Fly by the darkness.
Vertex-normals application for one-point lighting and reflection system
This video is about comparing vertex-normals technique and its applicability to games that uses one-point reflection system that does not allow normal maps/bump map. Here, my purpose is to investigate the problems with each technique, to provide information about the vertex-normals techniques being used, and what I would recommend to use when it comes to games that uses one-point reflection system that does not allow normal maps.
Note: This video and description does not apply to modern graphics game application where reflections from environment are applied to target meshes rather than texture at first-person angle.
What is transfer attributes: Vertex-normal modification using the 'transfer attributes' techniques involves the application of transferring vertex data from source mesh using sampled vertex information into the target mesh based upon spatial information.
What I do to achieve the results? I have used a modified sphere as the source mesh and then apply it to the target mesh which is shown at the video.
The pros and cons- It resolves certain problems that the calculation methodology have which is the fact that there are stretched reflected area while shading bumps can sometimes be a problem with the calculation methodology. It also resolves problems with projection technique as the projection technique results in inaccurate shading due to the vector technique where the direction of the vertex-normals is dependant from axis and vertex. However, this technique does have issues with darkened areas at long-range view and sometimes produced unacceptable reflection in flatter area that needs more polycounts.
What is calculated vertex normals: Calculated vertex-normals refers to vertexes ray direction being calculated from angular information from the surrounding areas.
What I do to achieve the results? Most programs nowaday alters vertex-normals as you edit 3D models by moving vertexes, so there isn't anything I need to do other than make the model as it is.
The pros and cons-This offers the best shading, but there are problems such as stretched reflections. As the results of how it gets calculated, in itself does not necessarily provide a smooth form of shading.
Mix of calculation and projection:
What is the mix? The mix is generated from mathematical calculations that calculates vector directions of vertex-normals and calculates the results based upon user-set ratios of mixed results from target mesh's vertex-normal to projection filters. Anything that isn't 100 percent projection, but the direction leads to there can be used as a example of mixed vertex-normals information.
Pros and con: Due to the large differences of vertex-normals between prior vertex-normals and projection when the shape isn't regular or uniform and without edges, the results shows problems within shading and reflection.
What is projection: This is a technique that involves generating vertex-normals direction from location of vertexes in respect to axis.
Pros and con: It can be useful in mesh which are low-poly (less than 600 polies)as differences between vertex-normals are large. The con is that the shading is extremely inaccurate.
It appears that the transfer attribute results is more preferable for old game applications due to the results of the reflection and the shading itself is almost as accurate as the calculate metholody. The overall darkness issues associated with this technique can be easily resolved by scaling vertex-normals or this can also be blended in as a reflective layer to calculated normals layer where the mesh has no reflections.
In the end, no matter what you do, it's all up to preferences. There's no correct answers.
The official website of NFS3 Expansion Pack
Last edited by Reptillian; 03-15-2013 at 09:54 PM..