DualJoe

I've run into the following feature in a couple of missions I've made. Using vanilla Arma2 CO. Sometimes instead of moving to the waypoint, the AI moves to the bottom left of the map and stays there. This may not be much of a problem with user-controlled AI, but in all other situations this is rather troublesome and usually breaks the mission. With DAC the problem becomes ridiculous, and I'm starting to get somewhat annoyed watching big convoys of AI migrating far away from the mission area. Funny thing about DAC is that it shows the waypoints the units are supposed to go to with a marker, but they completely ignore the waypoints. Any ideas or suggestions to work around this feature? Other than DAC I'm not using any AI-addons. The problem also occurs without DAC, all be it less frequent for obvious reasons.

Hello arma community members. I've started a little project to see if I could make my own "map" for Arma. I'm Dutch so I figured why not try some Dutch scenery, which I will try to recreate as fateful as possible in the Arma2 engine, sticking close to a 1:1 scale. I'm posting this thread to get some feedback and advise as I go along. After trying various different tools for terrain editing from the wiki and this forum I've ended up using Blender for pretty much everything and then exporting to a greyscale heightmap. Maybe I could expand on this if someone is interested in using the sculpting options in Blender for terrain editing or is interested in my workflow. Keep in mind this is my first try at terrain making in arma, I am however somewhat familiar with Blender. I've chosen a small area 8km x 8km around the hill called de Grebbeberg for a couple of reasons. First it's small (which may make it possible to work at a smaller cell-size), but also because it's a very interesting area. It is pretty diverse terrain (for Dutch standards anyway) with a river, channels, dykes, but also the hill as the main strategic area. Also it is an historically important area and key part of the Dutch defensive line called the Grebbelinie. In the Netherlands it is most famous from World War 2, where the German army concentrated it's forces on this area and the Dutch even managed to fight them to a standstill for 3 (whole) days. The germans suffered some pretty hefty losses and nicknamed the hill der Teufelsberg (devils mountain) which I'm considering of using as a name. I'll try to post some screenshots later today. I need some help with the following: First up, materials. I have gotten a simple version of the map into visitor3 and copied the materials from the tutorial island of the wiki. The terrain even shows up in buldozer with my crude satellite image as texture, but after packing it in a pbo and running it in arma2oa I get an error message about not finding an rvmat with a lot of zeros and I get a white untextured terrain. Any ideas on how I could solve this problem? Also when placing a unit in the editor and previewing the map, a couple of seconds after the map has started, my player character suddenly gets injured for no apparent reason, sometimes even dying on the spot. I'd like to know what could be causing this. Last I've got some questions about cell-sizes and performance. Ideally I'd like to use the smallest cell-size possible so I can add in trenches and gullies and such, but still have a pretty decent framerate albeit with a short view distances. I've searched the net but I couldn't find anything about a limit for cellsizes. Smallest cell-size for this map in a 1 to 1 scale would be 2m in a 4096x4096 grid. Before I waste a lot of time making small details I'd like to know how far I can go, before running into memory problems or something like that. Any insight in this is more than welcome. Another peculiar thing I've noticed is that framerates seem to be heavily affected by the smoothness of the terrain without changing cell or grid size. My first tests were with a low bit depth grayscale image and not using the full range from black to white, the result was a very ugly blocky landscape made up from flat terraces instead of smooth hills and horrible framerates. Later I managed to produce a better quality heightmap which resulted in a smooth landscape and proper slopes but surprisingly also a much improved framerate. Does that mean that for a decent framerate I'll also have to take into account the number of changes in elevation before deciding on a cellsize? EDIT Google drive link with fairly recent versions of this WIP

Apart from being occupied with other things, I ran into more trouble than I was expecting with my new setup. However after a lot of trial and mostly error I've finally managed to get VGA-passthrough running reliably in two different virtual machines. Now I can run linux and windows side by side on the same machine with full 3d accelleration. This should speed up my workflow immensily, because I no longer need to reboot for every little change/mistake. I also don't have to worry about windows destroying my system anymore. Add to this that back-up and restoring windows should be a much easier process. From time to time I can just save the entire windows virtual-machine to a single compressed file on an usb-disk. Hopefully meaning I no longer need to go through the usual lengthy reinstall of windows and all things arma, should windows die on me again. Apparently I had forgotten to backup my install files, so it will take a while to download and setup Arma2CO and TakeOn plus other bistudio products and tools. As soon as things quiet down a bit in real life and I don't get too swamped with my many other projects, I'll be picking this one back up. Bit off topic, but I have to express my astonishment about the performance of arma2co with windows 7 on XEN. Granted my new AMD 7950 is a faster card than the GTX560TI, but apart from the better graphics I was expecting at least a little hit on the cpu-heavy stuff. Especially since windows was running on 8 overclocked cores before and now only 4 cores at stock frequencies. Not only am I now able to crank up most settings to the max in arma2, but I can now also run more cpu taxing mods and missions withouth the same effect on framerates I was seeing before. I'm seeing 100%gpu load and cpu-power to spare on the 4 virtual-cores I've given the VM. Too bad NVIDIA-drivers don't work well on XEN. Nvidia drivers only support virtualmachines on their very pricy Quadro range. There are workarounds, but at the moment I don't feel like trying them. Which means that for the moment, I'll be running KVM most of the time, eventhough KVM-qemu seems to have nowhere near the same performance of XEN on my system. For me both have their own set of uses and advantages and I'm glad I went the extra mile to get the two of them running reliably. I can do so much more now with my single pc, that I'm having trouble getting to grips with all the new possibilities.

I have to correct myself about the horrible performance of KVM I mentioned above. I did some more experimenting with KVM today and it looks like the bad performance results were my fault. Eventhough I'm still using lower ingame settings than under xen and with considerably lower framerates, Arma2 is however perfectly playable under linux-KVM with 4 cores on a fx8120 cpu at factory clockspeed. Looks like I've still got some wrong settings resulting in low IO-speed and probably the ondemand-cpu-governor, but I got an average of 32 fps on benchmark 8, with Vsync turned on and simultaneously running the Farcry3 installer in the background on the same virtual machine. I had to lower the harddisk-bandwidth stuff like texture- and objectdetail, viewdistance to around 1200 and turn hdr to the highest setting, but framerates now don't seem to go below 28 fps. Because of the need to increase hdr and seemingly no effect of the extra cpu-work in the background, I suspect that running just the game may not be enough cpu-load to trigger the higher clocks of the ondemand cpu-governor.

If you got gpu-power to spare you can try turning off AA and setting the 3d-resolution to a higher value than your screen-resolution. Afaik this actually does what AA is trying to simulate, but without blurring details and without all the strange artifacts.

Actually nvidia cards are less likely to work than AMD/ATI ones. Something to do with vga-bios and drivers. Apparently NVIDIA Quadro have no issues, but those things aren't cheap. That doesn't mean it's impossible, but from what I've seen it requires compiling your own xen-version with built in bios. Vga-bios seem to be listed online or you can extract it yourself, but I haven't gotten to that point yet. That's why I was hoping to get kvm working, because now I have to run the nouveau driver when running xen. The plan for me was to have linux and windows running simultaneously with full 3d, but hadn't counted on the nvidia-driver ruining things. For those interested, I did some more tests in Xen today and it really does work eventhough my PSU and/or cooling is too weak for stable overclocking. I'm running a fx8120 at stock speed with a Nvidia GTX560TI and an AMD7950 passed through for windows running on 4 cores. I'm running arma2 on higher settings at a higher framerate (1920x1200 with most settings on Very High except terrain, hdr and postpro, sofar between 48 and 60 fps) then when I was running windows on the bare metal on all cores and overclocked to 4GHz with the Nvidia card. I don't understand why I'm getting so much more performance from Arma running in a virtual machine compared to native. Looking at the graphs the 4 cores are not being maxed out at all, so I might try some different malloc and exthread settings. I haven't even begun tweaking windows7 yet. I find this performance increase really, really odd.

I got stuck with the sprocket installers of the DLC's. Been a while since I've tested, but I couldn't figure out what I needed to install for those to start working. Then recently my windows 7 decided to disable all usb-devices, with no option to turn them back on, which for me was the last straw and I wiped it completely. Unfortunately I can't leave arma2 so I've found an alternative for wine in the form of virtual machines. I've removed windows and repartitioned my system with the objective of using virtual machines. My system supports IOMMU and I've gotten vga-passthrough to work both in Qemu-KVM as in XEN. 3D performance in KVM on my system is horrible, but I've seen reports on the internet of people with great results. XEN however works absolutely brilliantly. I only got it working this afternoon, but I'm allready blown away with the results sofar. One thing I don't understand is why I'm getting better performance in Arma2CO on a couple of old HDD's in a virtual machine with 6 of the 8 cpu-cores, compared to what I was getting in windows native with the game on a ssd and 8 cores at a slightly higher clock. To be fair though I was running a nvidia GTX560TI and I added an 7950 just for vga-passthrough in the virtual machines, but if anything that should've only increased cpu and hdd-load, not the other way around. Anyway if your system supports it and you've got an AMD videocard, I can highly recommend XEN for windows7 (gaming).

In short, depending on how exhausted I am after this friday, I'll try to get a bit of work done this weekend. As soon as I've got something ready for texturing, you'll be the first to know. Apart from that, I haven't had the time or the energy to work on this as much as I would have liked. My eye problem really came at a bad time. On top of that a couple of people I was depending on in real life have left me with such a mess as a holiday present, that I'm starting to doubt if I'll ever be able to fix it (which I'm expecting to find out this friday). But before all that, I'm waiting for my new card and then I'm reinstalling my pc for a fresh start. After some serious crashes while trying to get usb running again, I'm not feeling confident about my current windows installation at all. Certainly not enough to entrust it with hours of work.

USB issue is resolved! I don't know how he did it, but my brother in law found a microsoft support post detailing the removal of two lines in a very obscure windows registry number. So obscure that even the search in regedit could not find it. Anyway I've regained access to the Bistudio tools now so I can finally try to finish some buildings. "Unfortunately" I've already ordered an AMD-7950 videocard for use in my new windows-virtual machine which should arrive any day now. The combination of being separated from games plus the arrival of this new toy, might result in some ... errr ... delays .

@Domokun: Thanks, I've cooled down a bit and I did indeed take a walk in the freezing rain. I also suspect this problem to be caused by the new UEFI rubbish on my mainboard as well. The thing that gets to me, is that no matter what I do (or don't do) Windows always seems to self-destruct just by running it. My hobby OS Linux on the other hand only breaks if I personally break it. I'll figure it out eventually, I always do. Funny how your pc always breaks down at the worst possible moment, just when you need it and don't have time to fix it. By the way, good point about the SFX, hadn't even thought of that. But having the mill turn without sound would indeed be weird.

@Iarsiano: Right I'll make the buildings my first priority after I've solved a little windows feature. Finally sat down to get things set up and running in windows again. After a restart windows decided to drop support for USB all-together, for no apparent reason. I did update my bios, but the first few times booting after that I had no problems. This is what happened: I started windows no problems, got a phone call which would take a while, so I let windows poweroff the pc. After the call was finished, started windows again, usb gone. I've tried just about everything, but windows refuses to enable both usb2 as well as my usb3. (Btw ever tried navigating windows hidden options without a mouse and only with a ps2 keyboard?) Since everything works just fine in linux I can pretty much rule out any problems with the hardware. It's really hard not to hate windows when stuff like this happens. In my case I sort of exploded and it took every last bit of restraint I could muster to not stomp my pc to kingdom come. After I cooled down, I've been trying to find a solution, with no luck so far. Chances are I'll be using this opportunity to switch to virtualmachines combined with pci/vga-passthrough, which my hardware should support just fine. At least then the consequences of windows features can be contained and a windows crash won't necessarily result in a wiped bios or other absurdities like that. Anyway I'll need a bit of time to work this out, I hadn't really looked into virtualmachines before, but running windows on the bare metal is bad for my bloodpressure and my pc might not survive another windows feature. Excuse me for sounding a bit pissed of, but that's because I am really very pissed off at the moment.

Sorry for the lack of updates but my eyesight problem messed up my RL-schedule big time, leaving me with a lot less free time to spend on this. You could also say that I have too many projects running simultaneously. Pick whichever you prefer. Anyway I'm now fairly confident I'll be able to look at bright screens again, so I'll be slowly picking this up where I left off, starting with materials and the following effect which I may have mentioned earlier in this thread. Look at the sails, this is a video of the opengl 3d-view in Blender. I forgot to set the sail-planes to single-sided and discovered this effect by accident. If I'm able to get the scripting worked out, including on for rotation based on windspeed and direction, I might also be able to implement an animation to switch the backlighting effect on and off depending on the position of the viewer and the windmill relative to the sun. Otherwise you would also see backlighting when standing behind the windmill looking towards it, but with the sun in your back.

Sorry I've been away for a while, I'm still having trouble with my vision especially with windows-os and small fonts. Can't seem to find an easy way to zoom in in all applications, like I can in linux. Also the default black text on white is a major pain in the rear end. Maybe because I'm more familiar with linux, but I miss the easy negative color toggle and easy zooming in windows 7. Not only am I getting massive, massive headaches, but it drains energy like you wouldn't believe. Either I manage to get the entire BI-studio pipeline working on linux or more likely I'll have to wait till my brain adjust enough so I can work on this project for more extended periods of time. After that I'll look into the projects mentioned. I can understand the Dutchies here are getting impatient for some Dutch stuff, trust me I'm frustrated as well at the moment. It's also causing me some serious challenges in real life.

Yeah I'm making sure this heals up well, I'm pretty nervous when it comes to my eyesight. I'm not to keen about losing another one of my senses. I've always used the best screen displays I could afford, all pc-hardware is secondary to me. My current monitor is a 27inch wide-gamut DELL screen. I can't stand TN, but I have to admit I've been very impressed by the latest cheap IPS screens from LG for example. Especially compared to the prices of our Dells. Color quality is still far superior on our Dells, but for gaming I'd recommend those IPS-screens. (We could have made a virtual surround cockpit with a couple of 24 inch LG's for the price of one Dell) I was so impressed I've bought 2 this year, one for my arcade-cabinet-multimedia-coffeetable and one for my parents.

No I'm the one doing things wrong. I was mucking around with the sea-water shorefoam shaders and those errors are from defining empty textures. Work on pc's has been on hold for a while, because I'm still having some problems looking at bright computer screens after my eye infection. Probably because of a combination of eye-drops and my brain having to readjust for the difference between both eyes. I'm slowly trying to pick things up again and I'll fix those empty textureslots asap. Give me a bit of time though, because real life obligations have also been stacking up.

Be warned though this game is very addictive. Lopping of the heads of your friends is strangely enjoyable. In between spawns you can then relieve some boredom by killing peasants, burning down farms or kicking slow teammates in the rear. Also other than in most twitch shooter things, in this game getting a kill really feels like you were the better man, instead of the more experienced glitch exploiter.

Point was to show the performance of the fx-cpu when using benchmarks that don't have inbuilt favorites. The benchmark-suite itself is mostly cross-platform actually, so you could run it on windows as well. The nice thing about that benchmark-suite is that you can install and run it on your own machine and not have to take the sponsored words of popular review-sites for granted. Just like the OP I'm having serious doubts about the claimed inferiority of AMD FX cpus and I share his believe that the cpu is just not utilized to it's full potential. Don't know if the same holds true for an Intel cpu or not. As for the BF3 idling the cpu, I grant you that was a bit irrelevant, comparing that to the complexity and scope of Arma2 is completely ridiculous. It would have been strange if BF3 did indeed tax the cpu.

This post is probably mostly of interest for people wanting to add terrain-objects like buildings and such. Faced with the challenge of filling my Dutch terrain with objects that are neither East-European nor Desert-themed I had an idea to speed up content creation plus adding some optional variation. It started in my Grebbeberg WIP thread, but to avoid confusion I think this modular idea should be discussed in it's own thread. My idea was to attempt to transfer my modular workflow from Blender into the Arma-engine so that non blenderhead-arma-hobbyists could profit from it as well. It is somewhat like the way you can change loadouts and weapons on vehicles and soldiers, but this time for map-objects. In short I'd like to create a library of base-objects with a number of modular components for each of these base objects. The plan is that this would allow for the maximum amount of visual variation with a relatively small amount of work. On top of that, adding or changing modular components and/or their textures should be easier than having to create complete objects from scratch just to add some minor mostly cosmetic changes. I talked about the idea in a bit more detail in my Grebbeberg-WIP-thread here and here. Since then I've been trying to piece together a working mockup of a modular building, demonstrated in the gif-animations below. The first one shows the base building and the second a more recent version after I've played with it a little. Keep in mind this was just an experiment to see what was possible with the limitation of the basic building model (the one with all the things needed to get it to work ingame including AI-pathways and roadways and such). At first I was a bit worried that because it needed to be modular, the buildings would look too generic for this to be a viable option. But now I'm starting to realize the potential of this system, provided I find a way to make it work nicely in the Arma-engine. So I figured it was time to open a thread about this and see if I could get a ball rolling. I'm curious to see what others think about this and if there is an interest in something like this. Who knows, maybe this could turn into something like CBA for terrain-makers.

In that case it would be more practical to make the chairs into vehicles. AI can't walk around inside vehicles.

As it stands I haven't worked out exactly, how to implement the modular system inside arma yet, which I'm hoping I can get started with sometime next week. All the variations in the model I've shown would work with the exact same AI-pathway lod. (Door animations for different doors and locations is a different story though, but might still be possible if user actions can be hidden reliably) Geo and fire lods work inside proxies as far as my tests have shown. Haven't tested performance yet, but for now I'm not worried. I can have hundreds of my fairly complex windmill-model ingame and onscreen without a drop in framerate. In comparison these modular buildings will be ridiculously lowpoly, plus LODs work with proxies as well. If by hidden selections you are referring to the same system that is used for partial destruction, then the mechanics are most definitely working ingame even for realtime switching of modular parts. Now that I think about it, partial destruction works in the exact same way as what I'm trying to do and afaik also works fine with fire/geom-lods with AI. Concerning UV-textures; I haven't experimented with this part of the Arma-engine yet. I've been having some eyesight problems for a while now, which means I'm having problems reading at the moment. Once that clears up I intend to dig into it and figure out how the material system works. Especially areas that effect performance, for example if it's possible to share resources like textures without increasing the VRAM too much. For now I haven't got a clue how Arma-materials work yet, how many layers/textures can be used, different rendering passes, if I can have a separate lightmap UV or not, if things like on the fly hue changes are possible for color variations. However the only issue I'm a bit worried about is lightmapping, overlaying uv's to save on texturespace is hard for baked lighting. EDIT Tried to find something on hawken. Are you talking about the mech-game? Could you post a link for the modular bits then? Their youtube videos and screenshots are so bad I can't see s...! Almost as bad as the overexposed bright stuff thats on the BF3 buildings. Is that a new trend or something? Either produce random noise or destroy visibility to such a degree that you can't tell what you're looking at, not even what color it is. Keep in mind that it needs to work within the limits of the arma-engine. I've seen some other projects with very impressive procedural buildings including stress calculation and dynamic destruction (even one in the Blender-game engine), but that won't do us much good in Arma, unless I suddenly level-up into a sqf-coding-god.

I've got a fx8120 clocked at 4 Ghz and my GTX560TI is the bottleneck on my system as well. I've also never seen close to 100% usage in a game on any of the cores. At least Arma2CO taxes parts of the cpu somewhat, whereas it falls asleep idling for things like BF3. You could try installing MSI afterburner and turning on some gpu-statistics in the osd. I agree that people saying the bulldozers are shit cpus are indeed talking crap. (There are some benchmarks painting a different picture) That being said I seriously doubt you'll ever see close to 100% cpu usage on a game. According to one of the Bis-devs Arma3 with DirectX11 should be able to utilize more multithreaded hardware-power on modern systems, which might increase both framerate and cpu-usage.

To keep things clear I've started a separate thread for the modular object idea and I'll try to focus solely on the terrain-specific parts in this thread.

Raising the shorelines will not solve the issue unfortunately, already tried it plus I can't just add 2/3 meters without ruining the map or have unscaleable channel-banks because of the steep angles. Could be a nice dynamic to mimic slippery slopes, but don't know if the AI could handle it. Also on my map the LOD system reduces the cellsizes further away which means it will automatically lower/raise terrain further away depending on your terrain-detail-setting ingame.

I'll look at them later this evening to see if I can use them. On the other hand, if I'm able to work out this modular system, anyone should be able to produce new buildings by the truckload. I was a bit worried at the start that the modularity would impose generic and dull looking buildings, but now that I've tried a couple of different building designs my expectations have flipped completely. On me at least it has the effect of increasing creativity, because you can focus on single areas, without having to worry about the rest. I'm getting loads more new ideas now, whereas before thinking about all the little stuff like lods and stairs and such got in the way. As long as the part fits, you can go wild as much as you want. It shows a bit in the gif above, I obviously made more variants than strictly necessary for testing purposes. I've updated the image-link above this morning, couldn't resist making another roof-variant, upping the current combination total to 9. I'll make a few more facades and maybe a couple more upper-floor variants before forcing myself to test the concept ingame. Worst case scenario it will only work inside Blender, but for me that only increases the options, because that just removes the limitations of current arma-map-tools. Don't worry, I'll do my utmost to make sure that non-blenderhead arma-addon-makers can use it.

Ah pardon if I didn't make myself clear enough. These models are nowhere near finished. I'm just trying to work out a modular system and posting the results as I go. This isn't even at the proof of concept stage. I'm just showing my progress as I go along. This is a pretty complex 3d puzzle and all pieces need to fit. Also keep in mind that I've got a terrible memory so I'm unable to plan far ahead, it also means I can't work this out in my head. (It's also why I tend to repeat myself a lot, for which I apologize in advance.) Before my eye-infection I had already made a more detailed version with rooms, stairways and walls, but due to my terrible memory I have no idea where it is. So now I'm posting these updates, not only in the hope that I can get feedback and ideas, but this thread is also my online log/diary. It's the only reliable way so far in which I can keep track of what I was doing and what I discovered. The textures are just a couple of very quick procedural ones I made in Blender to help people visualize where I'm trying to go. I've noticed that quite a few people aren't able to visualize the finished product if I leave things completely untextured (present company excluded of course). The thing I'm trying to sell here is that it's not only easier to generate a number different looking buildings, with these modular parts. But it should also be relatively easy add or modify these modular parts which can very rapidly increase your options. For example the buildings in the gif-animation below are made with 2 roof parts and 3 facades. Imagine if you can change the textures as well for example to painted brick, plaster or wood and you'd have all these combinations times 3.