linecode/renderer

My software-only 3D renderer and raytracer (aka "TTSIOD renderer")

Screenshots, official tarballs, etc available at main renderer page

Introduction

This package allows interactive 3D visualization of triangle meshes.

It contains the source code for a straightforward implementation of basic principles in polygon-related graphics theory. You can use these sources to familiarize yourself with:

• 3D transformations
• Gouraud shading (complete Phong lighting per vertex)
• Phong shading (complete Phong lighting per pixel)
• Z-buffering
• Shadow mapping (with soft shadows)
• Raytracing, with shadows and reflections and anti-aliasing
• Portable display and keyboard handling through SDL.

The sources include support for OpenMP and Intel Threading Building Blocks, so they take advantage of multi-core CPUs and execute faster.

The supported 3D formats are:

• .3ds, the well known 3D Studio format (via lib3ds), and...
• .tri, a simple binary dump of vertex and triangle data.
• .ply (partial support, for 3D objects saved from shadevis)

The code is orchestrated with autoconf/automake, so it compiles and runs cleanly on many platforms (tested on Linux/x86, Linux/amd64-em64t, Windows/x86 (MinGW), FreeBSD/amd64, Mac OS/X and OpenBSD/amd64). For Windows/MSVC users, a separate VisualC/ directory includes an MSVC solution, created with the free Visual C++ 2008 Express Edition (configured to use Intel Threading Building Blocks).

The sources are available under the GNU license (read COPYING for details).

This is more or less a reference implementation. I wanted to write my polygon rendering code in as clear a manner as possible, and to also add support for shadow mapping and real phong shading. The code uses floating point almost everywhere, and calculates the complete lighting equation per pixel. When I first wrote this (late 2003), it was far from real-time... Then again, as of 2007, a Core2Duo can render the train object (see 3D-Objects folder) in soft-shadows mode at around 32fps (under Debian64 with Intel Threading Building Blocks). And since OpenMP/TBB will automatically use as many cores as are available, the multi-core future seems... interesting :-)

Compiling

Common configuration:

Edit src/Defines.h to change:

• window size
• Ambient, Diffuse and Specular levels

Edit top of src/Raytracer.cc to change:

• Whether Reflections are on (default:on)
• Whether Refractions are on (default:off)
• Whether Ambient Occlusion is on (default:off)
• also: parameters of the above

For Windows/MSVC users:

Just open the project solution (under VisualC/) and compile for Release mode. It is configured to use Intel TBB for multithreading, since Microsoft decided to omit OpenMP support from the free version of its compiler (Visual C++ 2008 Express Edition). All dependencies (includes and libraries for SDL and TBB) are pre-packaged under the "VisualC" folder, so compilation is as easy as it can get.

When the binary is built, right-click on "Renderer-2.x" in the Solution explorer, and select "Properties". Click on "Configuration Properties/Debugging", and enter...

    ..\..\3D-Objects\chessboard.tri

...inside the "Command Arguments" text box. Click on OK, hit Ctrl-F5, and you should be seeing the chessboard spinning. Use the controls described below (in the KEYS section) to fly around the object.

The compilation options are set for maximum optimization, using SSE2 instructions.

If you have the commercial version of the compiler (which supports OpenMP) you can switch from TBB to OpenMP:

1. "Configuration Properties - C/C++ - Language - OpenMP: (Set To Yes)
2. "Configuration Properties - Preprocessor - Definitions: (Change USE_TBB to USE_OPENMP)

...and simply recompile.

For everybody else (Linux, BSDs, Mac OS/X, etc):

If you don't have SDL installed, download it and install it first (http://www.sdl.org). It is required for portable display and keyboard handling. I used SDL version 1.2.5 but older versions of the library should also work just as well. If you use any of the Linux distros and install SDL from packages, make sure you also install the SDL development packages (that is, the SDL header files and libraries); so e.g. for Debian...

$ sudo apt-get install libsdl1.2-dev

The renderer includes the sources for lib3ds-1.3.0, so you don't need to do anything about lib3ds.

Optionally, if you have a multicore/multi-CPU machine and you want to take advantage of your extra cores, you must either have GCC >= 4.3.2 (which correctly supports OpenMP) or Intel Threading Building Blocks (http://www.threadingbuildingblocks.org) installed.

Executive summary:

    ./configure
    make

If you want to disable the special "snapshot" mode for the raytracer (i.e. if you have such a fast machine that you can "fly" even during raytracing) then add the --disable-brakes option to configure:

    ./configure --disable-brakes
    make

You can enable Intel's Morphological antialiasing using --enable-mlaa. Note that it depends on Intel's SSE instructions.

You can control the compilation by passing your own compiler flags to 'configure'. For any machine with e.g. SSE2 instructions, instead of plain './configure' use:

CXXFLAGS="-O3 -mfpmath=sse -mtune=native -msse -msse2 -mrecip -mstackrealign" ./configure

Compiling with Intel's compiler may also help (on my machine it speeds up rendering by 25% compared to the best result with GCC). Just edit and use the two build scripts in the contrib/ directory (i.e. fix the TBB and ICPC paths and run them).

Running

After a successful make, you can perform a quick benchmark for comparison with other machines...

    cd 3D-Objects
    ../src/renderer -b chessboard.tri

...or simply enjoy "flying" around the object with:

    cd 3D-Objects
    ../src/renderer chessboard.tri

You will be "piloting" around the object with these keys:

KEYS (available only when not benchmarking)

• Hit 'R' to stop/start auto-spin (camera rotates around the object).
• Fly using the cursor keys,A,Z - and rotate the light with W and Q.
• PgUp/PgDown and the (0-9 keys) change the rendering mode: (Points - Ambient - Gouraud - Phong - Phong and shadows - raycasters)
• S and F are 'strafe' left/right
• E and D are 'strafe' up/down (strafe keys don't work in auto-spin mode).
• ESC quits.

Cmd-line options

Usage: renderer [OPTIONS] [FILENAME]
  -h         this help
  -r         print FPS reports to stdout (every 5 seconds)
  -b         benchmark rendering of N frames (default: 100)
  -n N       set number of benchmarking frames
  -w         use two lights
  -m <mode>  rendering mode:
       1 : point mode
       2 : points based on triangles (culling,color)
       3 : triangles, wireframe anti-aliased
       4 : triangles, ambient colors
       5 : triangles, Gouraud shading, ZBuffer
       6 : triangles, per-pixel Phong, ZBuffer
       7 : triangles, per-pixel Phong, ZBuffer, Shadowmaps
       8 : triangles, per-pixel Phong, ZBuffer, Soft shadowmaps
       9 : raytracing, reflections and shadows
       0 : raytracing, with shadows, reflections and anti-aliasing

Have a look at the other meshes as well (inside the "3D-Objects" folder).

Shadowmaps

If you want to look at the shadowbuffer (curious, are we? :-) just uncomment DUMP_SHADOWFILE in src/Light.cc, recompile, run, and as soon as you exit, a file will be created in your current directory, called "shadow". Then...

../src/showShadowMap/showShadowMap

..and you'll get a look at your shadow map.

Tales of multi-core programming

This code was single threaded until late 2007. At that point, I heard about OpenMP, and decided to try it out. I was amazed at how easy it was to make the code "OpenMP-aware": I simply added a couple of pragmas before the for-loops that drew the triangles and the shadow buffers, and ...presto!

The only things I had to change were static variables, which had to be moved to stack space. Threading code can't tolerate global/static data, because race conditions immediately appeared when more than one thread worked on them.

Only two compilers truly supported OpenMP at the time: Intel's compiler (version 8.1) and Microsoft's CL. GCC unfortunately died with 'internal compiler error'. I reported this to the GCC forums, found out that I was not the only one who had noticed, and was told (by the forum guys) to wait.

While waiting for GCC to catch up, I kept researching multicore technologies. Functional languages seem particularly adept to SMP, and I've put them next in line in my R&D agenda (Ocaml and F# in particular). Before leaving C++ behind, though, I heard about Intel Threading Building Blocks (TBB) and decided to put them to the test. TBB is a portable set of C++ templates that makes writing threading code a lot easier than legacy APIs (CreateThread, _beginthread, pthread_create, etc). TBB is also open-source, so it was easy to work with and figure out its internals. Truth be told, TBB required a bit more changes in my code (OpenMP required almost no changes). Still, it is a vast improvement compared to conventional threading APIs.

I must confess I have not invested a lot of effort in using these technologies; I only enhanced my two main rendering loops to make them SMP aware. Still, this was enough to boost the speed (on my Core2Duo) by 80%! Judging by the effort invested, this is one of the best bargains I've ever met in my programming career.

As of now (October 2008), GCC 4.3.2 is now up to speed and compiles OpenMP code just fine. TBB is of course running perfectly (since it is simply a C++ template library), so choose freely between any of the two, and easily achieve portable multithreading.

When I say portable, I mean it: "./configure && make" is enough to create:

• OpenMP binaries for...

  • Windows (via TDM/MinGW GCC 4.3.2)
  • Debian Linux Etch (both 32 and 64bit, via manually-built GCC 4.3.2)
  • Debian Linux Etch 32bit (via Intel's 10.1.017 compiler)

• TBB binaries ("./configure --disable-openmp --enable-tbb") for...

  • Debian Linux Etch (both 32 and 64bit, via manually-built GCC 4.3.2)
  • Debian Linux Etch 32bit (via Intel's 10.1.017 compiler)
  • FreeBSD/64

• Single-threaded binaries for...

  • Poor OpenBSD/64: it doesn't have real, SMP threads. Not yet, at least. It only has user-space ones (as Linux did at some point). But it does compile the code, albeit in single-threaded mode.
  • Mac OS/X: Tested by a friend with a single core OS/X machine: Compiles and runs just fine (albeit in single-threaded mode). I don't know if OS/X GCC has adequate support for OpenMP; if it doesn't, GCC 4.3.2 will surely be available for it soon.

Talk about portable code!

If you're still in the dark ages and use legacy APIs (like CreateThread, _beginthread, pthread_create, etc) you are really missing out.

Enjoy!

Thanassis Tsiodras, Dr.-Ing.

ttsiodras_at-no-spam_thanks-gmail_dot-com