What are you working on?

[leileilol]

DP was never slow for me. I upgraded from a GeforceFX a long time ago.

DP's deferred rendering + shadowmapping = very very fast even the software renderer is quick.

Also we should be using/supporting IQM more. DP's iqm support is perfect and is fully supported, and i'd still use .framegroups anyway

[toneddu2000]

yeah I like iqm too, it's really complete, but, at least for me, the only problem is that it "scrambles" animations strips order, even with framegroup file. But maybe is a problem of the stupid quakec code I created to move characters!

[Spike]

Actually, AFAIK file formats cannot be copyrighted/patented; that's why LibreOffice can feature MS Office compatibility, BTW.

Its better than that. There are certain things which cannot be copyrighted, including facts. Structs and memory layouts and constants and things are mere statements of fact and thus not copyrightable (though I'm not entirely sure if that includes the names of members - it doesn't include logic in #defines, so you do still have to be careful).
You'll see this with all the mingw headers for ms-centric apis.
Supposedly raster fonts are not copyrightable either.
I'm not a lawyer though, and different countries have different rules.

Either way with valve, its not a case of whether you're doing it legally, but whether you can afford to defend yourself - valve do not need to actually win any hyperthetical court case. All they have to do is to make you too bankrupt to be able to defend yourself in court and they win.

The Android case should be fun. If oracle do manage to somehow win, quite a few things will need to change.

[Baker]

Google is doing some pretty fine work in the courts to establish "freedom".

Google being an information company is kind of aligned with sensible fair-use laws, reasonable and common sense interpretations of various laws and have the financial backing to do it in the courts with great legal counsel.

The Oracle thing is one example. Google has fought RIAA/MPAA silly wannabee interpretations of copyright in relationship to YouTube. Obviously they have been to the courthouses several times on things like "fair use" or how linking to files or within a web site is just part of the internet.

The deal going on with Oracle is a bit silly, a programming language is not a creative work and if it is, what about compilers

Either way, Google is getting some great case-law into at least the US legal system. I can't say I know too much about silliness it has had to deal with in Europe. Same story with Wikipedia helping more sensible interpretations of law and establishing precedent. Google does have financial incentive for reasonable interpretations of IP law, obviously, but they keep moving the bar forward and with a great deal of case-law established, judges have existing legal decisions they can look to for guidance.

[Baker]

-------

I've been mapping lately. Trying to figure out exactly what I "hate" about trying to natively use replacement textures through the Quake map format.

And converting textures like with something like textures2quake interferes with natural "make stuff on a whim" trains of thought.

What I would like to be able to do is have a few folders of textures sitting around that I can click something and it updates a few WAD files and creates appropriate pak files of replacement textures. So all you are really doing is using whatever image editor you like and those images automagically find their way into WAD files and into a replacement texture paks in the quake folder.

I don't even like opening, say, TexMex and importing images into a WAD. That doesn't feel spontaneous. I'd like to just be to copy an image in an image editor, modify it or whatever or make a new one, save it and more or less some automatic brainless zero-effect process handles things from there.

[goldenboy]

toneddu2000: Radiant is very usable for small indie teams / unpaid teams to make that level of game. Crysis isn't really an apt comparison since a small team will not have the resources to make Crysis anyway. While you can't use radiant to make Crysis, you can use radiant to make good games.

Regarding engine performance, I recommend to "timedemo demo2" various engines if you're interested and compare, it's not rocket science.

[mh]

Of the demos...

demo1 is the particle monster. It's capable of spawning maybe twice as much (coupla years since I did counts so can't remember) particles as the others.
demo2 has underwater effects that may skew the result depending on how the engine handles underwater stuff
demo3 is good for dynamic lights; lots of them going off.

None of them are good as an absolute benchmark. They all serve to measure different things, and depending on where an engine's primary bottleneck is you may get wildly different results with each. Do all 3 and get a more balanced biewpoint. Also recommend the big open scene when you come out of the first cave in Marcher Fortress for benching brush and brushmodel performance, ne_tower is a good engine crusher for somewhat general stuff, nothing (outside of some - as yet - unreleased RMQ maps) is good for real dynamic light benchmarking.

[WickedShell]

demo1 is the particle monster. It's capable of spawning maybe twice as much (coupla years since I did counts so can't remember) particles as the others.

demo1 has up to 4096 particles on the go at any one time.
demo2 has up to 2048 particles on the go at any one time.
demo3 has up to 3072 particles on the go at any one time.

http://mhquake.blogspot.com/2008/12/par ... -more.html

I happened to read that yesterday...

[Spike]

Actually, in vanilla quake (end even fte it seems), the number of particles is dependant upon framerate, as their rate of decay/change is in realtime, not demo time. The higher your framerate, the more particles at once, and the more time (proportionally) the particles will take to process.
This is a bigger issue with NQ demos than QW demos, due to multicast at least culling a few in QW.

demo2 generally isn't the best demo to use for benchmarking general performance, but it is good for peek performance due to its simplicity. Peek performance should not be the reason you'd use an engine.

If you want to benchmark an engine properly, you should play the demo normally and compare the slowest frame.

[ceriux]

speaking of particles and probably really off topic, i really like quakes square particles i think the round really look like crap, same with the more "modern" ones who look like exploded glowing jello.

[mh]

demo1 is the particle monster. It's capable of spawning maybe twice as much (coupla years since I did counts so can't remember) particles as the others.

demo1 has up to 4096 particles on the go at any one time.
demo2 has up to 2048 particles on the go at any one time.
demo3 has up to 3072 particles on the go at any one time.

http://mhquake.blogspot.com/2008/12/par ... -more.html

I happened to read that yesterday...

4 years. Time really flies.

The odd-looking multiples-of-1024 numbers may lead you to suspect that the figures are bogus, but the cause is actually multiple explosions going off at the same time, which are spawned in batches of 1024. The reason why demo1 is so high is the GL/2-ogre battle at the start.

Spike's totally correct about framerates, but that only applies to rocket trails which get updated, with more particles being spawned, every frame. Most other effects don't suffer from this ((dis)honourable exception - EF_BRIGHTFIELD particles).

[Baker]

Taking all the image manipulation stuff I wrote a 2 months ago and putting it to use, slowly ...

Code: Select all

//**********************************************************************************************************\
//
//   main.c:   Application entry point
//
//**********************************************************************************************************/

#include "project.h"


int main (int argc, char **argv)
{
   char      currentWorkingDirectory[MAX_PATH_IS_1024];
   int         numFolders, numFiles;
   text1024_t*   folderNames = NULL, * fileNames   = NULL;
   int         i;

   commandLineArgumentsCount = argc;
   commandLineArguments = argv;
   printDateAndTime ();
   if ( System_GetCurrentWorkingDirectory (currentWorkingDirectory, True) == 0)       return 1; // Error
   // Get folders list.  Get file list.
   folderNames = FolderMatch_Populate_Alloc (currentWorkingDirectory, "*", &numFolders);
   fileNames   = FileMatch_Populate_Alloc   (currentWorkingDirectory, "*.tga", &numFiles);
   
   ///////////////////////////////////////////////////////////////////////////////////
   // BEGIN
   ///////////////////////////////////////////////////////////////////////////////////

   
   // Turn every TGA into an LMP and write it
   do
   {
      fbool   Perfect = True;
      char   outfile[1024];

      for (i = 0; i < numFiles; i ++)
      {
         imageRGBA_t   myTGAImage;
         fbool   wroteOK      = False;

         // Load the TGA
         if ( !Image_Create_Opening_TGA_File (&myTGAImage, fileNames[i].text) )
            continue; // Skip this image

         StringLCopy (outfile, fileNames[i].text);

         String_Edit_StripExtension (outfile);
         String_Edit_Append_To_Name (outfile, ".lmp");
         {
            int      numBytes   = myTGAImage.pixelcount;
            byte*   myQPALImage = Memory_calloc (numBytes, 1, "LMP space");

            // 1. Convert to QPAL
            Image_Export_QPAL (&myTGAImage, myQPALImage);
            
            // 2. Write the File
            wroteOK = File_Write_Binary (outfile, myQPALImage, numBytes);

            // 3. Free the QPAL
            myQPALImage = Memory_free (myQPALImage);
         }

         if (Perfect && wroteOK == False)
            Perfect = False;

         // Unload the TGA
         Image_Destroy (&myTGAImage);
      }

   
   } while (0);





   ///////////////////////////////////////////////////////////////////////////////////
   // END
   ///////////////////////////////////////////////////////////////////////////////////

   Enter_To_Exit_DebugOnly ();

   return 0; // 0 = success of application
}

[qbism]

speaking of particles and probably really off topic, i really like quakes square particles.

[SCHPAM]

[/SCHPAM] i TEND TO AGREE [edit: well, can the square particles be liked without dissing other shapes?? ]

[r00k]

NICE ScreenShot!

Code: Select all

   //R00k: added square simple particle (oldschool look!)
   particletexture2 = texture_extension_number++;
   GL_Bind (particletexture2);

   // clear to transparent white
   for (i=0 ; i<1024 ; i++)
      ((unsigned *)data)[i] = 0x00FFFFFF;


   // draw a solid block
   for (x=0 ; x<16 ; x++)
   {
      for (y=0 ; y<16 ; y++)
      {
         data[x][y] = 0xFFFFFFFF;
      }
   }


This is the openGL version of above image particles I too use 90% of the time im online.
I have also reverted iDSoftware's inverse perspective scaling of particles with this
in r_drawparticles

Code: Select all

      // hack a scale up to keep particles from disapearing
      if (r_particle_scale.value)//R00k: cvar'd this
      {
         dist = (p->org[0] - r_origin[0])*vpn[0] + (p->org[1] - r_origin[1])*vpn[1] + (p->org[2] - r_origin[2])*vpn[2];
         scale = 1 + dist * r_partscale;
      }
      else
         scale = 1;//R00k: belay that hack!


cuts down on the noclip particles from nearby rooms

[leileilol]

really overdone muzzleflashes