone.
|
|
#
?
Nov 20, 2019 12:36
|
|
|
ToxicFrog posted:I've never played URW, but I was struck by the proposed 50k×50k so I did some math. That would actually be an amazing premise if the other contestants were players and you ran into others with that level of rarity, where you play for hours and hours and once in your entire life of playing it, you run into one other person.
|
|
#
?
Nov 20, 2019 12:52
|
|
|
Thirsty Dog posted:I am not a compsci grad or a games programmer but I have made a lot of bad posts on the internet, so allow me to say this isn't the kind of thread where you'll do well by trying to be superior or by yelling at people about stuff. It's a nice place full of lovely and smart people talking about a deeply niche games genre. Drop the aggro, it's not going to help. I see you’re unfamiliar with Cardiovax. Their posts are Not Good. I’m amused you all are doing some cool brainstorming on the idea. Putting more thought into it than I did, lol. Galaga Galaxian fucked around with this message at 18:51 on Nov 20, 2019 |
|
#
?
Nov 20, 2019 18:48
|
|
|
Rogue Royale seems like a fun concept. Here's my pitch. * Spawn in a bigass map but not Too Big, 100x100 is probably honestly fine * Consumables and gear scattered all over * Start as a class with maybe 4 key abilities (QWER) plus a passive or two * Many walls are breakable; some take one turn, others 2-3, depending on tools you have * Monsters can fight each other * Monsters can pick up gear and equip it * Consumable items that summon stuff like dummies, traps, caltrops, etc * You gain XP when you kill another creature, you gain more XP the fewer things are left (eg. killing a creature when there are 100 total is worth much less than when there are only 50 left) * Monsters have personality profiles related to their creature type: timid hoarders, aggressive attackers, campers, etc * After turn X, the map starts shrinking every Y turns * Of course, you can share a seed with friends so they can try the same map! Hell this would be fun to do in the Tangledeep engine.
|
|
#
?
Nov 21, 2019 03:39
|
|
|
zirconmusic posted:* You gain XP when you kill another creature, you gain more XP the fewer things are left (eg. killing a creature when there are 100 total is worth much less than when there are only 50 left) Also, after X amount of creatures are killed, all the surviving creatures level up
|
|
#
?
Nov 21, 2019 04:44
|
|
|
zirconmusic posted:Rogue Royale seems like a fun concept. Here's my pitch. this and there's a nemesis system or something where they get scarred or die, and you end up fighting their whole family or community in successive rounds
|
|
#
?
Nov 21, 2019 04:51
|
|
|
doctorfrog posted:this That sounds like something that would be at home in one of those Soldak games.
|
|
#
?
Nov 21, 2019 05:27
|
|
|
doctorfrog posted:and there's a nemesis system or something where they get scarred or die, and you end up fighting their whole family or community in successive rounds Hell if you're going the fantasy route just say everyone gets resurrected after each game and make it a Magical "Reality TV"-esque Bloodsport. Then either have a handful of reoccuring "rival" NPCs and/or just do the personality types and have a list of 100+ names that get drawn for each enemy.
|
|
#
?
Nov 21, 2019 06:14
|
|
|
I’d like to see it done by stealing Yahtzee’s idea, myself - you’re not a contestant, you’re the producer, and you’re trying to nudge people together in ways that make for good television.
|
|
#
?
Nov 21, 2019 11:01
|
|
|
But what if... that one episode of samurai jack?
|
|
#
?
Nov 21, 2019 11:07
|
|
|
Big money! Big prizes! I love it!
|
|
#
?
Nov 21, 2019 15:54
|
|
|
There must have been a SmashRL created for the 7DRL, right?
|
|
#
?
Nov 21, 2019 16:52
|
|
|
Motherfucker posted:But what if... that one episode of samurai jack? The one where him and the Scotsman team up and only get rescued by the Scotsman's wife who is even stronger than him? The one where a black dude claims to be an awesome samurai so Jack makes him fight with sticks and he only has points in katana so he loses? ...Samurai Jack episodes basically all seem like good emergent roguelike gameplay.
|
|
#
?
Nov 21, 2019 17:02
|
|
|
Anybody played Children of Morta? Pretty charming game. I'm extraordinarily inept at 2d games of this type though. Is this something that can be brute forced or should I refund it? Morta kinda has me hooked, despite dying 5 times in a row to the first boss thus far. Maybe it's because I grew up solely playing turn-based games and old-school fps, but I'm just irredeemably terrible at anything resembling Binding of Isaac or a platformer. Fruits of the sea fucked around with this message at 18:03 on Nov 21, 2019 |
|
#
?
Nov 21, 2019 18:01
|
|
|
I just feel like a decent many-optioned arena roguelike would go a long way. You can constantly add all kinds of new content and systems, biomes and powers, and let players and AI dictate better balances and more options for different modes. It can be small and contained, instead of an entire world from the start, and explore any kind of imagined setting and stuff if built for easy modding interaction and community content. Expand it from a small arena to a whole town, to an entire island. The build-your-own aspect and the idea of doing stuff like salty bets on AI fights feels like it'd make for a lasting and solid roguelike.
|
|
#
?
Nov 21, 2019 20:16
|
|
|
Roguelike / lite but with the nemesis system from Shadow of Mordor
|
|
#
?
Nov 21, 2019 20:26
|
|
|
Yeah exactly But apply it to any entity imaginable, until you can have a Ranger/Investigator dual-class Moloch riding atop of a leashed Chrome Pyramid being carried by Fire Orbs as your nemesis
|
|
#
?
Nov 21, 2019 20:31
|
|
|
Arena shooter using the DoomRL engine? 
|
|
#
?
Nov 21, 2019 21:17
|
|
|
So now that the update has been out for a while, does anyone have any thoughts on the new ToME class? Not really sure which trees are worth investing in since they all seem good, and spreading thin seems like a bad idea.
|
|
#
?
Nov 21, 2019 21:38
|
|
|
Ibram Gaunt posted:So now that the update has been out for a while, does anyone have any thoughts on the new ToME class? Not really sure which trees are worth investing in since they all seem good, and spreading thin seems like a bad idea. I won on Insane earlier today on a stupid brainlessly sloppy faceroll run, was a good enjoyable way to learn the new meta of Things That Will Absolutely Kill You in 1.6. Out of the locked trees, Artillery for rocket pods is absolutely far and away the strongest option, as a tinker class with ranged attacks you have access to stacking up a ton of secondary damage that isn't reduced by the half weapon damage modifier of the rocket pods. (Rocket Pods should perhaps get a proc damage penalty, sorry folks.) You will be able to focus on bursting down the most dangerous target in a room, and before you're finished doing it everything else will be dead too. The mecharachnid seems cool but you can't shoot through it which is a big bummer in a game with hallways everywhere. Grenades are way too strong right now. The tree self-synergizes too much, going all in on Barrage and mixing in Lock On will give you something like 530% weapon damage at a little over 50% aspd on your primary target, plus a 20-25% damage taken debuff from incendiary grenade, and you're a weapon class so your base damage can absolutely explode with good investment. This is an impossibly high amount of very fast burst damage, with plenty of reaction time from the overtuned speed boost to neutralize danger and then resume slaughtering. Between the Incendiary Powder armour shred and optionally an acid groove tinker (remember that you can swap tinkers situationally) high armour can't counter your weapon DPS, and if it could Incendiary Powder also adds even more nonweapon fixed damage just in case. I took Windtouched Speed then PES for prodigies and absolutely blendered the entire game, killed Master through just splash damage shooting his summons (which were also instantly annihilated) before he could take a second turn. Including resurrection. Before PES I would occasionally get confused because I was dealing damage as if I certainly must have had PES or Arcane Might. But I didn't! Like many initial releases of Razakai classes (I assume this is a Razakai class? I absolutely love your work, buuuuut) the shield tree also synergizes with itself far too much. Everything good in this class right now synergizes too much. Lightning Web combines with passively increasing your block value from Capacitor Discharge to give a continuous 6 turn flat damage mitigation for roughly your shield's base block value, which is like if Eternal Guard lasted 6 turns and didn't cost you a prodigy and also passively gave you a damage field that triggered debuffs from melee procs you otherwise wouldn't have access to as a ranged attacker. Let me emphasize, the damage mitigation from this is absolutely ludicrous. In the early game, it will be giving you more damage mitigation than a lategame Antimagic Shield stacked on top of Cloak Tesselation or Brawler Striking Stance. In the late game absolutely nothing compares. The class also gets access to 3 very strong overlapping sources of critical mitigation (Magnetic Field, Last Engineer Standing and the Ablative Armour tinker) which checks off another safety box of "neutralizing a thing that will kill you," being almost completely immune to criticals is huge safety. The fixed damage shielding stacks up with the Exoskeleton damage splitting to make you basically immune to most occurrences of spell damage, especially since you can't be crit. And to complete the set, Reactive Armor is an incredibly powerful effect, while limited to only weapon damage it is much much stronger than similar abilities given to light armour type classes, except you can use it with heavy plate. So you are also immune to weapon damage, and anything with scary weapon damage is taking bonus retaliation damage from your grenades, playing in to your "burst down everything immediately don't even worry about it" playstyle. Specializing in only weapon damage isn't a fair "drawback" to justify how powerful this is defensively when your other class features give you perfect coverage against non-weapon damage. The only thing that can kill you is instances of extremely high burst non-weapon damage that don't rely on criticals, for example watch out for the Vor Freeze+Meteor turn one combo. But that kills absolutely everyone right now and will probably be tuned back down. You have very strong self-renewing always on passive defenses, and Lightning Web as an even stronger completely broken temporary defense will almost never wear off before you have killed the poo poo out of 99% of the game's most dangerous obstacles because you also have absolute pinnacle burst damage. I waded straight in and facetanked the Room Of Death overpowered wyrms without having to use any inscriptions, did the same for the final bosses Elendar and Argoniel (though ate a death because I forgot that his new overtuned Disruption Shield will hit you for 8k damage a turn when you finally break it,) did the same for ATAMATHON which is absolutely ridiculous, and even the Hypostasis of Entropy which is the Cults endgame bonus boss. You're supposed to play that fight as a long haul, accumulating a vicious stacking damage over time as long as you're exposed to the vacuum of space, then moving to special temporary floor zones to remove the DoT from you but temporarily supercharging the boss. I could just ignore the intended mechanic and burst it down, then I ended the run by movement infusioning myself onto its last fired projectile that was still out after I killed it (did I mention you also have enough passive projectile slow that getting hit by projectiles is completely optional?) I had I think 31 stacks of the Total Collapse DoT. At the same time the Turret and Heavy Weapon trees are really undertuned right now. They need help, currently the class is 50/50 split between way too strong and way too weak. I really like the look and flow of the class, all the design elements are there, but it seriously needs not only a big amount of numbers tuning (turn down literally everything in Grenades, and if Lightning Web is going to continue scaling on block value it needs a harsh diminishing returns soft cap or something as a sanity check, seriously,) but a careful assessment of how many stacked up synergies it can bring to bear. I strongly suspect that some of it can't be fixed by just numbers fiddling without completely neutering the power of some abilities into unfun territory. I don't like unfun nerfs! I think in the end it'll turn out great, Raz is a great class designer, this one just needs a lot of feedback. My character vault file is here if you want a look https://te4.org/characters/5574/tome/2823e780-7d1c-498d-bd17-c47ddffdd39d every death is well earned beautiful example of me being a lazy turdlord, I had to work for them.
|
|
#
?
Nov 21, 2019 23:01
|
|
|
Fruits of the sea posted:Anybody played Children of Morta? Pretty charming game. I'm extraordinarily inept at 2d games of this type though. Is this something that can be brute forced or should I refund it? I really enjoyed Morta when I played it recently. I spent about 17 hours to beat it, but could have gone faster if I wasn't goofing around some. There are two tips I can give for the first boss. 1: Always have a dodge stocked for the pounce attack and 2: When she's hanging on the northern end of the arena she's totally vulnerable, so try and be there as she's coming down.
|
|
#
?
Nov 21, 2019 23:13
|
|
|
Lambo Trillrissian posted:-snip- Thanks for the writeup!
|
|
#
?
Nov 22, 2019 00:27
|
|
|
Metal Meltdown posted:I really enjoyed Morta when I played it recently. I spent about 17 hours to beat it, but could have gone faster if I wasn't goofing around some. There are two tips I can give for the first boss. 1: Always have a dodge stocked for the pounce attack and 2: When she's hanging on the northern end of the arena she's totally vulnerable, so try and be there as she's coming down. Eventually figured that out on my own, but thanks! Nice to know it's a relatively short, should hopefully be able to bludgeon through it. I have enough forever games on my plate as it is.
|
|
#
?
Nov 22, 2019 21:19
|
|
|
The Annihilator feedback is pretty accurate (albeit amusing when one of the top feedback posts on the forum atm is complaining they deal no damage). I've got a bunch of bugfixes for them coming up that also buff turrets just by making them less dumb, but will do balancing after that. In thanks, have a picture of the Annihilator pet because the I can't stop giggling at the angry face and tailgun.
|
|
#
?
Nov 22, 2019 22:07
|
|
|
I've been away from roguelikes for a bit, but I wanted to revisit Tangledeep with the upcoming DLC release. I'll be trying to relearn it tonight, so if you want to come yell at me about what I'm doing wrong, feel free: https://www.twitch.tv/goldplatedgames EDIT: Done for tonight! Man, what a good game. I've got the DLC (including the new one), and I'll be looking at that on Monday. Too Shy Guy fucked around with this message at 10:11 on Nov 23, 2019 |
|
#
?
Nov 23, 2019 06:01
|
|
|
|
| # ? May 23, 2024 17:32 |
|
|
How do you get new steam injectors in the base campaign of tome? I need a new steam implant so I can actually use both my sustains but I dunno how to get more since they aren't in shops like they are in EoR.
|
|
#
?
Nov 23, 2019 08:07
|
|