Developing Multitile Creatures in Roguelikes | Hacker News Reader

Developing Multitile Creatures in Roguelikes

gridsagegames.com

142 pointstodsacerdoti30 comments

The entire dev blog [0] here is really great for anyone interested in game dev, roguelikes, graphics, etc. etc. Lots of in depth information and great images/gifs. Such a nice resource and to give to a community of small game developers, lots on there on my reading list!

(It is from the developer of Cogmind, Josh Ge, who also has a nice video [2] on their journey to becoming a game developer. I love how the game looks, but haven't yet tried it.)

[0] https://www.gridsagegames.com/blog/

[1] https://www.gridsagegames.com/cogmind/

[2] https://www.youtube.com/watch?v=9yJflbg0V38

I am a huge fan of roguelikes and I casually develop one of my own for fun (in a very very early stage of development) so anything like this instantly gets my upvote.

This article is super sweet to see and the cogmind creator writes more about development for Roguelikes more than anyone else I know. Truly a wonderful resource in the community.

I am however surprised by the omission of what I consider to be an obvious example of multi-tile creatures which would be vehicles in Cataclysm: DDA

I suppose a vehicle is not a creature, but I think that for all intents and purposes, that the implementation details will be very similar. I think it's also possible to get AI controlled vehicles which are functionally very very similar to a creature.

Of course, he may have left out Cataclysm DDA because its implementation is funky. You'll notice this if you take a pickup truck and turn in diagonal, suddenly opening up half of your vehicle to zombies despite it being closed when the vehicle oriented in any of the four standard directions (up, down, left, right)

The article did reference CDDA vehicles in a list item near the beginning of the "Pathfinding" heading.

I've been waiting for CDDA to support multi-tile creatures for ages, but that's likely to be some time coming as there isn't much in the way of common logic between creature and vehicle entities. :(

Wow! Yup! I totally missed it! My apologies!

Pretty thorough but with one disappointing gap (disappointing for me at least). While it does cover my game Dumuzid (probably the first roguelike focused around multitile monsters) it skips over another I made a couple years later which is a little more notable for placing 1st in 7DRL 2018: DEAD FACE (a turn based bullet hell with hacking and multitile mechs). https://jere.itch.io/dead-face

Considering I made each of these games in a week I don't think making such mechanics is particularly difficult. At least not technically. They do require some new kinds of design challenges (how to deal with getting stuck, what's actually interesting about big enemies, representing tiles visually). It's a mechanic with a lot of unrealized potential.

If you dig this kind of content, they talk a lot about roguelike design + development on the podcast Roguelike Radio: http://www.roguelikeradio.com/

Handling a single entity (eg, a city) that occupies multiple tiles is a classic problem in grid-based games which blows up a lot of naive approaches, it's cool to see people writing about very specific issues like this.

Is there a TL;DR of why this is so tricky?

The article talks about several issues, such as:

- Pathfinding when taking up more than one tile

- Area-of-effect (is damage multiplied per tile?)

- Collisions (particularly if the entity is not square/rectangle)

- Visibility

Honestly none of those sounds particularly challenging. The hardest is probably the path finder but even that is just 1 tile patfinder using a corner of your square except now you need to check more squares if you are able to move in a direction. But combined together, those and many other small code & design issues add up and becomes a pain

Imo the reason why roguelikes dont add them is simply because it adds very little value comparable to head aches it can cause

"Honestly none of those sounds particularly challenging"

Individually? Sure. Together? Kinda rough. Interacting with every other system you've ever put into your game and with every other system you ever will? Yeah, it can be challenging.

I think the point about things visually looking bigger than you is important. DCSS (tiles) has the kraken's as pointed out, but there's also the lords of pandemonium. They're just one tile, but they're about twice as tall at as a normal unit, and visually appear much larger than anything you've seen in the first 95% of the game. I thought that was really effective and obviously the simplest.

Here are some of the basic minor lords. They're the ones whose heads are a little taller than their own square.

https://imgur.com/N4a9VaN

> I think the point about things visually looking bigger than you is important.

Initially, I dismissed this when building my remote D&D app [0] but it's been a persistent question from beta testers. It's the sort of thing that sounds trivial, but isn't - especially if you start with the 'everything is a unit on a grid' assumption.

I still haven't implemented it, but this article has given me some ideas on how I could.

[0] https://tableofsending.com/

This looks cool, but I'd rethink the name. It's clever if you're in the know, but as someone who's pretty in the know, I didn't immediately assume that this is an rpg kind of thing from the word "sending".

I also kept reading your abbreviations as Terms of Service which has a funny but icky implication.

Thank you! That's another bit of consistent user feedback! :)

And I agree. I'm rebranding it for the move from MVP hack -> v1.0 product.

Do you already have the new brand?

You might consider "scrying": "scryingtable.io" or "scryingmap.com" or "scrymap.com" or "scryi.ng" are all available.

Is DCSS any good these days? I remember playing it back aruond 2014 (0.14-0.15?) and beating the game a few times, but it felt like each update just increasingly dumbed it down

If you thought it was dumbed down in 2014, you probably will still prefer another roguelike.

Considering they're removing food completely yea. I do think most of the changes are thoughtful and make sense but sanding all the edges off makes the game seem a little bland these days.

I mostly play brogue currently but I've been getting into angband variants. Sil is probably my favorite of those so far

I was actively playing DCSS at that time, beat the game with every race and every class, did 15 rune and all and I don't think it was getting dumbed down. A lot of pointless stuff was removed but the tactical depth of the game was preserved quite well.

Hm, maybe I should boot it up again and see what's been changed.

I still really enjoy it. I feel like one of their guiding principles is "if doing something tedious or annoying would make you more successful, let's remove that". So, for instance, instead of having to take notes about what's in your stash, you can just search all items you've ever seen and where they're located.

Some people like(d) DCSS because it was a complicated system with lots of entropy that sort of gave an illusion of complicated decisions. A lot of recent design decisions have been around removing decisions that aren't really that interesting if you know what you're doing.

There's merit to both views. I don't have a strong opinion one way or the other. The biggest change recently was removing food. On one hand, I've never finished the game without a huge surplus of rations. Food did not affect the game, almost ever, even for trolls and spriggans so long as you were not overly cautious about eating food when needed. Once you understood this, food didn't really serve any purpose in the game for 99.9% of game time. On the other hand, newbies might think its interesting, and overindex on, say, spellcasting to reduce hunger costs and hoarding rations and think its a fun added challenge.

Looks like this article is written by the creator of Cogmind. That game has been on my steam wishlist for a while but I haven't actually tried it yet.

Go for it. I say it's worth the sticker price, easily. Incredibly deep, lots of replay value. The number of builds/strategies in this game are astounding. And it's very actively developed, with the developer being very involved in the community.

Do go for it. Definitely worth it. Also once you've tried it far enough, try out the different "options"/"mods" to play.

I really like reading his writeups on roguelike developments. I also play Cogmind and it's been a timesink too. With each major update, he also introduces a "mod" option that lets you play the game in different ways and that refreshes the game so well.

For example, Cogmind is a game in which you blow up other robots and take their parts in as yours. Each time you would go up a level, you would be able to put points into different categories that let you accomodate more parts, etc. One time however, the game added a different way of playing that lets you "buy" upgrades as you level up (level up as in going up a level). That made the game quite different and I had fun playing it all over again.

This is reminding me of a couple of old tile-based RPGs from the 90s - Realmz and Exile 2/3 - that both did multitile creatures. I do think that there was a bit of a lull between then and the more recent resurgence of (tile-based) roguelikes

Thank you for mentioning Realmz. I played a demo sometime around '97 and have been recently been hoping to find it again, but could never remember the name of the game or developer. I'm glad my search is finally over - looks like one of my projects for this weekend will be finding an OS 9 emulator.

Your best bets are SheepShaver and BasiliskII. The former for emulating PPC Macs and the latter for 68k. The best resource for setting these things up is the e-maculation wiki and discussion forums [1].

[1] https://www.emaculation.com/

Reply on news.ycombinator.com