aeonios

Not the last time I checked. It was an idea being thrown around, something that probably ought to be done but which hasn't yet. I don't know if anyone is even working on it.

I think the most important thread to have is for networking, at least on multiplayer. Not having networking on a separate thread causes major problems including some notable disconnect issues.

No, physics should handle that. The pathfinder becomes extremely stupid when trying to calculate paths around other units. It gets stuck very easily and ends up causing huge traffic jams on some maps. Buildings/terrain/trees/etc are fine for pathfinding since they don't change much, but with units you don't really want it to worry about something that might have moved out of the way by the time it gets there. That just tends to give you really bad paths. Well there was that too. I know that it was very WIP last I checked on it, but I dunno how many people have had much motivation to work on anything when nobody even knows when stuff will start getting reviewed or merged again.

Correct, this could easily break sync. GPUs and CPUs alike have various quirks when it comes to floating point. The standard itself is broken. We've dealt with that CPU side already, but moving it to the GPU you'd probably lack libraries for ensuring floating point consistency. Correct. If you want to mess with the GPU you'd be better off doing real graphics instead. The current pathfinder is actually inefficient in a lot of ways, not the least of which in the way it handles collisions (which really shouldn't be part of the pathfinder at all). Apparently it also uses flood-fill in a lot of situations which is horribly inefficient. @wraitii has been working on some sort of complex multi-stage patch to fix all of those issues, but along with a lot of other things it got stalled by the halted development. From what I saw of it the pathfinder is also ridiculously complex.

Not to be a downer but 0ad is really not a good AI development platform. The current AI interface in 0ad is basically a horrible hack designed specifically to support petra, and petra is a nightmare of spaghetti code. If you're just looking to get into AI I'd suggest looking into Zero-K instead. It's much, much more friendly for AI dev and even has a very competitive AI dev community. I'm also kind of curious as to what you mean by "distribution functions". That's not an approach that I'm familiar with.

ReShade actually has a depth-based SMAA which would be a great reference for building a depth/normals based one, so I may end up doing that sooner than expected.

I didn't know that was a thing. If we could split grass into individual leaf pairs it would look a lot better in a lot of ways, since the current strip grass creates ugly strip shadows. This could be pretty epic. I'll see what I can do. The current patch doesn't need that much work. An issue I'm running into repeatedly now though is that the graphics options panel simply doesn't provide enough space to display all the options. I'm going to have to merge the water reflections and refractions options into one just to be able to fit the GLSL option back in, and then there still won't be room for the AA or texture filtering options. What's this presentation/conference about anyway?

>Why no AA I was working towards that months ago then the project came to a screeching halt because something something EUSSR totalitarian thought police laws. >Texture filtering That's not something I've given much thought to but it shouldn't be all that difficult to implement. I think it's just a minor draw option so it should be possible to enable it centrally. >Grass There is grass, but applying it to maps is a pain. There are some problems with grass, some artistic in nature and some due to performance. Eventually we'll have an openGL 3.3 option with drawinstanced which should take care of the majority of gfx performance issues. The art thing is actually more of a pain because if you place grass on the side of a hill part of it will be floating off the ground which just looks awful. Getting even grass coverage is another challenge, but fixing that requires first fixing the floating grass issue, because it makes shadows (and low angle cinematic shots) look bad and grass looks really bad without shadows.

eh? I have no idea when I'm going to be able to get any work done again. I honestly want to do a more sophisticated AA using depth and normals to detect edges more precisely and for less cost than MSAA. I dunno about ReShade I'll have to check that out to see how that works.

AO is a pain, and possibly expensive. There are filterable versions that are promising, but still a pain. I might play around with that eventually but the extended feature freeze has pretty much forced my work to a halt.

Sure, right next to the skeletons of the people who left them there. Random treasures are a weak way to balance a map though. You can't expect random treasures to supply the tens of thousands of wood that each player needs in order to compete reasonably.

Sand dune maps are so cool but so impractical from a gameplay standpoint. That said I'm not a huge fan of the orange sand, and it's a bit boring both from a texture standpoint and in terms of terrain. I think there should be random open/lowland areas that aren't covered in dunes. Maybe even oases. That would make it more interesting and also create an opportunity to add trees and things to the map to make it less resource poor. This too, although that's really easy to do.

Scaling for 3D models has to be done around a reference point. Since models aren't spheres they may have odd types of symmetry so a naive scaling algorithm is certain to fail. I'd suggest using a program specifically designed for working with 3D models to do the scaling with, like wings3D or blender. Those programs have better facilities for managing scaling to make it easier and to allow you to see what you're doing.

This sounds like a stance issue rather than an issue with unit ai. Also, not having units attack aggressive animals automatically would mean they'd always be stuck fighting lions and things 1v1 even if they have a bunch of friends nearby that should be helping them. That wouldn't be very smart. Probably stances should be split into range and behavior so that both can be controlled better. Also allowing users to set default stance settings for each unit would be good too.

Technically there was no Japan during that time period. The Japanese islands were at that time mostly inhabited by the ainu, who were progressively displaced by immigrants from mainland Asia. None of those people identified as being "Japanese" in any sense. Japan didn't even become a thing until almost 1600 AD. There wasn't really a Korea either, it was like 4 independent states back then without any sort of central cultural identity. They were distinct from China but they were not Korean. China did exist back then. The great emperor Qin Shi Huang united China in 220 BC. While his imperial line was extremely short and ended in 206 BC with only one heir making it to the throne, the imperial system he created lasted until 1912 with varying degrees of stability. Both the Qin dynasty and the Han dynasty occurred within the general time span of 0ad and would be appropriate to include, although the Han dynasty lasted much longer and is probably more characteristic of imperial China in general.

That's not actually true. Most of the foliage uses transparency not only for soft edges but also to blend with terrain and other foliage behind. I messed around with foliage a bit and found that the bleed-through effect was much more prominent than the soft-edge effect (which doesn't seem to be used particularly well at all). Fixing all of the art to use alpha test would take a lot of work and the results might be questionable. Particles can be fast but you have to use drawinstanced and geometry shaders, which are GL3.

That's a common misconception. What the romans defeated was not the fully equipped and well trained Macedonian phalanx under Alexander the Great, but rather a pathetic shadow of it that was barely maintained by corrupt and failing remnants of the empire he founded. If Rome had faced off directly vs Alexander then Alexander would have stomped them into the dirt, quite easily. Rome also had a well trained, well equipped professional military for the majority of its existence as a republic and then later as an empire. Rome also did beat down a lot of lesser states during its reign but it didn't always win against the larger military powers of its day. Rome's military power during its peak came from good coordination between archers and infantrymen among other things, not because they had superior strength or technology or anything. It was just a plain old well trained, well equipped military with competent generals. That said I don't think a sidearm mechanic would be appropriate from a balance standpoint. I've been working on an AoK-style balance mod but the work is pretty slow. A lot of game mechanics either had to be modified or thrown out entirely to make it work well and it's going to be a lot of work to beef up unit rosters sufficiently to get good diversity and balance.

Eh, if you want more than a couple of lights then you do. I don't know if it absolutely needs openGL3.x but it does need to do some fancy data structure access in order to know which lights it needs to draw and to access the data for those lights. I don't know how all of that works unfortunately, and it seems to be more complicated to code than deferred rendering is. Sure, except that they won't be lit by dynamic lights at all, which would look really bad given the amount of transparency used in 0ad. Trees, bushes, even things like shields would stand out like a sore thumb. Everything under water as well. Forward+ would allow everything to be lit dynamically, deferred would not. That is pretty cool. Lightning looks kinda bad without dynamic shadows though.

Not with that fractal placer thing. Trust me I tried and it was only disappointing and frustrating. It might be possible to generate better mountains and things but it'd be difficult and expensive and way beyond my expertise at least.

Eh, you could obviously but they wouldn't really fit in to any map usefully that way.

Eh, how? Mapgen doesn't have anything that can really support making those kinds of mountains. I'm assuming the mesa thing was just a model but the rest of those things require mapgen and texturing capabilities that the game doesn't really support.

Day/night isn't actually that simple to do. It'd require figuring out how to get the sun to rotate properly at a given angle as well as adding new values to atlas/map settings for morning/noon/evening/night colors to interpolate between. Interpolating between colors is pretty simple but getting the sun rotation to not look like crap not so much. The sun angle also would need to present a usable parameter to mapmakers. Also doesn't work so well without a dynamic skybox, which requires a lot more fiddling to allow phasing between night and day. Dynamic skybox needs some form of dynamic clouds, again not so simple and there are at least a couple of different ways to handle that. Eventually probably, but there are other important things that need doing first. Dynamic seasons just forget it. Every tree and bush model as well as all buildings would need 2-4 seasonal variants, which are not supported by the engine currently not to mention the art just isn't there. That'd also require having exact matches for inter-seasonal terrain textures, again not there, and interpolating between the seasons would require loading twice as much texture data with a lot more data switching during the game as the seasons changed. The performance would be intolerably bad even if you managed to get the art assets. It would be nice if we had snowy building variants though. It's awkward having a snow-covered map while the buildings are spotless.

Animal fattening could easily be done in a mod. It would require a bit of work and a few sanity checks, but still totally doable.

Hm. Without using some form of complexity reduction lights become very expensive very rapidly. 0ad's art heavily uses transparency so deferred rendering isn't an option. Forward+ could work but I'm pretty sure that requires openGL 3.x.

In what universe? 0ad's formations are extremely rigid. They don't turn or reverse direction properly, there's no middle ground between completely breaking down and maintaining the formation rigidly when attacking, and they can't handle obstacles at all. There's also no practical use for them besides keeping troops together. In total war formations do turn and reverse smoothly, they handle obstacles just fine, and you can do all kinds of cool and centrally important things with them like directional bracing, shield castle, and charging. Total war also has epic cavalry charges that are awesome for mowing units down. It also demonstrates why units need to be managed in batallions, because each batallion takes on the given formation individually, allowing you to place formations strategically on the battlefield, to flank and surround the enemy. Total war also has a vision penalty for trees, allowing you to hide units in forests to ambush the enemy. The realism and depth to total war's combat system is very impressive, and 0ad has none of that. 0ad also has none of the straightforwardness or simplicity that aoe2's combat system does, and at least aoe2's formations could maneuver properly and handle obstacles. It's literally the worst of both, with territory and citizen-soldiers thrown in just to trash up the gameplay even more and remove all kinds of interesting strategic options that were standard in aoe2. That's why people are playing aoe2 rather than 0ad. Two decades after aoe2 was released and they've still got people playing professional tournaments and we only have 2 high level players total even with all the improvements we have over it.