[CS:GO] de_coredump

    Hey all. I've been mapping on and off since ~2011 but this is the first time I've deemed one of my maps "good enough for release" (i.e. I can't stand the sight of it any more). I've never been to Mapcore before either, but this froody dude on Twitter suggested it as a good way to get feedback, which (as you can probably guess) I'm in dire need of. I'll try not to be 'that guy' who shows up, plugs his stuff, and leaves forever.

    So, Core Dump. Most of my earlier maps had disastrously poorly thought-out layouts, so for this one I tried to dial things back and went for a safer, Mirage-y layout, albeit with a few twists that I hope go down well. I wanted to go for a kind of original aesthetic (well, original in the sense that I've yet to see a map that does it) but I can't make textures or models to save my life, so I was working with whatever I could pilfer from royalty-free sources or slap together in Paint.NET. It's also terribly optimised, because... well, I don't really know what I'm doing. This map languished almost-finished in my folder for a few months—hence the lack of Newke assets—before I finally decided to just put it out there and accept its shortcomings. I've moved on a bit since making this, but I'll still treasure any and all feedback.

    Anyway, pics.

    If you fancy a runaround, you can find it on the Workshop here.

    Cool, you signed up!

    I'll try to remind myself to check it out ingame tonight.

    From the screenshots I'd suggest more player-level lighting as there a few patchy-dark areas and other areas look bland due to this.

    P.s - minor point but I'd have the overview in the workshop screenshots. People like when that's done..

    Quote from Vaya

    16 hours ago, Vaya said: Cool, you signed up!

    I'll try to remind myself to check it out ingame tonight.

    From the screenshots I'd suggest more player-level lighting as there a few patchy-dark areas and other areas look bland due to this.

    Yeah, having more interesting lighting is definitely something I need to practice at. When I recognise that a part of the map is visually dull I tend to reflexively try to populate it with more props/overlays/brushwork without considering other options. In terms of readability I thought everything was more or less sufficiently illuminated though. Anywhere in particular you feel is too dark?

    Quote from Vaya

    16 hours ago, Vaya said: P.s - minor point but I'd have the overview in the workshop screenshots. People like when that's done..

    Ooh, good call. Unfortunately the overview is a bit scrappy at the moment (some areas don't show up properly/at all) but I've chucked it up into the image rotation for now. Should give people at least some idea of what they're in for.

    I love the theme on this map and I think I've worked out your optimisation problems

    Theme - strong deusEx vibes which I like!

    optimisation - as a first point of call you quite simply you need to func_detail a lot more of your building facades. Lots of messy visleafs being made because building fronts aren't flush cubes.

    Texturing - a little mushy with the same colours used too much/ Obviously the levels theme means too much variation would look out of place but some detailing is being lost due to this- for example the balconies beside B. I also noticed in one place you're using a texture as a trim right beside somewhere else where you're using as a floor. I think adding a little colour (limited pallet of course) would help differentiate areas.

    [Blocked Image: http://i.imgur.com/rpn1FJy.jpg]

    Also - some textures have been stretched to point of blurliness - like the crate on A site. It might add to the deusEx look a little but plz no..:P


    Quote

    Yeah, having more interesting lighting is definitely something I need to practice at. When I recognise that a part of the map is visually dull I tend to reflexively try to populate it with more props/overlays/brushwork without considering other options. In terms of readability I thought everything was more or less sufficiently illuminated though. Anywhere in particular you feel is too dark?

    CT spawn is a good example of what needs fixing. I would replace the dead car with an alive one (with headlights pointed at the wall) at an off angle. Easy way of making the area lighting more intresting..

    I think it's less about visibility and more about guiding the players - just now you have super illuminated areas where I can't tell where the lighting is coming from and then you have kinda-muddy dark areas that could benefit from more overhead lighting. Maybe add a couple street lamps etc- just to guide the players eyes more. For example - maybe in the screenshot below replace the lamps with more head duty ones and bump up their brightness.

    [Blocked Image: http://i.imgur.com/OCKOAUY.jpg]

    [Blocked Image: http://dailyphotoblog.net/media/1819/back-alley-night.jpg?preset=fullsize]

    [Blocked Image: http://orig02.deviantart.net/e077/f/2011/187/f/d/public_alley_101_hdr_by_andrew_23-d3l0qk1.jpg]


    sorry for paint annotations :)

    Quote from Vaya

    10 minutes ago, Vaya said: I love the theme on this map and I think I've worked out your optimisation problems

    Theme - strong deusEx vibes which I like!

    Haha, thanks. I didn't want to go overboard on the cyberpunk aesthetic because I still wanted it to make a believable contemporary setting, but I'm glad the primary inspiration still shines through. I hope the references weren't too heavy-handed.

    Quote

    10 minutes ago, Vaya said: optimisation - as a first point of call you quite simply you need to func_detail a lot more of your building facades. Lots of messy visleafs being made because building fronts aren't flush cubes.

    This is something I struggled with a lot, mostly because I wasn't sure what was the 'right' approach. I assumed that the extra visleafs wouldn't matter too much as long as I closed them off with a hint surface. It'll be tricky to refactor the geometry, but probably worth it.

    Quote

    10 minutes ago, Vaya said: Texturing - a little mushy with the same colours used too much/ Obviously the levels theme means too much variation would look out of place but some detailing is being lost due to this- for example the balconies beside B. I also noticed in one place you're using a texture as a trim right beside somewhere else where you're using as a floor. I think adding a little colour (limited pallet of course) would help differentiate areas.

    [Blocked Image: http://i.imgur.com/rpn1FJy.jpg]

    Also - some textures have been stretched to point of blurliness - like the crate on A site. It might add to the deusEx look a little but plz no..:P

    I can actually remember joking while making this map that I never wanted to see another concrete texture as long as I lived :P. You're right, the theme makes variation difficult. I tried to combat this by altering architectural styles from building to building, but obviously there's only so much that can do. I was also a little unsure and apprehensive about making a night-themed map, so I hesitated to experiment with darker or more colourful surfaces in case they hurt readability too much. I'll give it another shot.

    Interestingly, the two surfaces in your screenshot really are two different textures (milwall004 and tilefloor009a), but I can see how their similarity in close proximity might look pretty sloppy. I was never really happy with the crate on site A, but I had to make it out of brushes and there aren't many good 'crate' textures to choose from (since they're mostly props these days anyway).

    Quote

    10 minutes ago, Vaya said: CT spawn is a good example of what needs fixing. I would replace the dead car with an alive one (with headlights pointed at the wall) at an off angle. Easy way of making the area lighting more intresting..

    I think it's less about visibility and more about guiding the players - just now you have super illuminated areas where I can't tell where the lighting is coming from and then you have kinda-muddy dark areas that could benefit from more overhead lighting. Maybe add a couple street lamps etc- just to guide the players eyes more. For example - maybe in the screenshot below replace the lamps with more head duty ones and bump up their brightness.


    sorry for paint annotations :)

    Sounds good to me. I felt like I was going a bit overboard on streetlamps at times, but maybe I just need to make the edges of their spotlight cones more narrow/distinct so all the light sources don't bleed together. Your paint annotations were helpful, don't sweat it!

    Newbie question: if I go ahead and implement some of the suggested fixes (not likely in the near future, I have uni work and another map to finish) would it still be okay to post updates in this thread? I know that this board is for more or less 'complete' works—and in my defense, I thought it was—but it's evident that I still have a lot of work to do.

    Again, thanks for the feedback.

    I would say it's fine as long as you're not constantly bumping it. If you're doing that it should probably be moved to the 3D forum ( @Sprony or @FMPONE normally do this I think) as it's still clearly a WIP.


    In regards to the building facades:

    Quote

    Quote I assumed that the extra visleafs wouldn't matter too much as long as I closed them off with a hint surface. It'll be tricky to refactor the geometry, but probably worth it.

    @will2k is probably the best person to confirm this but I *think* lots of extra visleafs means that source needs to work harder to render the stuff onscreen- it means more complicated visclusters and its spending more time working out what can and can't be seen when a player is moving around the level. World brushes should all be cube-like and flat as possible- extra details should be func_detailed where possible to avoid too many splits. All a hint brush does is split the visleaf at point of intersection- kinda helping source split up the level in a logician way - it's not really a mask for complex visleafs :P

    Thanks Vaya for weighing in with all the info :)

    @Trashbang: read my optimization papers in my signature below; 2 specific articles should clear up all your questions about visleaves and PVS.

    Demystifying Source Engine Visleaves

    Source Engine PVS - A Closer Look

    I'll quote one line of the first article

    Quote

    Quote The immediate conclusion you can draw from all this explanation is that the smaller the PVS, the better fps you will have. A smaller PVS means fewer visleaves to draw and fewer visleaves’ content to render (brushes, textures, props, entities…).

    Unless the extra visleaves are practically empty (such as upper skybox leaves), additional visleaves means more content to render, therefore in general, lower fps.

    For hints/areaportals, check my 2 papers on "practical guide" series as well as my paper Man vs Engine for the full scoop.

    Knock yourself out and enjoy the read :D

    Hey, thanks @will2k. Unfortunately I didn't pick up much from those articles that I wasn't already aware of by the end of making this map (apart from vvis's exact calculation methods, which were always something of a black box to me) but stressing the importance of horizontal leaves has certainly spurred me to go back and see if I can't work a few more in somewhere. I'll definitely peruse Man vs Engine when I get the chance, too.

    Quote from will2k

    37 minutes ago, will2k said: Unless the extra visleaves are practically empty (such as upper skybox leaves), additional visleaves means more content to render, therefore in general, lower fps.

    Do additional visleafs still matter if the amount of content rendered is the same either way, though? Obviously a more complex PVS of the same size will take longer to compile, but would it adversely affect in-game performance? @Vaya's criticism was that I wasn't func_detailing the facades of my buildings, and I'll absolutely agree that it's bad practice to have all those indents everywhere, but I don't see how it changes the amount of content in the PVS.

    Quote from Trashbang

    5 hours ago, Trashbang said: Do additional visleafs still matter if the amount of content rendered is the same either way, though? Obviously a more complex PVS of the same size will take longer to compile, but would it adversely affect in-game performance? @Vaya's criticism was that I wasn't func_detailing the facades of my buildings, and I'll absolutely agree that it's bad practice to have all those indents everywhere, but I don't see how it changes the amount of content in the PVS.

    IF (that's a capital if) your map is tightly optimized then there will be a slight dip in fps if you leave everything as world brushes; but if your map is not well optimized then expect a bigger hit.

    I have already tackled this issue in a previous paper of mine Comparative fps study in Source Engine Optimization System and you can check in the paragraph titled func_detail that the fps dropped around 10 fps when I switched my func_detail back into world brushes in a well optimized map. You are basically forcing the rendering engine to search the BSP tree with a bigger and messier PVS to render the content of leaves tagged 1 in the PVS array. The smaller and simpler the PVS, the quicker the engine can draw the information.

    When you start implementing hints and areaportals, your visleaves will be divided and if you already had a big number of leaves, then expect the number to even grow much bigger, further adding to the above effect.

    Lastly vvis will likely take forever to finish and you will rage quit long before the end :D

    Vaya gave you a nice pointer about func_detail :); don't be afraid if large chunks of the level are func_detail as long as you have backbone world brushes to cut visibility, help implement hints/areaportals/skybox brushes, and seal the level. All this is nicely explained in my paper Man vs Engine (paragraph III.5) and in my article about common misconceptions in Source optimization.

    Hope this helps and good luck with the map :)

    One thing that sticks out to me is the server farm sitting in that outdoorsy garage area, seems kind of out of place for that sort of setting. But that's just me nitpicking.

    What a cool concept for a level. It reminds me of Kane and Lynch 2 without the asian setting (or crazy lens flares).

    I like the idea of a rainy nighttime level a whole lot Its too bad the rain solution in source is so limited. I think you could sell the rainy feel alot better with low reflection on the ground and high reflection on puddles. And add distance fog! I know its always a tricky balance between play ability and art when it comes to competitive levels but its so tempting to take it slightly further with this one =)

    Quote from spa

    On 4/25/2016 at 3:29 PM, spa said: I like the idea of a rainy nighttime level a whole lot Its too bad the rain solution in source is so limited. I think you could sell the rainy feel alot better with low reflection on the ground and high reflection on puddles. And add distance fog! I know its always a tricky balance between play ability and art when it comes to competitive levels but its so tempting to take it slightly further with this one =)

    Yeah, rain is tricky and I agonised a fair bit over how dense I could take it without obstructing visibility. Ideally I'd also like to make custom soundscapes so it sounds muted indoors, but I'd have to research that first. I'll definitely experiment with reflectivity when I get around to updating it, though again, that's something that I'm new to.

    Unfortunately I have another tricky mapping project that I need to get around to releasing ASAP, so most of the suggestions in this thread haven't been implemented yet, but I'll certainly be touching it up a bit when the time comes :)

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!

Register Yourself Login