Home 3D Art Showcase & Critiques

LANDNAV simulation game development

2

Replies

  • sacboi
    Offline / Send Message
    sacboi insane polycounter
    Like the Hummer's addition, to my mind adds a point of reference for the player, in a sense something too aim at whilst navigating en route.
  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter
    Doing some playtesting: 



    There's plenty more I'd like to do regarding graphics and the world but too many other things take precedence for the moment. Once I basically finish the game and get mostly focused on marketing, then I'll start adding in more variety to the vegetation and more hand-crafted areas to make it feel more like a real place and less like procedurally spawned foliage.

    Besides just adding more content I want to see what can be done to get smoother looking appearance. Especially the wind animation grass looks kind of choppy - i bet there is some stuff i can do with anti-aliasing to get a cleaner look. 

    Couple more months I'll have demo ready that includes training. 
  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter
    A demo is available now, you can get it at the LANDNAV discord here : https://discord.gg/mz5WZz9KHW

    If you aren't on discord and would like a copy send me a PM and I'll send a link. 




  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter
    Some quick running around after making some changes to how the forest is spawned: 


    I saw there is a tool on the marketplace for more user friendly procedural spawning compared to Unreals built in Procedural Foliage tools. The plugin is called "Simple scatter" and really helps cut down on iteration time so I took a few hours to update the forest.

    I fiddled with post processing a little bit too, trying to get a sharper look. My recording quality isn't super high so its hard to tell, but it looks a little more crisp compared to before. Overall I am happy with it - if I had millions of dollars I'd get some help with the lighting and pay a couple enviro artist to go to town making lots of interesting things to look at, but considering the scope of the game and the development circumstances I think it's job well done for now.

    I got plenty of coding left to do but looking forward to eventually making some new terrain types. 

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter
    I was doing some playtesting and noticing a lot of weird graphics glitches related to the DFAO being used in conjunction with origin shifting. I have a script which should update all the mesh distance fields each time origin is shifted but nonetheless just seeing a lot of ugliness.

    That was a whole thing to fix...anyway, it got me digging into graphics side stuff again so I got to tweaking a little more... one thing that's been bugging me is the colors had too much contrast and it felt a little off. Sort of cartoony almost. Things weren't blending together in a natural way - it was just bugging my eyes I felt like it wasn't nice to look at. 

    So I went through all the textures and made little tweaks. Nice thing about unreal is you can do that right in the editor. I also increased the dithering on the tree shaders. Sometimes this is very obvious to see on billboards in the distance, but it overall looks much nicer than a bunch of chonky trees showing off their cards too much. OVerall I feel like this looks more natural and pleasing. 

    I also made a LUT for warmer fall colors feeling. I originally thought a cooler feeling would be most appropriate, kind of get that damp fall feeling, but after awhile I feel that gets old and kind of sad to look at. This is a game where when you play, it will usually be a couple hours at a time and its slow paced so I think it needs to be a pleasing experience as much as possible.


  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter
    I made a few changes that helped get rid of the ghosting artifacts. Game reads much cleaner now, especially during high wind the leaves aren't turning into a blur. 

    That was accomplished by changing these rendering parameters:

    r.BasePassOutputsVelocity=True                // Enables rendering WorldPositionOffset velocities in the Base Pass, significantly reducing blurring of moving vegetation, for a large increase in image quality, at the cost of some performance.

    r.BasePassForceOutputsVelocity=True           // Enables the computing of motion vectors in the Base Pass. Required to be combined with r.BasePassOutputsVelocity=True for a large increase in image quality, at the cost of some performance.

    r.TemporalAACatmullRom=1                      // Enables an alternate filter kernal for TemporalAntiAliasing(TAA) for an increase in image quality.

    r.Tonemapper.Sharpen=0.5                      // Enables a subtle sharpening of the screen via the ToneMapper for an increase in image quality.

    Idea for that came from the Rural Australia project on marketplace. I forgot the authors name but that project has helped me out big time so I have to thank him some how once I publish this game.

    Screenshots won't show the improvements that much but it does look sharper and cleaner overall. I recorded a video but I forgot the bit-rate was turned down so it's chunky looking. Later in the future I'll do plenty of marketing video anyway. Overall game is looking quite nice now. In some areas it's beautiful actually. It looks better in motion of course. I think the colors are working nicely. Using the LUT helped a lot. Other than that there is no post processing. 

    I also disabled distance meshes and the associated DFAO. That does decrease a sense of depth, but the performance saving is huge... but the big thing was, with DFAO I was getting a lot of weird artifacts after shifting the origin a few times. Plus the fact that when shifting origin I also have to shift all built meshes as well, which causes a heavy hitch. Perhaps it is something that could be fixed but that's a little more technical than I am so I go with the simpler solution to just disable it for now. 

    The water shader I was using was beautiful but it depended on distance meshes so it had to go. I replaced with the basic water shader from Unreal standard assets. It's fine enough, gets the job done. In future I'll go over the creek sides and beautify them, right now the transition is lacking. That's low priority stuff though.

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter
    Been refactoring the project. Working on redoing the character controller - this time we will (I've hired some help) also include an option to swap to third person perspective. 

    The controller is still being worked on but I've done some quick rigging to test out what it will look like. 



    and a blurry video because i forgot i had recording settings turned low.  Movement is really stiff - its just the default third person controller in unreal. Our implementation will be more smooth and modern. Maybe another week or two before it's ready, but I'll also need to make a couple custom animations which will be done much later.

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter


    Here's some video that demonstrates perforamnce gains after I've changed the rendering method to forward rendering. For this project it's made a significant performance boost. Since I only have one directional light and skylight it seems well suited for the project. Game is almost always staying above 60fps on max settings on my 1060 gpu machine, that is about 20-30 frames on average better than before. (there is a regular hitch but thats in the internet upload, not the game)

    I haven't pushed this update to play testers yet but one user was able to get previous version running on very low-end hardware, so I think we are in good place performance wise. But there is still quite a bit of work I can do still on that front.

    For now, I am still refactoring the last bits of game framework code. In another day or two I'll be finished with that headache and then I can start production on tutorials/training again. But the day when I can make some more terrains and environments is on the horizon :)
  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter


    I had some errors when trying to build, coming from the plugin I use for foliage spawning. I ended up nuking some of it cause it was some corrupted references (I expected anyway), so once again I reworked the foliage spawning just a little bit. Truth be told that's something I never mind doing, and I kinda needed a couple easy days anyway after so much refactoring.

    I've tweaked lighting and many materials and I think it's looking nicer than before. I added some procedural wetness and snow to most materials that gets updated from the weather system. But the specularity values were all over the place so I went through and tweaked it all until it looks natural again.

    I also tried out the new megascans trees. They are fantastic but I have some shading problems related to forward rendering it seems. So I forget that for now. In the future I bet I can get those working, and at that point there will probably be more available too. For future maps that will be nice.

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    Been working on the tutorials for LANDNAV. It's a lot to write. I've been able to put World Creator to good use in helping make some of the instruction more visual though. In image below, for example, I'm showing an example of terrain features in a way that helps understanding the relation between contour lines on a map and real world geography:


    A lot of new play testers have joined the community and this helps me a lot to see where people have trouble with the game. Based on community feedback, I changed the Control Point model to something more easily seen:


    It seems that even people with some experience doing landnav in real world aren't going for the harder difficulty exercises, so I am mostly working on offering more beginner content. Eventually people who stick with it will need more challenging content but I've got to make sure there is enough stuff to get people started.

    I haven't done a ton with the art, but every time I do some play testing I always end up tweaking some little thing. Usually it's something like, "that one type of grass is jumping out at me like it doesn't belong," and so I tweak the textures a tiny bit. Over time I feel like these little nitpicks are making the game look nicer. Once I finish the tutorials then it is onto stat tracking, and then... maybe a new map!


    For a new map I am thinking to make something much easier, probably a mountainous area with great visibility and less foliage.

    Two benefits: easier for beginners in doing landnav, and also better performance for the lowest end hardware. So far the game is seeming to run great for most people, and at least one person with very low end hardware is able to run it... but barely. I do want for practically anybody to be able to play the game since it is educational first and foremost, so having some maps that dont demand so much GPU power ebcause of lots of foliage overdraw should help with that.


  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter


    Exciting time for the project - new content. New map, new characters, new game modes.


    Also I am adjusting direction of the project a bit. So far it has been 100% realistic simulation of military land nav training. For a select few people, this is interesting but more useful as a training aid than fun.

    I think a broader audience could enjoy navigation based gameplay, but it has to be delivered in a more exciting package. So my primary focus now is on making the game fun.

    So we are headed to the Dolomites for a career as an adventure racer. Navigation will still be the core aspect of the challenge but because we are in the mountains, it will be easy even for a beginner to make use of intuitive terrain association skills. With less brain bandwidth on hardcore map reading, we'll be able to make more meta-game decisions pertaining to survival and character maintenance.

    Anyway, all that game design stuff isn't purpose of this forum but I am slowly going to be chipping away at this dolomites level. For now I have confirmed the production pipeline for iterating on the tiled landscapes. There is a total of 16x16km, broken into 4km chunks. This seems to offer plenty of loading speed but isn't to onerous to add content to for a solo developer. I think the size is plenty big so I may restrict play area a bit, and then we still have plenty of backgroudn for epic views.

    Materials are place holder - because we are trading many trees for lots of rocks, I'll probably implement a displacement/tesselation shader to make the ground pop more - especially on talus slopes.


    The goal is to match the feel of the Cortina area in spring, seen above ^

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    Since my idea for new career adventure mode will necessitate larger maps, I wanted to make sure I'll actually be able to do the work in a reasonable timeframe. So I setup the landscape materials, setup foliage procedural spawners, and am satisfied enough with how it looks and how much time the workflow will take to populate 256sqkm. I should be able to get all the basic procedurals done in just a few days, which is important because there is some more time consuming things I want to add to support gameplay - namely roads, rivers, and a few other man-made landmarks. All to help support easier gameplay.

    I also noodled around with some post process of course. Impossible not to do that a bit. I think I like the more colorful one where the greens really pop. Just makes me feel happy.

    I am also happy with the definition on teh mountain slopes. I would like to go and manually place meshes but I think that would be massive time sink given how many mountains there will be. And I don't think these look half bad. Looks more or less like actual rock.

    Well, the character of the mountains doesn't scream Dolomites but I am not going to go crazy about it - after all I have the whole game to make so I'll have to compromise somewhere. But if I can get the colors to match I think that will be nice enough.

    In next few days I just stress test with the full size world, make sure no surprises, then back into gameplay logic.



  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    Couldnt help putting the coding off for oooone more day to work on the art. Actually it isn't total procrastination - I am going to make a pitch for some grants soon so I wanted a little more variety to show in videos.

    Anyway, I think I did manage to get a little closer to that distinctive Dolomites type of rock - without manually placing any meshes, just relying on the landscape shader and the heightmap from World Creator.

    I actually resorted to some obscure filters in teh "alien worlds" tab or whatever it's called. The thing is, most filters do something to subtract from the geometry, but I needed to sort of build up the verticality plus get more sharp ridges.

    This is making use of 2k heightmaps, and there is quite a lot of them so it really increases size of the project unfortunately. But it's still small than almost any AAA game so I guess I don't need to worry about it.


  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    There is a lot of information out there regarding the world composition tools in UE4 that seems wrong to me.

    I just did some test today comparing a single 8km terrain compared to one broken into many smaller tiles and streamed using World Composition, and the performance is at least 3x better on average compared to a single terrain.

    Just putting that out there - I think a lot of the info written is just people repeating things they heard elsewhere and didn't actually test. Or they are trying to sell you something...

    I still can't figure out why people consider world composition so broken and unusable. It's pretty straight forward and easy to be honest. Perhaps for a large team it adds complication to the version control workflow? That's not something I have to deal with.



    You can see that stats there. I have it limited to 120fps and it stays there. I stream the tiles at 3km distance. You could do less and hide it with fog if you needed. The landscape lod is set like this:


    Anyway, I just wanted to share because if you do any googling to find info about open world level streaming in ue4, there is a lot of people confidently saying that it is broken and unusable. I did some test and that just doesn't seem to be the case.

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    Real happy with new map so far. I got many coding/gameplay things to do in coming months but it will be a happy day when i can just go through this map and beautify it.

    I am getting things ready to make a pitch for some grant money. If I am lucky and get some I may try to get a little help to make some nice environment props. A few rifugios, unique trail markers, stuff like that. I'm kinda dreading making different clothing and backpacks too, but that would be too expensive to hire out. So I'll just have to fucking do it I guess.

    After some play test I decided to reduce map size from 16km to 8km. I think it would take hundred hours or more before you really got to know an 8km area because of lots of backtracking. And at that point probably better just to offer more maps than more of the same.

    Smaller map decreases iteration time a lot. Less time waiting on moving things back and forth.

    Like mentioned previously, I found a massive performance gain by loading lots of 2km chunks rather than one 8km chunk - despite what articles I read about performance. I am not sure what the difference is except that with a single terrain there is many times more primitives drawn. I guess that is the total triangles drawn? I thought unseen components of the terrain would be culled but perhaps not. In any case, now I am able to get the terrain looking as good as I can and there is practically no performance hit.

    I also did a better job in world creator creating the various heatmaps to control the texture placement. I struggled to get realstic talus placement before because I didn't realize you could output the talus simulation as a mask. It's a little convoluted workflow but basically you can assign a simulation as a mask for a texture layer.


    A lot of the survival gameplay stuff involved making a new inventory and UI. In an effort to make the game more braodly appealing and approachable I went with a more clean and colorful design. Well, design means I made up something new as I went along but I think it gets the job done for a stock-only UI. All items are placeholder images of course.


  • teodar23
    Offline / Send Message
    teodar23 sublime tool

    Hey man,

    Good progress so far, i've been following for a while.

    One thing that bugs me about the terrain is the visible edge on the right side

    Maybe you can use a background mesh for very distant landscape that will give the idea that the terrain doesnt abruptly end.


    Also, i get the feeling that the terrain z scale is off and could be a lot smaller.

    Keep up the good work!

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    Hey you are right. that is near the edge of the map and i havent made background meshes yet. But I will eventually.


    And you are right about the z scale being a bit overblown. I mainly did that for gameplay purposes. Otherwsie getting over mountains feels too fast/easy.

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    LANDNAV dev update - YouTube

    some new video showing new gameplay. I've also refined graphics a bit including the UI too. No sound, and may take youtube awhile to process full HD version.


    Much work in previous months has been programming related but now it is more focused on gameplay and soon I'll be working on new maps + polishing existing maps. Very nice to go through and add unique touches to each area.

  • Alex_J
    Offline / Send Message
    Alex_J veteran polycounter

    I’ve been overhauling the UI, among other things.

    I 've honed in on a consistent style a little more. My goal is to keep it as clean and simple as possible. Just like when you are doing landnav in real life, you want a simple plan without extraneous details to get in your way and cause confusion. So I try to make the UI reflect that mindset.

    I'd done a lot of back and forth on the general theme of the game. I didn't want to lock it down to military theme because landnav is not only a military thing. I then considered some generic simulator themes with flat, clean appearance and that's alright but it wasn't quite there. I think I've settled on something I'll stick with now.

    A bold orange diagonal immediately gets attention of any orienteering enthusiast, and also just generally feels exciting I think. Landnav is a sport where you make bold decisions fast, so I think this diagonal shape helps communicate that.

    For the most part the UI is clear, fat text. My idea is to keep it as minimal and focused as possible, but also make sure its easy to see. So I am using pretty big fat fonts with an outline and drop shadows. In some cases I add a slight background darkening and/or blur but I try to avoid that because if the screen gets too busy I feel that goes against the overall style. To do good in LANDNAV you have to have laser focus - you can't get distracted or lose your train of thought, so I want the UI to adhere to this same principle.

    I've added juice to all the UI buttons. Just simple stuff but it makes a big difference in making it feel more like a polished game.

    Well, a bit more coding to do but pretty soon I'll work on another new map and then it will be final polishing phase.

2
Sign In or Register to comment.