The realization of complex scenarios and games with high realism demands the use of sophisticated techniques that avoid saturating the bandwidth available on graphics cards.
Reducing the number of polygons to represent one of the priority objectives for improving performance.
Una técnica que permite representar un objeto con un número de polígonos que depende de la distancia del objeto al punto de observación se denomina LOD (Level Of Detail)
En este artículo profundizaremos en estos conceptos que permiten optimizar la representación de complejas escenas con gran realismo.
The color and lighting can appreciate different details and nuances of the surfaces. Called “Buny the Standford” (rabbit) is a surface model of repository created at Stanford University, used as a reference in the comparative test to validate the various techniques.
The level of detail (LOD) of an object (the geometric shape of su) impact on the cognitive aspects of its image recognition, and in the computational cost related. Improving representation should be based on the right balance between these two aspects. A good picture but cheaper.
- Since cognitively is about giving “credibility” representation, regardless of the technique used. The distance to the object, the aesthetics of the image, environment, movement and other factors directly influence this process.
- Since terms of computational cost is intended to achieve “speed”, so the resolution of the model, the number of objects in the scene, calculated the type of lighting etc., may be limited by the availability or ability to process both the CPU and the GPU.
More model 20.000 facets will be reduced to just 400. Although the new model calculated with a notoriously bad algorithm distorts the initial geometry (right), a determinada distancia la diferencia es apenas apreciable, assuming considerable savings in resources to its graphic treatment.
The use of different models with more or less detail, of the same object, to be used depending on the convenience in each scene could be a technique that optimizes certain resources, as calculations and projection display, however, it can negatively affect other aspects, as the use of memory and transfers data for such models.
Techniques allow intermediate levels of detail at run time application (dynamically calculated) Looking for resource balancing. How to simplify an object, reducing the number of facets, is currently the subject of numerous studies and papers are published in specialized forums on synthetic image. The techniques try to look as close as possible to the object approaches, error functions determined for evaluation. The cognitive aspect is more subjective and therefore more complex structure.