Announcement

Collapse
No announcement yet.

New Unfinished Map: TH-Plateau Jumeau

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • New Unfinished Map: TH-Plateau Jumeau

    Hey all,
    All right, I'm probably on the wrong board...my apologies. I thought if I posted this in the mapping section noone would notice. Like I was telling DP, I'm not sure I'm going to be able to finish this sucker any time soon. I was laid-off during the summer and I'm too busy trying to survive right now.
    Just thought, WTF, and decided I wanted some feedback. Problem is you will only be able to go into the map as a thief and there are no Guards, no loot, and no winning conditions. Sucks eh? Ah well, if your interested just go for a stroll through the map and check it out. If your an expert mapper, or even a semi-competent mapper, then give me some constructive criticism.... hell, just tell me what you want no matter what. There's lots to learn from here - especially what not to do when mapping.
    I would like to say that the best advice I've been given so far was from Nachimir long ago (not remarking on my map) who said, "Best thing to do is make your first map.... and then ditch it.... alot of first maps aren't worth trying to save. What you learn on your first will translate over to your next map." Or something like that....
    So Nachimir, if your reading this, I'd like you to check out whats done so far and leave some comments if you've got the time, thanks.

    Anyhow have fun all, I know its not much now but maybe sometime when I'm in better shape and have the cash for a comp that doesn't argue with me.......

    ( oh, and try to find some catfall somewhere before you start running around the joint, you might find it handy in spots.)

    Have Fun!!!
    ...and all is silent, save the voice of the clock...

  • #2
    Oooopps!
    erm... go get the rather large file (over 3MB compressed/ 13 MB unzipped) at Tears of Blood maps section.
    ...and all is silent, save the voice of the clock...

    Comment


    • #3
      hmm i think the best way to learn "what not to do" when mapping is going to Cranky Steves on SomethingAwful ^^
      My leet Thievery Map
      My leet UT3 Map
      My leet AS Map

      Comment


      • #4
        if you want to go guard do what I did and summon tknight in console

        Great map thought, it's certainly got lots of promise. Very like the lost city thief2 map (never seen thief tdp's lost city so cant compare with that). When I got the warning about guardians- scared- i expected zombies to suddenly come attacking me :grin:

        Your advice about catfalls is true i went guard ran a few meters and cratered :lol: oh and that spiral staircase makes people feel sick but nice map- now start on version 2
        [E.D.G] "Eaves Droppers Guild - we're on the edge."

        All New Thievery-O's!

        Comment


        • #5
          Apologies for the unanticipated delay. I had to reinstall UT, and there was a minor hiccup, which is still unsolved.

          I tested the map offline, and see one major problem technically: it's fecking huge!

          For optimal performance, your node:polys ratio should be around 2.5, as it stands 2.3 is close enough. However, the large open space drives occlusion time way up past 40 in some areas, when it should be no higher than 10.

          (in case you don't know the relevant commands:
          stat occlusion
          stat zone
          stat fps
          )

          The nodes:polys ratio can be found under the statistics tab of the build options.

          It's a nice idea, but wide open spaces don't work well at all in UT1. The larger an area you're working in, the less detail you'll be able to pack in.

          By way of comparison: the final version of Breandor has almost the same number of brushes and zones as your map. The zones are much more restricted than Plateau Jumeau, yet some people (including me) still can't run it too well out in the gardens.

          To explain a bit about zones, occlusion and CSG:
          Zones help the engine to decide what to show and what not to. Rather than checking every surface and actor in the map, the game will first check zones. It will decide what to render and what not to (what is occluded) in the zone occupied by the player, and will also look at all visible portals and make the same decisions for each visble zone.

          Occlusion time is how long it takes to make all of these decisions, so a lower occlusion time is better, (roughly) above 10 means it's going to start noticeably dropping frames. Having more polys and nodes on show will, of course, give the renderer more data to sift through, increasing the occlusion time. Of course, your system specs also have a large effect on this. A map that runs acceptably on my system (a K6-2 500 with 32Mb of v-ram and 256MB ram) should get an extremely low occlusion time on an up to date system.

          After that, you get on to lighting, and for each surface it has to handle a lightmap in addition to the texture. Dynamic lights will slow that down some more, too... It's easy to make a map that looks fairly good unlit and seems to perform okay, but then you light it, it chugs, and you go "Oh, shit". I've certainly done that before. You have to account for actors, lights, and textures in advance while building the geometry.

          Remember that semi-solids will reduce your node count, but will also not occlude.

          A few tricks I find helpful when zoning:

          Buffer zones:
          This is simply, rather than dividing an interior and an exterior with a portal, putting a small, third zone between them, usually with a corner in it and a portal at each end. This way, whichever way someone is looking, the game only has to calculate for one complex zone and a simple one, rather than two complex ones that would each run well alone. Not a perfect solution, but it works to some extent.

          Divided exteriors:
          Mapping is simulation, meaning you can get away with a hell of a lot of unrealistic building that players will rarely see, engage with, or quibble over (Save the odd Master Builder with a crate tower ). It's overwhelmingly tempting to make a large space, then recreate the setting within it, but this almost always runs like a dog (TH-Grange is a rare exception that I'm still having some trouble figuring out: Ulukai seems to have kept the geometry quite simple, yet used good aesthetics to squeeze as much "Ooh!" as possible out of each poly and light). Go out into the snowy open space of TH-Folly and you'll see it doesn't quite work so well there.

          The first version of Breandor was a massive open space with some ruins and a manor shoved into it. You could climb over the ruins and get onto the roof (without having to use crates). The drop in performance simply wasn't worth being able to fire arrows over the house. There is almost no worse technique than making a large subtraction and dumping your map into it. The 436 UT engine is best at handling interiors. Static meshes, terrain infos, antiportals, and distance fog were all invented to allow the Unreal engine to do large exteriors with detail. To do them in UT, you have to compromise, a lot, and that compromise is usually swapping detail for size.

          The last version of Breandor had the north and south gardens completely separated and also heavily divided by large bits of geometry, with buffer zones between the gardens and main interiors, and fake backdrop walls descending from the ceiling of the map onto the top of all the highest walls and roofs. Despite all that, it still runs badly on low end systems like mine, because the garden zones are large, so learn from my example and make something better. My exteriors, despite their problems, are much, much smaller than the exterior of Plateau Jumeau

          So, I don't have many harsh words for you beyond: try something less ambitious, and it will probably be much better.

          Comment


          • #6
            Oh oh! Th-20 had a huuuuggggeee outdoor area!

            Comment


            • #7
              No shit Sherlock!

              Comment


              • #8
                Yeah, well...
                Anyways, what nach is trying to tell you: If you want to make HUGE outdoor maps, wait for nightblade.
                Ut2003 features a nifty heightfield terrain feature that allows creation of large outdoor areas in a matter of minutes... (Maybe hours...)

                Comment


                • #9
                  Thanks Nach, much appreciated! Some of what you said I anticipated and saw really early on - other parts I didn't forsee and what you've said is very clear and helpful. Your absolutely spot on. This was an ambitious undertaking for a newbie mapper. I found so much of the map was giving me problems towards the point I'm at right now, that I didn't forsee the insanity of adding guards, actors and loot, never mind the lighting, as being any deterence to framerate and occlusion, etc...until it was way too late.... :cry:
                  On the upside, I now know what I should plan for on the next map. This is something that is often sadly left out of tutorials for Newbs. An ambitious new mapper with a hot-head full of idea's hasn't got a clue of what they are about to come up against and this can be devestating after months or years of hard work on something the poor sod doesn't realize is out of his league. - (and outside of the limitations of the software.) ...untill it is much too late. I can see why ditching the map and starting fresh is probably the wisest course of action as going back to the table - and original map - for full-open-surgery is another very risky business as well. I find chopping large chunks and revamping huge subtractions very troublesome, especially when they were done early on. Whole new families of BSP and HOM's rear their ugly heads as well as overall messy chaos.

                  Thanks again Nach. Will be waiting to see what NB and a new comp have to offer me and what I've learned, (or not).
                  ...and all is silent, save the voice of the clock...

                  Comment


                  • #10
                    Glad I could help

                    Comment

                    Working...
                    X