The first week of our aggressive experimental development cycle has been a challenging one. With bug fixes from the first release we only had a couple of days for the new build. Half of the features didn’t make it in which we have been working on over the last few days. We will put out a patch tomorrow that will add some point to playing on the experimental servers this week.
We’ve been collecting performance data from people on the city build, and as expected we have a lot of optimization to do there.
Experimental Content Update 3 Plans
This Friday we are heading back to the land of survival. We will be introducing some new blueprint mechanics, doing some more work on our procedural map workflow to hopefully make the map much more interesting (and big, and improve performance). Our aim is to start tweaking the balance of full loot so that its not so punishing as you progress further into the game.
Something new I tried with the current map is adding more negative space to the playable area by adding very large wasteland type areas around focal points that require vehicles to traverse. As there weren’t really much reason to move between cities, these didn’t serve much purpose.
I did however end up pushing our tech a bit further with support for very large maps. The current nullius is 12km x 12km, I am going to keep this size for week 3, and add back in some of the classic Hurtworld biomes as smaller chunks among larger wasteland.
What an epic week! So we bit off a bit more than we could chew this week, but we still got some stuff out there to test with the community. It was an epic push on Friday to get as much of our stuff into the test map to try and see how it runs. While we saw a lot of problems with the build we made, we learnt a lot from it. This was the idea. I learned that my cities are waaaaaay too big. While we will keep the big city asset in case we want to use it later, we may be removing a lot of height from it, by knocking out the filler floors. We quickly found that running up all those stairs was just so tedious. Not necessarily the running up, but definitely the running back down. Also the size horizontally may be too much as well. I had to do a specular & gloss map for the city textures. This is no small feat. The Albedo (Colour) map is 259 layers by itself. The Specular & Gloss texture is that but x 2 so 520-ish layers for that one. Man what a mission. The reason for this is because the city runs on 4096 textures, to improve performance and use less draw-calls. This was quite a slog to get everything not only looking right but looking the same. Not only are there these mega-layered textures, but parts of the city have tiled versions of the texture within the 4096 texture sheets but as separate textures. These were created to save poly counts on places like the outer windows on the building exteriors. These also had to look the same as the other textures. Example of the Cities’ 4096 textures below.
I’m also progressing nicely on the AWM Sniper Rifle. Top is the High Poly, Bottom is the Low. I really like this gun. *drool
This week I’ve been working on bugfixes and our experimental city survival map type.
Unfortunately we bit off a bit more than we could chew trying to get a new game mode happening in about 2 days and we weren’t able to get everything in that we’d planned leaving it feeling a bit empty and aimless.
We’ve got an update coming out very soon which should start to address some of these issues, the compass is working now (see below) and will show the locations of all machines within 2km. Gear machines sell gear crates, gun machines sell gun crates and ATMs allow you to deposit and withdraw scrash (although the ATMs have a few bugs still to address).
I’ve also given the guns a balance pass cranking up the power of the ar15, fr18 and shotgun, I’m interested to hear what you think of them now.
We probably won’t do too much more work on this gamemode this week, just whatever things will port over to the main game like balance changes.
I’ve also been working on a bunch of ItemV2 changes and bugfixes including changing structures with an attachment dependency (like a door or a window) to depend on its attachment even when the construction cell is claimed. We made this change because we think bases constructed this way look odd, don’t make sense and we don’t want such an unintuitive tactic becoming the dominant strategy.
Another change we made was reducing the maximum field of view we support from 100 to 80. We made this change because we think 100 vertical field of view is excessive for a standard 16:9 setup and it was causing us issues with people being able to see through walls with their fov cranked right up. Rather than make the character controller fatter or bring the camera’s near culling plane in closer (which would reduce the precision of the depth buffer causing more shadow flickering problems) we thought this was the best solution.
I’ve been seeing reports however that this isn’t working so well for our players on ultra-wide screen setups so we’ll look at switching it over to a horizontal field of view picker and limiting it this way.
This week I’m going to start working on the performance issues, continue to build on our loot trees and item rolling and keep working through the backlog of bugs and changes.
I also plan to patch some exploits that have popped up in our main branch and roll this update into the experimental map update at the end of the week.
Hey folks. This week I’ve been searching for memory leaks, getting the city experiment out, and looking into changing the road workflow into something less destructive.
Firstly we had some nasty memory leaks on the server that needed a look. It turned out we weren’t pooling some high frequency objects correctly, namely the paths that creatures used to navigate the world. So every time you chased a creature, the server would leak. This was an easy fix at least. The second biggest leak was items, which Spence made some good optimisations for.
We got the city out to you all on Friday, and I had a good play Saturday. Definitely a long way to go, but an interesting experiment that definitely stress tested our pipeline and showed the places that needed improving. One of these places is with elements of our map creation process being destructive, especially with the road tool. To mitigate this I’ve started putting together a non-destructive workflow where the road tool captures snapshots of an area, and only stores deltas when it makes changes. With this, we can let the road tool mess around with the terrain as much as we want without worrying about having to regenerate the map all over again. Check out a gif of it in action above.
Pushed out a lot of assets this week trying to get all the assets we need for these experimental game modes. Balancing fast production and assets that don’t look awful can be difficult. The funnest things this week were definitely the Utility box and ATM machine. The Utility box was/is going to be used as a minable resource node in the city, so it had to stand out while also fitting in. I’m terrible at any sort of Graffiti, so the graffiti was done by Mills and is all Hurtworld related which is cool. The ATM machine (and slot machines not shown) were needed straight away due to it being the night of the build so we agreed a back alley make believe style ATM would be a great fit, the slot machines were done in the same style but if they are going to become a permanent part of the game I would like to re-do them as proper slot machines.
Next on the list was making a new loot crate that looked like the old one but was much lower in poly count. The original box was about 1200 triangles while the new box is a mere 12 triangles and with the help of a normal map comes out looking pretty similar even with 6 flat faces. I also made a small scrap metal pile world item, it kind of looks weird when it’s upside down but it’s not too bad, I would like to touch up the specular map a bit.
I finally got to something that’s been on my to-do for a while now, re-UVing the military vest. The military vest was one of the first assets I made and I was still in the habit of optimizing every single bit of UV space for mobile platforms, this creating horrible symmetry on the camo patterns, so I’ve finally gotten round to fixing it and I’m much happier with the result.