qolelis on 15/11/2017 at 09:00

---

It all started with the idea of my game character being able to control physics in some places, so why not also being able to rebuild certain objects (to some extent)? Then I unfortunately started thinking about maybe doing it using boolean operations and allow more general rebuilding. Had I known when I started how complex that really is, I might have skipped it, but I'm apparently insane, so here I am...



It is a nice type of butter knife, isn't it!? Although, there is something with metal that makes spreading the butter harder than with a wooden butter knife. I think it's the lower friction of metal compared to wood that makes the butter not cling well enough to the blade, so the blade sort of just slides over the butter (or the butter slips off and ends up not on the bread, but on the floor)... Anyway, butter knife physics aside, I think I got it from my grandparents a long time ago.



Yeah, I did some research before I started to see what had already been done and couldn't find any that didn't need manual adjustments afterwards (like the one in Blender at the time). In modelling software that is sort of okay, since the modeller can fix it afterwards and the player will never see any of it, but in my case, the modeller is also the player, so it has to be perfect. I don't know if what I have is better than anything else, although I guess it is pretty good -- but that probably just means I haven't yet tested the cases where it fails miserably.



There sort of isn't any new geometry, I'm just combining what's already there (and also cut triangles that need to be cut and re-triangulate the polygons (which sometimes have holes in them) and so on and so forth...), which is hard enough, but Unreal does a few things to help: I can for example access all triangles of a mesh without having to care about how it was modelled (as far as I know, Unreal converts any quads to triangles). My code is made to work in Unreal, so it might not work at all in other situations.

The main problem is that there are so many cases to handle and it's all so finicky (even the slightest change can mess things up). At the moment I have around 20 test cases of which 3 or 4 are failing, which means one or two triangles disappearing or being in the wrong place. The errors aren't huge, but I can't afford anything going wrong. Sometimes I despair, but, on the other hand, I have come pretty far.

After I had already been at it for a few months I learned that this is something people (as in solo developers) are always adviced not to even try, so maybe it was a bad idea. On the other hand, even if I have to scrap everything, I have gotten a lot of practice (both math-wise and code-wise).



In which way is it "complicated and rigidly defined"? Does it, for example, work only on a specific set of objects (like, let's say, only cubes and spheres)?

qolelis on 16/11/2017 at 22:35

---

Well, if that's the case, my thing might actually be better -- but it's still not good enough for in-game use. I'll keep on working on it, though, until it either is or I reach a point where I have to shelf it.

I just checked Blender's boolean tool again and while it works for basic shapes, it can't handle more complex shapes (whole sets of surfaces disappear or nothing happens at all), so maybe I am on to something after all.

Neb on 17/11/2017 at 01:51

---

I've started making a character-driven cyberpunk strategy game. Nothing visual to show off yet, because I'm screwing around with data structures and working out what I'm going to need (a tonne of agents). Probably won't make much progress until the middle of next year. I think by then, and after a bunch of prototypes, I'll know whether it'll be worth it - because what I'm aiming for is incredibly ambitious.

Setting: Medium-term future in a cluster of coastal megacity states. All urban territory.
What it won't be: City builder, economic/management sim, war game.
Agents: Characters with their own personality stats - RPG. Start by generating your own (pick ideologies, vices, etc). Create and control an organisation: Political party, religion (UFO cult, cybernetic, etc), corporation (nothing too dry, variety of relevant types). Multiple characters inhabit each organisation.
Core gameplay loop: Favours for other characters with mutual interests will earn you 'tokens' - like collecting cards in a card game - which represent favours they can give in return (ie, an intelligence org can dig up dirt on someone for blackmail, etc). These can be spent during turn-based encounters. Encounters include things like court battles, civil disorder, election campaigns, cyberspace actions etc. I want to make them punchy pay-offs.
Dynamic factions: Characters align based on mutual interest. Factions tend to form dynamically within - and between - organisations. This creates conflict and drives events. Orgs can have their functioning affected by infighting.
Interface: Main city view with territories. Graph view visualises orgs, characters, and their relationships [(https://media.licdn.com/mpr/mpr/p/1/005/0b5/027/14c729e.png) rough ballpark].
Politics: Playing around with the idea of political system with separation of powers - executive, judiciary, legislative. Laws to be nudged. Gerrymandering districts. Civil society movements. Cyber-sovereignty. All the way to the wacky stuff like AI judges, Jonestown style cults that upload themselves to cyberspace, etc. Balance between sci-fi and authentic politics.

My notes have way more detail. The aim is to make some core loop prototypes in a small world to test whether I can make something fun, and then scale the complexity from there. See where it goes.

Nameless Voice on 23/11/2017 at 01:41

---

UE4 is great.
All of its tools and subsystem editors are really powerful, well-thought-out, and user-friendly.
I especially like the Blueprints visual scripting language, which you can use to make your entire game logic instead of writing C++ code, if you are so inclined.

I believe Unity is more popular with indie devs because it was first, and also UE4 initially wasn't free.


Caveat: I haven't used Unity myself, so I can't say how good it is.

Nameless Voice on 23/11/2017 at 11:34

---

Either way, it seems to me that it really makes no sense not to use one of the professorial engines to build indie games these days, especially 3D ones.
Why reinvent the wheel?