Author Topic: Over Haul  (Read 2908 times)

Legacy_KlatchainCoffee

  • Sr. Member
  • ****
  • Posts: 372
  • Karma: +0/-0
Over Haul
« Reply #135 on: March 17, 2013, 05:43:12 pm »


               I have been admiring the lovely work in this thread for a while, but seeing that fox totally made my day!  Can't wait to use it in what I've been working on.

Thank you! '<img'>
               
               

               
            

Legacy_Draygoth28

  • Hero Member
  • *****
  • Posts: 574
  • Karma: +0/-0
Over Haul
« Reply #136 on: March 17, 2013, 05:45:45 pm »


               Anyone know any tricks to get around poly count? I am haveing hitching when another model is rendered in game. I think the count on the skin is around 5K...
               
               

               


                     Modifié par Draygoth28, 17 mars 2013 - 05:46 .
                     
                  


            

Legacy_OldTimeRadio

  • Hero Member
  • *****
  • Posts: 2307
  • Karma: +0/-0
Over Haul
« Reply #137 on: March 17, 2013, 05:56:41 pm »


               

Draygoth28 wrote...
Anyone know any tricks to get around poly count? I am haveing hitching when another model is rendered in game. I think the count on the skin is around 5K...

Yeah, you're going to want to compile your models in advance.  Here's a kit with a couple of model compilers and Cleanmodels.  You're probably going to want to use the DLA model compiler for this (the Bioware one is superb for regular geometry but often borks skin mesh items).

What I generally do is a command line like this:
nwnmdlcomp.exe -cn mymodel.ASC mymodel.mdl

I copy the ASCII file from my override directory (where I export with NWMax from) to the compiler directory and change the extension to ASC prior to compiling.  If people want to edit the models to make changes, best they use the ascii model that came out of NWMax but for players to play with, compiled models are superior on a number of counts.  One of those benefits is that a compiled model can be many times the size and complexity of a normal model and yet load almost instantaneously, as compared to the lag during run-time compilation that most are used to.

Love the fox, btw, and I'm glad to see you still working on this wonderful project!
               
               

               
            

Legacy_Draygoth28

  • Hero Member
  • *****
  • Posts: 574
  • Karma: +0/-0
Over Haul
« Reply #138 on: March 18, 2013, 12:42:44 am »


               Thanks OTR, I was hoping to have not hit a dead end. I have some really nice mounts I want to release in the near future but the poly count is killing them.
*I can't seem to get this to work for me. I am more than likely doing it incorrectly. The compiler only flashes open then closes. Any help is appreciated.
               
               

               
            

Legacy_OldTimeRadio

  • Hero Member
  • *****
  • Posts: 2307
  • Karma: +0/-0
Over Haul
« Reply #139 on: March 18, 2013, 01:01:54 am »


               

Draygoth28 wrote...
Thanks OTR, I was hoping to have not hit a dead end. I have some really nice mounts I want to release in the near future but the poly count is killing them.

If we're just talking about a complex skin mesh, especially a model which was made with video game usage in mind, then compiling them should be the answer to the lag.  If they have complex shadow-casting geometry (i.e. a complex mesh which has Shadow ticked on in the Aurora Trimesh Modifier and which is not a skin mesh- think a separate mesh possibly not associated with the skin, like a pair of horns), then there are still going to be problems or worse, slowdowns which people might not be able to attribute to that model.  Really, really good not to have shadows on meshes like this unless you're willing to put what can amount to quite a bit of work in to prevent the big hit that shadow casting objects cause with any level of complexity.

*I can't seem to get this to work for me. I am more than likely doing it incorrectly. The compiler only flashes open then closes. Any help is appreciated.

NWNMdlcomp.exe needs to be run from the command line.  In Windows, hit windows key - R to bring up the run menu, type cmd and hit return.  From there, navigate to the directory where nwnmdlcomp.exe resides and use the command line I provided in the prior message to run it.
               
               

               
            

Legacy_Tarot Redhand

  • Hero Member
  • *****
  • Posts: 4165
  • Karma: +0/-0
Over Haul
« Reply #140 on: March 18, 2013, 01:28:32 am »


               I tend to use a couple of batch files that I've set up for the jobs of compiling and decompiling. The compile one contains ->

nwnmdlcomp -c Decompiled\\* Compiled\\

while the decompile one has ->

nwnmdlcomp -d Compiled\\* decompiled\\

These 2 batch files are placed in the NWN Model Compiler directory above the compiled and decompiled directories. I also use Ant Renamer Portable for renaming a number of models all in one go.

TR
               
               

               
            

Legacy_Draygoth28

  • Hero Member
  • *****
  • Posts: 574
  • Karma: +0/-0
Over Haul
« Reply #141 on: March 18, 2013, 01:31:27 am »


               when i open it , it says 'nwnmdlcomp.exe' is not recognized as an internal or external command,operable program or batch file.
ok now it says unable to locate super model "h_ba".
               
               

               


                     Modifié par Draygoth28, 18 mars 2013 - 01:57 .
                     
                  


            

Legacy_Tarot Redhand

  • Hero Member
  • *****
  • Posts: 4165
  • Karma: +0/-0
Over Haul
« Reply #142 on: March 18, 2013, 02:31:39 am »


               As it works for me, I'll expand (hopefully without appearing to be condescending, I'm just trying to be thorough here) on what I said above. Create the 2 batch files above and place them in the actual NWN Model Compiler Directory. This directory should contain 3 sub-directories - Compiled, Decompiled and OldVersion. Apart from the sub-directories there should be the following files - nwnmdlcomp.exe, readme.html, Updated.txt and (once you've created them) your 2 batch files. Place your model to compile in the Decompiled directory, appending '.ascii' after the '.mdl' extension. Double click on the batch file containing the text 'nwnmdlcomp -c Decompiled\\* Compiled\\'. You should now find the compiled version of your model in the Compiled sub-directory.

This has worked for me with both XP and vista.

TR
               
               

               
            

Legacy_Draygoth28

  • Hero Member
  • *****
  • Posts: 574
  • Karma: +0/-0
Over Haul
« Reply #143 on: March 18, 2013, 02:57:45 am »


               i created the batch files, and compiler opens. It still says unable to locate super model "h_ba". and i get no compiled model.
               
               

               
            

Legacy_The Amethyst Dragon

  • Hero Member
  • *****
  • Posts: 2981
  • Karma: +0/-0
Over Haul
« Reply #144 on: March 18, 2013, 07:09:32 am »


               I've found that I often have to have the model for the supermodel in the same directory as the model compiler.
               
               

               
            

Legacy_Tarot Redhand

  • Hero Member
  • *****
  • Posts: 4165
  • Karma: +0/-0
Over Haul
« Reply #145 on: March 18, 2013, 10:56:57 am »


               While I am glad that the batch files now work, I am sorry you still have problems. Problems I am definitely not qualified to help with as all my stuff is much simpler and doesn't have a super model. I hope AD's solution works.

TR
               
               

               
            

Legacy_OldTimeRadio

  • Hero Member
  • *****
  • Posts: 2307
  • Karma: +0/-0
Over Haul
« Reply #146 on: March 19, 2013, 12:49:21 am »


               Sorry for the tardy reply, Draygoth!

Okay, as Amethyst Dragon indicated, the h_* supermodel sereis are going to need to be included.  Their decompiled, ASCII forms.  I believe they're natively compiled, anyway.  That's going to be a bit of a pain in the butt for a couple of different reasons.  Again, while it's preferred to have models compiled in advance nomatter what, the goal here is shaving off loading time.  Any model which hooks into h_ba is going to cause at least a little hiccup because of the (relatively) enormous size fo the h_ba animations to begin with.  However, if you already have mounts in your scene (which certainly already plug into h_ba), any additional loading time lag is probably due to on the fly compilation- because the animations are already in memory.

Just wanted to get that out of the way.

Crash course, two different routes.

First, give your model a shot with the Bioware compiler.  The Bioware compiler will almost certainly bork it, but it's worth a shot.  Just copy the model into the "in" directory, run the processmodels.bat file and check in the "out" directory.  Also look at processmodels.log.  If there is any error at all, it'll probably be a "**MAJOR ERROR**".  Fortunately, processmodels is very good at sniffing out any problems, inconsistencies or lack of "best practices" on any models it attempts ot compile.  Unfortunately, it throws up its hands for pretty much the same way for major errors as well as minor ones, reporting the same inaccurate message in the log for both.  If your model DOES compile, copy it to your override and test it out.  If it works, you're lucky.  There's a pretty good chance it won't, though.  Processmodels and skin mesh (and some types of animation) don't seem to work right.

Second, adding the supermodel chain to either the Bioware processmodels "in" folder to the "DLA" model compiler's model folder.  As Amethyst Dragon mentioned, you're going to need ASCII versions of the h_ba model series present at compile time.  The reason for this is the deep down way supermodeling appears to work.  More can be learned about that here (from Torlack), BTW.

Here is a list of the h_ba supermodel chain:
h_ba.mdl
h_ba_casts.mdl
h_ba_coat.mdl
h_ba_custom.mdl
h_ba_med_weap.mdl
h_ba_non_combat.mdl


I think that's it.  This ful chain of models will likely have to be placed in the directory for compile.  I had forgotten this appears to be the one time where compiling models in advance can get really...irritating.  Usually it's so easy it's not even funny.  On the assumption that the compile goes through with the DLA model compiler with these additional files, what you want to do is to test your compiled model (probably by copying it to the override directory) but  you do not want to test with the resulting compiled h_ba* models into your override or distribute them.  Again, don't do that. 

If all goes well, your fox model should compile fine and work well, standalone.
               
               

               


                     Modifié par OldTimeRadio, 19 mars 2013 - 12:54 .
                     
                  


            

Legacy_Draygoth28

  • Hero Member
  • *****
  • Posts: 574
  • Karma: +0/-0
Over Haul
« Reply #147 on: March 19, 2013, 10:26:13 am »


               If I can figure this out them I will be able to add several updated mounts to the game.
'Posted
               
               

               


                     Modifié par Draygoth28, 19 mars 2013 - 10:26 .
                     
                  


            

Legacy_Hekatoncheires

  • Full Member
  • ***
  • Posts: 219
  • Karma: +0/-0
Over Haul
« Reply #148 on: March 19, 2013, 07:44:17 pm »


               HEPHALUMPS!!1!

omgomgomg
               
               

               


                     Modifié par Hekatoncheires, 19 mars 2013 - 07:44 .
                     
                  


            

Legacy_The Amethyst Dragon

  • Hero Member
  • *****
  • Posts: 2981
  • Karma: +0/-0
Over Haul
« Reply #149 on: March 19, 2013, 08:40:29 pm »


               That elephant and camel are superb looking, and NWN has needed good ones for a long time.  With those, I might be able to build out my PW's "African savannah type kingdom", and my desert empire would be perfect for camels.