instagoat

Merkava should look like this ingame:

I'd upvote that kind of thing. I've made an example ticket here: http://feedback.arma3.com/view.php?id=12555 I will possibly make some examples and update that ticket there with some new examples, if that would be okay? Edit: Looking at it now, because of the way the masks work, I am not sure that you could make overlays that work with every screen resolution unless you fiddle with the reticles again. They are good as they are now, and in general I'd rather see improvements to the driver and the way turning out and engine and weapons are handled, rather than these overlays that are purely cosmetic. Edit: this ticket http://feedback.arma3.com/view.php?id=13397

http://feedback.arma3.com/view.php?id=13397 There's a need to take a look at these issues. Right now, they are a nuisance to commanding AI in vehicles, and the ticket has been open for a long time and received surprisingly little attention. Am I the only one driving around in vehicles together with AI?

Will there be a dedicated armor thread at some point, and some kind of feedback about the implementation of thing like Active protection systems and ERA?

What the vehicle needs isn't more armor. From the frontal arc I've survived three to four ATGM hits, and similar amounts of APFSDS strikes. It needs a mortar, and ATGMs, and maybe HE-F-T ammunition. Or Canister. What the game in total needs is better handling of vehicle crew. Ie, don't abandon a vehicle that can still fight just because the engine is dead or the tracks are down. Unless it is about to explode, don't leave. And more control for vehicle crews: http://feedback.arma3.com/view.php?id=12806

Hot Fusion is expensive. Hot fusion is difficult. We are trying to do something that takes stars 40 times the size of Jupiter to get going out in nature, but on a scale measured in meters instead of thousands of kilometers. Cold Fusion is a non field, no matter how many press articles and deluded dreamcastle builders you link to. Unless there is an open discussion about how this works, and somebody actually explaining how the posed problems to the theory are not problems (such as the thermodynamic issues, the lack of obvious chemical pathways of fusion and because of that the lack of appropriate byproducts and/or end-products), no point in paying attention to this anymore unless you're an actual researcher trying to push the field. With hot fusion we know in great detail how the process works, and we have Ideas about how to make it happen. The challenges are mostly technological, and tech problems can be overcome. Progress is slow, but considering the payoff this technology would bring, well worth the effort. None, because they do not have the money. However, somebody needs to build these reactors. They are cooperative efforts between dozens of research projects and often hundreds of companies providing technological hard- and software. There are no individual companies building individual reactors, there are hundreds of dedicated engineering companies providing services and design solutions to developing research prototypes. That somebody is trying to sell something is no argument. By that logic purple plate energy fields are real because somebody is making and selling them. Sorry, but nope.

I just wonder what the tactical sensibility of equipping a completely different gun (of smaller caliber than the default one no less) on what surmounts to a TUSK kit on a normal Slammer...? Why not go all the way and give it an autocannon and an AT launcher instead, make it some kind of NATO super heavy BMT?

Numbers 5, 6, 8 and 9 are also virtually indistinguishable from each other.

Personal SF Combat Coverall (Bundeswehr), Helmet, Glasses and Vest are BWMOD, gun is Toadies M60. All modded weapon sights are by Alwarren from his Attachments pack. Personal SF Uniforms with milspecmonkey style PVC patches. Facemask is from kunico's hidden identity pack, gun is Toadies M60. Personal SF Uniforms and SF Coveralls. Facemask again by kunico.

Better on all points and improving. 10/10, but I am biased.

I don't envy you for having to make these vests fit. I've seen some of the really slim ones that only had fabric cummerbunds with stretch material to velcro under the back or front flap, really tight at the sides of the uniform, no bulk at all. Even the less bulky default vests are rather bulky, and even they have clipping issues on some uniforms. Are you going to just model your vests so they fit perfectly on your uniforms and not take the defaults and other community uniforms into account, or are you working on making them fit somehow still? After looking at the pictures there I'm finding it difficult to imagine how some of those can be modelled without clipping issues :/ Or is there a way to make the vests find certain memory points on the uniforms and stretch? That was how I was imagining the system to be when BI announced the modular Uniform system back in 2011, but it doesn't look like that is the case at all, with so many default vests already having such problems (especially on the FIA guys with the leather smocks, for example.)

I find it kind of sad that this dreamy dreamboat that has produced nothing of serious impact aside from a the pons & fleischman scandal-esque situation gets so much attention compared to hot fusion. There is an actual theory there, which has been verified, and genuine engineering and science problems related to the topic. There are tons of things happening in that field, which actually have resulted in construction of prototype powerplants and testing rigs for various types of confinement systems. But no, for some reason, magic draws more attention than genuine mystery and genuine problems. Just because some government mentions this non-theory does not make it valid. Anybody mentioning it does not make it valid. Albert Einstein could personally walk out of his grave and proclaim that there is something to this theory, it would STILL not be valid. The only thing that would validate it is openly presented and discussed problems, addressed by an openly presented, clearly described, testable theory, which would then lead to its hypothesis being confirmed (ideally.). Nonsuch is happening here.

ctrl+g to select different grenades.

The british SF soldiers use it. It's a retextured Blufor crye vest, visually.

What may be happening is that the AI is "thinking" in too concrete categories. It tries to really BE clever, rather than trying to APPEAR clever. There is a difference, because the former requires very concise and situation specific, abstract considerations and thinking while the latter just gives an outline for a basic threat and then the AI reacts in one direction or the other. I can't quite describe how I see this, since I am not a programmer and have no familiarity how the Ai really works at heart, but in general it seems to me that some basic things it needs to be doing are all that is necessary. A massive reduction of input variables, and resulting reduction of output variables. Unfortunately, it is exactly that reduction that seems to be so insurmountably impossible to fix. Just trying to think of a way to make an AI deal with obstacles in their line of fire is a nightmare, as the friendly fire and friendly running-over accidents prove, which are way too common.

They do it very marginally. It's not much, and some anims are really sub-par anyway. A good example is toadie2ks M60 reload, though. Very natural movement because the torso moves along with the arms, gives it a lot of weight, as if the character really has to use strength to pull the charging handle back and slam the feed tray cover shut in the end. Animations like this are an art form in themselves. In the CS and other fps communities, these anims are usually built by other people than those who develop the guns. Maybe it'd be worth to check out Toadies anim for a gun like this once he makes them available? He seems to intend to publish all his animations for the community to use, so maybe there's something in there too. Not wanting to talk too much into your work though... I appreciate this gun along with the attachments pack, so thx again!

I actually am beginning to think that the one thing to make the reload animation look more "alive" is to also animate the torso. Most tutorials don't help with that though, most games only have to deal with animating the arms, not the whole torso :(

Uh, well, thing is that the bots in CS are using pre-planned path nodes baked into the map by the developer. The AI does nothing but select which path to go along in which number, and then they aimbot without having to cope with bullet drop, or any weapon fancier than a shotgun. CS has no AI to speak of, and again, it has extremely restricted maps with not even a little bit of the complexity that Arma's AI has to cope with. CS:GOs AI couldn't walk down a street in this game. Show me a different game that actually has an AI. The Cryengine AI is an example of a freeform AI, which is rather competetive but still not working with as much complex issues as Arma's AI, and also in more restricted environs. The combat distances are also much, much shorter in that game and the combat priorities are different. But on the whole, having not much experience with these AI apart from short stints in far cry and crysis, I think this is close. However, is it really better than Arma's AI? I remember much complaint about the stupidity of the AI all the way through Crysis 1, its addons and Crysis 2. Again, a familiar story, and it should tell us that Crye's developers are probably not better off with their AI than BI's are. So, what other games are there? OF:DR or RR? Both suffer from shortfalls, both have their advantages, but in no way is either a distinct improvement over what is offered to us with Arma 3s vanilla AI. FEAR? They even went so far as to remove clutter from their levels and designing the levels around the AI, rather than the other way around, to the point of being able to strip the AI down to bones and still have it work brilliantly. Not possible in Arma. Just claiming that it is easy to find an AI "better" than Armas without putting forward any clear examples is rather cheap, don't you think?

I've just noticed a very minor thing with the reload animation that irks the physics know-it-all in me. When tipping the weapon to check the extraction port, the character rotates the gun about the axis of the wrist. This would be difficult in real life because you would have to rotate the center of mass of the gun around your wrist. It's much easier and natural to rotate the gun about its center of mass (ie, around the axis of the bore), with the hand moving around it like a lever. 09 seconds in, -very- short moment but you can see it briefly here. The rest of the video is interesting from an anatomical study perspective, since in Arma we don't just have to deal with the arms to animate like in all other games, but with the whole body more or less. Going to far, though, really all I wanted to point out that it would look more natural if the reload would rotate the gun about its center of mass, rather than the arm-wrist axis.

AI in general is the toughest part to develop, because there are zero frameworks and standards. Plus, the stack-and-tack way of developing it so far has led to it having become unwieldily complicated. I think this AI, by any standard is among the most complex and extensive AI systems in use in games today. To my knowledge, there are no other games where the AI is asked to do this many things in such varied environments. Other games develop maps and gameplay to fit the AI, where it is used, and where it doesn't work they have it cheat. With Arma it is the other way around: the devs put down a world and open up gameplay completely, and then go in to make the AI work with that. Putting down a four room apartement is something the AI's of other games have been written entirely around, with each level meticulously designed to fit the capabilities of the AI to deliver the best player-AI interaction possible. Again, not so with Arma (I know I am repeating myself here, but this point needs to be emphasized I think.), where the world is placed with at best minor attention to the AI's ability to work inside it. (Alone changing the compass direction of approach can change a certain terrain's tactical significance so much that there is no benefit to trying to optimize the AI for any given location on any map.) Then the AI is built with the goal to make it perform good, on average, across the entire map and the entire range of situations that they might find themselves in. Basically, they are attempting to make a non-random code that can address a completely chaotic, non-predictable environment with chaotic distribution of tactically significant points and avenues, and then make this system appear as if it is felling clever decisions in the situations it faces. And contrary to you, the human, this code has less brainpower than your average worker ant. Consider the amount of different processes and filters you are running in split second timeframes when felling tactical decisions in gameplay. For example: Cross a street: Enemies are nearby, none are currently seen. There is a wreck in the street, and open road left and right, with buildings and doors to backyards for a hundred meters in each direction. There are windows observing the street, and there is gunfire nearby. As a human, you know all this, you will check your benefits of being on the other side versus the risks (Sniper observation, likelihood of enemy contact based on memory of enemy location, direction of gunfire, terrain knowledge and reasonable extrapolation of enemy movement, again based from memory and experience, mines, cover, different routes with similar cycles to consider, etc, etc, etc) and you will also be able to dynamically filter ALL of those information based on your training and mental threat response level down to things that only matter, right now. Note that I am presupposing the posession of the knowledge, and discarding the fact that first you also need to -aquire- this information and sort it for what is relevant and what isn't relevant. In terms of AI, memory cycles. At the end of all those processes, usually within one to five or so seconds, you will have felled your decision, either crossing the street and coordinating with your squad, or staying where you are. And you will reevaluate what you are doing at every moment from then on as your senses pick up new information. Now imagine all this has to be done by a silicon chip incapable of simultaneous processing as our sensory apparatus is, that is using a very different way of analyzing and prioritizing this information (hand-written too!) and which has to do this not for one "person", but potentially hundreds of them, all at the same time, each with their unique tactical situation. That includes different weapons, worn gear, optics, protection, presence of friend or foe, vehicles, silencers, etc, etc, etc. The more I learn about AI and also about both Human and non-Human animal perception, the more I respect the fact that the AI is functional at all. Maybe it even is easier than I imagine it to be now, but at this point the only hope I think is that somebody calls some important EU Universities AI departement and begins a doctoral work with developing a completely new and revolutionary AI system using cutting edge code and tech for the RV engine. Because I fear that the limitations of the AI in many respects are not limitations of code and ability, but of technology.

16 AI inside a town fully cut my fps in half. 160 AI drops me down to 8 fps. Without any AI on the map, I get 60 fps on stratis and 40~ on Altis. They are drains on the performance.

Making animations like this run smoothly and lifelike is difficult. Even some of the BI animations are weird (MG reloads). The only community made reload anim that's up there with all the other FPS titles is Toadies for his M60. But the guy also has a lot more experience than most Arma modders have with regards to these things. Arma's only gotten reload animations now, CS modders have been doing them for ten years now. It could be improved, but it is far from "too X" anything.

I am so sad about only knowing two other people who have Arma 3 :( This would be such a blast with a bigger group

I think it's much the same as a GM in a D&D or Dark Eye game. If they do this correctly, there'll be enough things to do for him to keep him entertained along with his players. Fully depends on what kind of abilities he's given, along with what kind of sensible restraints.

Live AI is one of the most resource intensive things you can do, especially when the AI gets complex. Most very advanced AI is run on supercomputers, and AI in games is actually nonexistent or cheating a lot of the time, if not most times. Consider the amount of decisions and action feedback loops encountered when simply crossing a street, and once you've that visualized, try to imagine that for dozens of individual AI in a nonpermissive environment. AI tech for games in general is piss poor. What would be necessary for a game like Arma to really shine in the AI departement would be dedicated AI hardware, like a graphics card but specifically designed to run AI routines, with some sort of standardized engine behind it. However, there are no games providing incentive for the development of such, and in general, competetive AI is not desireable because if your average enemy in a shooter is as competent as a human player, the difficulty would ramp up massively. In some genres, prohibitively.