Відео

That's the plan! But thanks, much appreciated!

Those aren't pointers. They are .NET handles, for interop with the editor app which is written in C#. Looky here: en.wikipedia.org/wiki/C%2B%2B/CLI

Personal preference. Zig is low-level and powerful, but modern and comfortable to work with. I like it a lot!

Thanks, you as well!

You are very welcome!

Making videos takes work --- bro made a full game engine, making all the fancy animations to model a concept to newbies takes a lot of extra effort he could direct to just building a cool game --- Lague just makes videos, it's a different objective

I have the utmost respect for Lague and what he does, but it's safe to say our focus is quite different. He makes very cool bite-sized projects exploring a certain concept, and then makes highly polished and entertaining videos about them. I "just" build a game and make occasional videos about the process. But my game project is much larger, and takes up most of my time if I ever want to finish it. Sometimes I put more effort into my videos such as the two parts on video playback in Basis (ua-cam.com/video/zBhhckL5Q4k/v-deo.html and ua-cam.com/video/cgFCS9t41vU/v-deo.html) but they take a long time to make. In the end, my "product" is the game, not the videos.

Yeah, that's why i added "if you're serious about UA-cam" part, I didn't know your priorities.

Thank you! Yeah, it's coming together quite nicely. Honestly, most of this is stuff I have come up with myself, as I wasn't able to find much info on suitable vehicle AI online either. I basically started small and tried to identify issues with the approach, and then tried to solve them one after another. That's why the driving AI has a bunch of sub-systems, many of which tackle one particular issue or case, as you might be able to see in the code section at the end of the video. Is there anything in particular you struggled with? I don't know how Unreal expects/wants you to set things up, but describing the whole system is obviously too large of a thing to do in a YT comment. Maybe I could dedicate a video going through the AI code, if there is enough interest. I've also toyed with the idea of open-sourcing parts of the game code, and AI could be one such part. Obviously it wouldn't build without the rest of the code and engine, but it might be of interest to other people as a reference. Finally, I know that Mercuna (mercuna.com/ground-navigation/) has some middleware which is supposed to handle driving AI quite well, in UE5.

Absolutely! The older I get the less I feel the need to write "clever" code, favoring dead-simple stuff instead. Many times it ends up being more performant too...

Yeah, this is comes from the local offset applied to the predictor, and it's especially noticeable with the Mother vehicle. You can see the offsets in the AI profile JSON data at around 11:30, at the bottom, where the mother vehicle pushes the detection forward and backward. If you don't offset it at all it lines up with the transform of the vehicle, but then most of the points will be inside the vehicle and of no real use. On the other hand, predicting far into the future tends to be unreliable. I've toyed with the idea of offsetting left/right or perhaps rotating the predictor based on steering value or angular velocity, but we'll see if that is necessary.

@@MadrigalGames why not keep your current offset system but do your prediction after the offset instead of before? right now you have a predicted path of what seems to be the center but offset to the front, what if you instead had a predicted path of the offset position? immediate problems / differences would be that the points could become significantly further apart when large vehicles are turning, and the predicted path would be more accurate to the real path than it is currently, which the looseness of your current system may be helping the pathfinding as it rejects dangerous paths faster by overshooting turns

Your initial comment got me thinking. What I might want to do is have a set of offsets, per AI profile, eg. center, left and right, specified in local space. These would represent the front/back bumper of the vehicle essentially. Then I would predict without any offset and apply those offsets to the predicted transforms. A potential problem is that every predicted transform would have to make 3 nav mesh queries instead of one, but that might be okay. I probably need to test that out. Thanks for the tips!

@@MadrigalGames unrelatedly you should consider making a discord server or something similar for discussion of the game

Yeah, the plan is to set up a Discord server when the demo release gets closer.

Thanks!

Thanks!

It's about driving, exploration, puzzle-solving and generally messing around :)

@@MadrigalGames uuuh

Probably could use @TypeOf() in a few places yeah, but I find it makes some interfaces easier to use if they take a comptime T parameter, and an instance (or pointer) of T, rather than taking an anytype. Often, in the case of anytype I have to look at the implementation to see what I was thinking at the time. Personal preference, I know... And yeah, the exclusion list is for things you don't want it to poke at.

@@MadrigalGames Was more so thinking it can reduce potential for user error, but maybe the compiler would catch it anyways. Not very familiar with zig. Cool to see you stuck with it though, I only recently realized how cool it is. Same for some of cpp..

Yep, it's Recast and Detour.

Hi Juan, thanks! Let's see if I can answer all of your questions in order. The NavMesh doesn't take any initiative over any gameplay event or system. It just sits there and answer questions like "Can a car drive here?" or "What is the path from here to there?". The physics system also doesn't technically drive the vehicle. It just simulates how the vehicle moves when certain inputs are applied, such as accelerating, braking and steering. The vehicle is driven by giving these inputs to the physics system, and it works the same whether the driver is a human player or an AI agent. The physics system doesn't know or care which one it is. That's what makes this so tricky to implement an AI for, because I do not (yet at least) "cheat" by teleporting the car around. Instead, I calculate what the inputs need to be to get the car from point A to B. Regarding the tesselation resolution, it depends on what you means by "resolution". I use Recast and Detour for building and querying the NavMesh respectively, and Recast has a bunch of settings you need to give it when building the NavMesh (so yes I build, or "bake", the NavMesh in the level editor). I can list the settings here but I don't think that's what you were asking for. If you'd like me to go into more detail, I can do that. Not sure what you mean by "granny crane". I do have a bunch of ideas on how to tackle the issue you refer to. One such idea is limiting the AIs ability to make reverse-turns in tight spaces. Instead, they would switch into a sort of "reverse mode" where they follow the driveline normally, but with all the inputs flipped, to make the vehicle back out of the tight space until it reaches a larger area where there is more space to make reverse-turns. I might have to make a follow-up video when the Mother vehicle knows how to get out of that area :) For the network transport I use ENet: github.com/lsalzman/enet. It's buried pretty deep in the game engine and I have my own RPC and propagation layers built on top of it so I rarely have to interact with it directly. However, as it is a C-library, you should be able to use it directly from Zig.

We'll see. The engine based around a client-server model which handles some of the heavy lifting of multiplayer support, but actual multiplayer gameplay isn't really a priority right now. A co-op campaign + messing around in multiplayer in the sandbox mode would be pretty sweet to have :)

I use and have used a bunch of different ones. The engine technically supports multiple different ones since the physics backend is loaded as a plugin. However, the most capable of these backends is PhysX and it is what Traction Point uses at the moment.

Soon!

I did, yes.

@@MadrigalGames damnn impressive!

The realtime shadows (eg. sun shadows) work with the grass, which you can see at around 23:38. The discrepancy comes from adding lightmapping (which is alleviated by the grass painting) and SSAO (which is alleviated by the SDF stuff) to the mix. And yes, like I mentioned in the video, I'd like to automate the painting to a certain degree, with the option to go in and touch it up later if want to.