PC Developer Discussions: Alpha 17

Developer Discussions: Alpha 17

  • Newly Updated

    Votes: 1 100.0%
  • Check out the newest reveals by Madmole

    Votes: 0 0.0%
  • Over 100 new perk books with set collecting and bonuses

    Votes: 0 0.0%

  • Total voters
    1
Status
Not open for further replies.
Its not like we don't want to support large or infinite maps, but lets be realistic. A very high quality high performing map beats an infinite or very large and ugly slow performing map. I prefer smaller maps because it takes hours of real time to meet up with your friends or hours between frags which is boring for pvp.
Seeming most starts used to be inside of 2.5k the longest it would take to meet up with someone would be about 15 minutes not hours. (If you move with a purpose)

As far as boring for PvP, that what WAS great about PvP, even if you started a server on day 100 you could travel far out and hide until you catch up enough to defend yourself against the leveled players. Reducing the map size only gives less places to loot, and makes it harder for players to progress and formulate defensive and offensive strategies.

Couple that along with the bonus xp for all teams and you effectively shun any solo player who wants to PvP. Whats the point, everyone's base is only a short jog or bike ride away from everyone else. (8k map)

That team of 5 who level twice as fast, unlock all content twice as fast and have you outgunned by 500% will find you and wreck you no matter what before you have the chance to get on your feet. Just saying smaller maps without the option to have what we had before does nothing for PvP. (I guess maybe you can meet up 2 minutes faster on start, big advantage)

 
Still need to load a massive height map into the game then to get the geography, roads, biomes... so you mean exclude potato pc owners from being able to join your server? Its not that simple, people EXPECT it to work if you make that option and the ram needed to generate a map like that is more than most people have (at least currently). Anyhow this stuff is above my pay grade. I'm not saying no, but we're not focused on size right now, but bugs and performance. Without the latter the prior is useless. How happy will be people be if I said "heres your infinite map with floating pois, moses parted seas, roads that run into cliffs", etc? We're fixing bugs, optimizing map gen time and making it pretty, THEN we can look into size. Priorities.
A hybrid approach to map caching would be huge. In A16 the map generated as you explored it. I totally understand the thought process and value to precaching the entire map for A17. However, it has led us down a path where you CAN'T really support big maps for the reason you mentioned.

The solution is to pre-cache the map up to 8k x 8k or whatever, and then allow for the map to generate itself beyond those limits via A16's framework.

 
Your preference to one alpha over another has what to do with map sizes?
Random gen was complete ♥♥♥♥ in a16, just in different ways from what is currently in a17. No real caves. No rivers. Random towns that made no sense. The list is endless.

 
Random gen was complete ♥♥♥♥ in a16, just in different ways from what is currently in a17. No real caves. No rivers. Random towns that made no sense. The list is endless.
Every randomgen has been better than the previous one so far. A17 RWG is questionable, but on its way to be a better one.

 
The traders are great in the game. They allow you to buy and sell stuff which gives you more options. Because no one wants a boatload of useless junk, handguns for example. After going on a few raids though buildings, I ended up with a chest full of them. Thankfully they will buy a few daily.

But the best thing is that they will sometimes give you cash for free. Yes that's right, free money. How you ask? Well I go into there shop and sometimes find a cash register loaded with Dukes and bank notes (cash) which I promptly empty and sell the bank notes back for more Dukes. Nice. All this right in front of the trader.

Now I got more Dukes to buy stuff from said trader, nice, thanks buddy!

Too bad this doesn't work in real life. :)

 
Every randomgen has been better than the previous one so far. A17 RWG is questionable, but on its way to be a better one.
I have to disagree. The A17 road system is a giant leap backwards. That's not what I dislike about the A17 system though, better roads are just a matter of time and effort.

My concern is that the underlying system is difficult to scale without trying to pass too much data from the server to the client and trying to generate too much land at once leads to ridiculous generation times for maps.

I can see the benefits of using static heightmaps, you could play using the layout of the land from wherever you live by making a heightmap from freely available data which could be fun. I'd prefer if we were given the choice though, either use a static heightmap or generate it all at runtime.

 
I have to disagree. The A17 road system is a giant leap backwards. That's not what I dislike about the A17 system though, better roads are just a matter of time and effort.
My concern is that the underlying system is difficult to scale without trying to pass too much data from the server to the client and trying to generate too much land at once leads to ridiculous generation times for maps.

I can see the benefits of using static heightmaps, you could play using the layout of the land from wherever you live by making a heightmap from freely available data which could be fun. I'd prefer if we were given the choice though, either use a static heightmap or generate it all at runtime.
I like your point. And having a choise between these 2 ways ot generatin terrain would be more than awesome. Imagine the possibilities.

 
The OP didn't consider 8K to be a large map. He was talking about 16K maps (or larger). Ship five maps like that and you've just added an extra 2Gb to the game - which half the players would probably never use.
(Edit: I guess technically five maps is about 2.5Gb since each 16K map is roughly half a gig. Also, just for reference, the entire 7D2D game is under 8 gigs now.)
Fair enough i did mean 16X16 maps, but who cares if even another 5gb is added to the game? If you think logically another 10gbs would literally mean nothing as most games releasing of considerable size are 50+ gigs. Imo id much rather have a select few “large” maps thats tested for quality and that has a sense of epicness to them and detail, then the crappy super small RWG maps thats created on the spot that we have currently have. Just my thought.

 
If possible i would love a fixed 2x2 km center that you can take with you to the next seed
I hate to be the one to break this to you but that would probably look horrendous.

Even a minor change to a noise algorithm can completely the land that's generated. Trying to smooth the terrain between the transported land and everything else is damn near close to impossible. You'll end up with the edge between the old part and the new part covered in cliffs and valleys, in the same way that the terrain generator used to break when you loaded a map from a previous build.

 
Its not like we don't want to support large or infinite maps, but lets be realistic. A very high quality high performing map beats an infinite or very large and ugly slow performing map. I prefer smaller maps because it takes hours of real time to meet up with your friends or hours between frags which is boring for pvp.
Very much disagree. the 17.2 smaller world is rubbish. Sorry to say. An island really like every other game that is out there that people lose interest in after 5 minutes of play.

if the small map as you say be great for pvp have that as an option in the game menu. dont restrict your players choices because of your own play style. I would rather a huge map then a as you say high performing map it def does not beat a infinite or large map. your play style seems a very very small minority and limits so many things a large world gives.

 
I hate to be the one to break this to you but that would probably look horrendous.
Even a minor change to a noise algorithm can completely the land that's generated. Trying to smooth the terrain between the transported land and everything else is damn near close to impossible. You'll end up with the edge between the old part and the new part covered in cliffs and valleys, in the same way that the terrain generator used to break when you loaded a map from a previous build.
I don't know what algorithm 7dtd uses, but at least perlin noise could support what Royal is suggesting. All you'd have to do randomly reset the parts of the gradient grid outside the 2x2km area and then regenerate the land using those. The algorithm will automatically interpolate the old terrain and the new terrain.

 
I don't know what algorithm 7dtd uses, but at least perlin noise could support what Royal is suggesting. All you'd have to do randomly reset the parts of the gradient grid outside the 2x2km area and then regenerate the land using those. The algorithm will automatically interpolate the old terrain and the new terrain.
There's ridged-multi-fractal in the mix along with others. Smoothing the transition isn't impossible, I never said it was. The problem is that the more smoothing you do the more detail you lose. Finding the right balance that will blend the bottom of a valley from algorithm A into the top of a mountain from algorithm B without smearing the terrain so much it looks like it's made of warm butter will be difficult.

Not impossible but it's likely to look horrible,

 
Every randomgen has been better than the previous one so far. A17 RWG is questionable, but on its way to be a better one.
I never said it wasn’t getting better. I know it is. What started this was people complaining that large maps are more important than a working game.

 
There's ridged-multi-fractal in the mix along with others. Smoothing the transition isn't impossible, I never said it was. The problem is that the more smoothing you do the more detail you lose. Finding the right balance that will blend the bottom of a valley from algorithm A into the top of a mountain from algorithm B without smearing the terrain so much it looks like it's made of warm butter will be difficult.
Not impossible but it's likely to look horrible,
I don't feel like smoothing algorithm is required though. If you copy the gradient data of the higher octaves, doesn't the small-scale detail get preserved?

EDIT: Alternatively you could interpolate each octave at a time, starting with the lowest and going up.

 
Last edited by a moderator:
I'm still having issues with vehicles. While these issues didn't show up when riding the bicycle.

They seem more present when riding the mini bike and the motorcycle.

Things like the game crashing or jumping to random locations on the map and getting stuck on the motorcycle, sometimes still moving on it's own. These seem to happen when riding for several minutes.

The bicycle moves faster when you press the shift key but this also works with both the mini bike and the motorcycle.

I am not sure if that makes it worse or not.

Has anyone else noticed this?

 
I don't feel like smoothing algorithm is required though. If you copy the gradient data of the higher octaves, doesn't the small-scale detail get preserved?
EDIT: Alternatively you could interpolate each octave at a time, starting with the lowest and going up.
That would be a hell of a big job for a questionable feature.

There's more than a single perlin noise algorithm at play for any given point on the map. You're not only mixing the octaves of however many perlin noise there are, you'd also have to mix the octaves of however many multi-fractals, billows, and possibly voronoi that you're using. This must also take into account any operators performed while mixing the original algorithms.

This also assumes that both server owners are running exactly the same mix of noise algorithms.

 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top