faatal

They are doing preset parts, so you would pick a head and body and then vary some aspects like hair. Roads are not planned to be changed unless we stumble on something or need to fix a bug. I did speed up road gen speeds about 3x, so I can now make an 8k world with default settings in <50 seconds. I do have a fast PC. Looking better, but running worse is often how it goes. In this case it may cost a few FPS, but it is not a final tweaked feature yet. The lower quality shadow settings may have contact shadows disabled.

I have not posted in a while, so here is some info for January: RWG is getting some improvements for A22. Alloc and I have reduced memory use a huge amount. I have also been working on performance with stamping now many times faster. I plan on looking at road generation speed next as is the slowest system now. Better messaging. Some fixes for seeds not being consistent and possibly other bugs like biomes not matching percents. I started prototyping sleeper volumes having a simple embedded scripting system for more control over spawning. Our gfx programmer is adding new hair shaders, contact shadows and ambient occlusion. The player profile editor has been updated for the new player models. Console features are getting more functional and refined each week.

No one currently. May not be changed much for A22.

Development is not slower. Releasing information is slower as it is not wise saying X is coming at the beginning of an alpha, only to find months in that there was some big problem delaying the release. As we get closer to release, then features should firm up to where we can say here is a list of X, Y, Z that should be in A22. A good example is DLSS support. I have it kind of working for Windows, BUT it has a variety of problems working with our multiple cameras and postprocessing steps, so there is a lot of trial and error getting it finished. As I don't know if this will be shippable with A22 or ever, I would not want it on any list of coming features yet. That appears to be terrain, if so, then that is a core issue with how terrain works, so it may not be something that changes. I have seen that and have some prototype shader changes I tried many months ago, but Unity was being a punk and it was just not working, so I put it on the back burner.

Nice to hear. There are still more coming. A recent one I did, is block entities (props) spawning from a chunk can now happen over multiple frames which helps with tall buildings trying to spawn many hundreds of them in one frame leading to a big FPS frame spike.

I don't know. I guess both.

Something has to be at the bottom of the list.

That looks interesting, but their FSR 2 asset had issues in 7dtd with motion blur that require rendering/shader changes that we have not yet tried to do. We have more important rendering issues we are working on now. Enabling our old block AO for A22 required me to write a light update manager and change how we apply AO as it was slow with the much larger amount of props we have in the world now. Sure, it might be possible, but we are working more on improving our general performance, not finding ways to slow it down Excluding the new character outfit and animal redesigns as that is art requirements to get those up to the standards of zombies.

We don't use FindObjectsOfType in any code requiring performance, but yes there are some tools and init code that use it. We already do object and data pooling. Our object pool is more advanced than Unity's. Vector math is even faster if you change the individual variables directly (x y z), which is what we do in our higher frequency code. Thanks

Not planned for A22 as was basically said in that video.

There is a lot of work to do.

I was satisfied, so I stopped messing with it. I do tweak stuff for a good amount of time and generally have to make myself stop and call it good enough. It is basically a few numbers, but the problem with changing tuned numbers is you often unleash some other bad edge cases. 1 Last thing was some Vector3i and 2i inlining optimizations. Have a few similar ones in chunk code for next. 2 Sure. Probably a few times a week.

It would seem those type of D3D errors are typically either Unity or gfx drivers. Alpha 22 should be on a newer version of Unity, which may fix the crash.

Unity's wheel colliders are fairly basic. They are not a racing game simulator. Two wheeled Unity vehicles fall right over as soon as you turn. The wobbling is simply vehicle balancing forces keeping you upright otherwise you would be laying on your side most of the time. Unity vehicles in general would be upside down or on their side a fair bit of the time if not for the side and front/back forces pushing the vehicle upright.

...I could change the protection size or add new ones like the placement protection and not have to care that all your chunks have the wrong data in them. The seed is always 0, so you would get the same result each time. Count changes would give you more or less of that same result.

respawndelay is parsed as a float and multiplied by 24000 to convert to a world time delay. World time is a long, so the delay being an int is correct. The x24000 gives it plenty of precision. I think it is a 4 piece system with mods. Two other programmers are working on it, not me, so I don't know the details.

It adds 2 at level 1 and that number goes up to 10 at level 5 like so: <passive_effect name="LootProb" operation="perc_add" level="1,5" value="2,10" tags="sledgeSkill"/> There are some that only have 3 or 4 levels. The base value is 1, so if there were 10 items in the list, each being 1 they would add up to 10, with each being a 10% chance. Adding 1 would make it a 2 out of 11, so now a 18% chance and so on. I think there are 23 magazines in the full list. Your sample size of you vs them is not large, so randomness could be a big factor in what you saw.

Yes, I generally do, but I have not been keeping up with this thread for about a month now, as I am busy elsewhere. Some topics I am not going to say much about and tend to skip over, like art style or design directions as I don't make those decisions or contribute much to them. I may complain about them like anyone else on the team but those are not my call.

Trader area data is no longer stored inside chunks. It calculates if a position or bounds overlaps a trader when one of those functions gets called by using each trader bounds. That is the dynamic part. Not pre storing it in chunks.

Maybe. That is way old and I was not even a player of the game back then.

Problematic. When you turn on the 3rd person body, the camera is inside the head, so you see edges of your face and eyes, which is ugly.

For example, if we say we want these 3 features in A22, then yes, we would have to take whatever time is required to complete those 3 features to a releasable quality level. In the past it was dozens of feature changes, each with varied dev times leading to devs with shorter tasks switching to more features, further extending time and making for long releases.

That just adds complexity and we want food to be consumed with whatever you are doing in the game. What features will be released in any specific alpha is not a given. A22 is planned to be short duration with targeted features.

I am not sure how many places in code/xml would need to change to just get a stub.

Yes. The plan for A22 is to be much shorter than A21, which means there would be less room for 21.x updates.