Final compile, without final vis

Discussion in 'Mapping Questions & Discussion' started by cornontheCoD, Mar 19, 2009.

  1. cornontheCoD

    cornontheCoD L7: Fancy Member

    Messages:
    437
    Positive Ratings:
    70
    I've been trying to figure it out, but I can't figure out what I should do for a final compile, with vis still running on "fast" mode. I know I should go to expert, but I don't know what to do after that, that dialog box just confuses me.

    Also, any estimates for how much longer a final compile would take compared to a normal compile?
     
  2. Ezekel

    Ezekel L11: Posh Member

    Messages:
    818
    Positive Ratings:
    244
    don't put it on fast mode, you want it on normal.
    the difference between the 2 is that fast mode skips steps that are normally done, and as a result, i belive it means you'll only have direct lighting on the map, rather than having it bounce/reflect off of surfaces in a more realistic manner.

    regarding compile time - it depends on manner factors including the map geometry (i.e. the size of the map, how much of the map is on 16x16x16 grid, did you use func_detail intelligently, etc), your CPU speed, number of CPU cores, RAM and the amount of other programs running in the background.

    for lighting, you should also consider the amount of models and func_details as they need to be lit too. - and of course the number of light sources and the lightmaps on your brush faces.

    for me, with a 2.0ghz dual core and 2 gig of ram, i usually need between 5-20 minutes to do a compile for a release version of a late beta stage map. - but i don't compile with HDR turned on (it bloats the file size but adds a feature that, though looks really nice, is rarely noticed in the thick of battle, and custom maps need to balance filesize with other trade offs to a certain extent).
    if you are compiling with HDR turned on, then the vrad part will take double the amount of time as it has to run 2 times.
     
  3. StoneFrog

    StoneFrog L6: Sharp Member

    Messages:
    395
    Positive Ratings:
    81
    -fast VIS is dandy for quick dev compiles and possibly even Alpha and Beta versions, but it's better to do a complete VIS passover with the final rendition of the map.

    Although you may not see major differences, they are there. Maps tend to be much more optimized with the final VIS than the fast VIS. Fast VIS is sloppy in that you can get inconsistent performance that may not always really be the least bit similar to how your map is really being processed by the engine. In Goldsrc maps, we used r_speeds to benchmark a map and how it's rendered, and sometimes a terribly laggy map compiled with fast VIS becomes quite playable with final VIS.

    Also, sometimes miscellaneous little errors are also corrected with final VIS, such as displacement anomalies and whatnot.

    You should take the time and do the final VIS compile. I think I heard once long ago that it takes "2.5 times as long as regular (or was it fast) VIS", but obviously your mileage may vary. :)
     
  4. Earl

    Earl L6: Sharp Member

    Messages:
    284
    Positive Ratings:
    37
    I always compile with full vis because it makes VRAD goes faster. Vis should be quick anyway, if you're func_detailing properly.
     
  5. Icarus

    aa Icarus

    Messages:
    2,246
    Positive Ratings:
    1,182
    You mean to tell me there is no difference performance-wise between fast and normal vvis?
     
  6. StoneFrog

    StoneFrog L6: Sharp Member

    Messages:
    395
    Positive Ratings:
    81
    Who are you referring to? I said in my previous post that there IS a difference between fast and normal VIS. It's like RAD compiling. The amount of time it takes could be rather daunting, but you should do it from time to time to get a good idea of how your map truly looks.

    Fast VIS is not an accurate representation of how the engine processes your map.
     
  7. cornontheCoD

    cornontheCoD L7: Fancy Member

    Messages:
    437
    Positive Ratings:
    70
    ok, maybe I should re-state it. unfortunately, my map cannot be compiled on normal vis, because it is poorly optimized. It is fine with normal RAD, but not VIS. I have accepted this fact, and continued anyway, because it would take a complete rebuild to make it optimized.

    I just want to know what settings to use for a final final compile, but with "fast" VIS
     
  8. Apom

    Apom L6: Sharp Member

    Messages:
    366
    Positive Ratings:
    62
    Your question can not be answered, as there is no final compile with fast VVIS.

    Also, VVIS should really never exceed a handful of minutes (VRAD can reach a couple of hours, that's different). Anything more indicates that your map will run at 15 FPS on most computers.
     
  9. Sgt Frag

    Sgt Frag L14: Epic Member

    Messages:
    1,443
    Positive Ratings:
    294
    There's gotta be a way to optimize you map without rebuilding.

    Sure rebuilding might make it better. But I know for instance there is that hole from B to C that you can see through. But WHY? Even if you look through there and see someone by the time you get to that area you're not going to have any idea of what you'll run into.

    Just stick a big plain building there if you think players will 'think' they should be able to see over there.

    And that can only help the laggiest part of your map. 'C'. Because you wont be rendering B also.

    If you want I can take a look, but probably wed. at the earliest. Hoenstly I think you must be missing something in optimizing. Your areas really aren't that much larger than gravelpit areas, and they are all seperated by tunnels that should block vis.

    If you want me to have a look send me a PM and have the vmf file ready to email.
     
  10. MrAlBobo

    MrAlBobo L13: Stunning Member

    Messages:
    1,059
    Positive Ratings:
    219
    i could also take a look at it...but I could always offer a compile

    I have a 4 hour shift at work on sunday...and ive got a quad core @ 3.2Ghz, it will compile, then you could use this to see if the full vis is worth the effort...
     
  11. StoneFrog

    StoneFrog L6: Sharp Member

    Messages:
    395
    Positive Ratings:
    81
    Correct. There's an old saying somewhere along the lines of, "If your RAD time is longer than your VIS time, you should be proud. If your VIS time is longer than your RAD time, you need to go back and rethink some things."

    It may be better for you to go back and optimize your map, even if it means rebuilding. In the long run it'll make the map much more tolerable when compiling, and probably a lot more enjoyable to play. It will also give you time to go back, rethink, and reorganize.

    Still, just how long is VIS taking you currently? If it's only 2 or 3 minutes, you've seen nothing. Can I get a screenshot of the top-down 2D view of your map? I may be able to judge whether or not you truly need to redo the geometry.
     
  12. cornontheCoD

    cornontheCoD L7: Fancy Member

    Messages:
    437
    Positive Ratings:
    70
    in terms of on my computer, vis literally gets stuck on portalflow. it just won't move for hours after a certain point. By now, I don't even want to imagine how long vis would take.

    Here's the situation:
    1) This map is a school project. I realize it is not good to rush a map in any way, but I am basically done with timber, I want to move on with the lessons I learned from timber and pretend that timber never happened.

    2) I want to make the map look as nice is possible when I present it. A final compile would help with this.

    3) I know that timber is not a worthwhile map to pursue. It is not that fun, so best-case scenario after rebuilding it would be that I would have an optimized map that plays mediocre and looks mediocre. I would rather start from the beginning with a new map and make sure that that map is optimized. Then I could possibly have a map worth pursuing

    thanks for offering the help guys, but I just wanted to know if I could do a final compile with fast vis
     
  13. Apom

    Apom L6: Sharp Member

    Messages:
    366
    Positive Ratings:
    62
    While fast VVIS will not look as good as normal VVIS, the difference is minimal, if even noticeable (technically, visleaf cutting has an influence on leaf ambient lighting, hence overall lighting, but that's a long shot). Its main purpose isn't beautification, but performance gain, possibly by a significant margin. If you only want to take screenshots and/or run the map without players on it, you can live without it.
     
  14. Forthex

    Forthex L2: Junior Member

    Messages:
    80
    Positive Ratings:
    12
    You absolutely NEED to read this;

    Controlling Geometry Visibility and Compile Times

    - turn as many brushes possible into func_detail (basically, anything that isn't a MAIN floor/ceiling/wall
    - use hints and area portals and use them often
    - use func_viscluster
    - avoid insanely large, open areas

    I was in the same boat. I left vvis running for an hour and it only reached 40% completion, so I canceled it and tried again with a few modifications. I placed a func_viscluster brush in the large open area of my map and vvis TOOK LESS THAN A MINUTE.
     
  15. cornontheCoD

    cornontheCoD L7: Fancy Member

    Messages:
    437
    Positive Ratings:
    70
    I've already read nearly everything on optimization, including that, but thanks for the help
     
  16. Apom

    Apom L6: Sharp Member

    Messages:
    366
    Positive Ratings:
    62
    func_viscluster is kind of a cheat. In most cases, you shouldn't use it, because if you did a good job with hint brushes and nodraws behind displacements, then you shouldn't have that many visleaves in your open areas anyways. Generally speaking, func_viscluster = good for compile time = bad for run-time performance.