MDK2 has four major sections in the rendering code. The first is object transformation. This is where the object is translated and/or rotated from it's original location (this is just a local to world coordinate system change.) The second is camera transformation where the object is transformed into the camera's coordinate system (changing the coordinate system from a world to eye.) The third is the projection transformation where the object is projected to the screen (changing the coordinate system from eye to screen.) The fourth and final stage is the lighting where all the lights which affect the object are collected and applied to the object.
When T&L is disabled, stages 1,2 and 4 are done in software in the Omen engine. The only place where OpenGL is use (which may or may not be hardware) is in stage 3, the projection transformation. So with T&L disabled, most transforms and all lighting is done in the Omen engine.
When T&L is enabled, all 4 stages are done through OpenGL (hopefully with hardware). There are also several OpenGL specific optimizations such as display lists which help to improve performance with T&L on.
Under T&L, the lighting should be more precise, because we took certain shortcuts in the Omen engine to make it as fast as possible in software mode, but in most cases, you can barely tell the difference. As for number of light sources, the Omen engine can handle more than 8 light sources (which is usually the limit in current hardware designs) so in certain cases, the software version may be using more lights than the T&L version. As for tessellation, the models never change in either version. There is no further tessellation under T&L. So basically, the biggest thing you get with hardware T&L is a significant speed improvement."
When T&L is disabled, stages 1,2 and 4 are done in software in the Omen engine. The only place where OpenGL is use (which may or may not be hardware) is in stage 3, the projection transformation. So with T&L disabled, most transforms and all lighting is done in the Omen engine.
When T&L is enabled, all 4 stages are done through OpenGL (hopefully with hardware). There are also several OpenGL specific optimizations such as display lists which help to improve performance with T&L on.
Under T&L, the lighting should be more precise, because we took certain shortcuts in the Omen engine to make it as fast as possible in software mode, but in most cases, you can barely tell the difference. As for number of light sources, the Omen engine can handle more than 8 light sources (which is usually the limit in current hardware designs) so in certain cases, the software version may be using more lights than the T&L version. As for tessellation, the models never change in either version. There is no further tessellation under T&L. So basically, the biggest thing you get with hardware T&L is a significant speed improvement."
