In the swiftly evolving digital era, we are constantly seeking ways to streamline and optimize our processes. With the aid of cutting-edge technologies, we’ve created a rendering process optimizer that brings significant benefits to our rendering farm.
Our new optimization tool considerably shortens the time of rendering a single frame, which results in faster task execution and more efficient use of our resources. This means we can now deliver our clients’ products faster, giving us a competitive edge.
However, our optimization doesn’t stop at speed. Our new optimization tool is also economically advantageous. By reducing the amount of computational power used, our costs are now lower. This translates into cheaper services for our clients, as they pay for the computational power we use. So not only do we deliver their products faster, but we also do it for less.
Our achievements are not limited to speed and economy. There is also a significant environmental aspect. Less computational power used means less CO2 emissions, making our solution more environmentally friendly. In the era of global warming, this is a key aspect that more and more companies are paying attention to.
EU funding has enabled us to purchase top-quality servers, which further contribute to the efficiency of our solution. These servers not only increase our performance, but also minimize our CO2 emissions, making us more sustainable.
In conclusion, our new optimization tool brings numerous benefits. It’s a faster, cheaper, and greener solution that allows us to better adapt to the dynamic rendering market. It exemplifies how technological innovations, backed by appropriate funding, can lead to significant benefits on multiple levels.
Our rendering optimization algorithm works based on the hyperrectangle method, which is one of the direct search methods for functions of multiple variables. We’ve applied it to optimize VRay parameters – maxSubdivison and minShadingRate, which have a key impact on the quality of rendering and the time devoted to it
Initially, the algorithm chooses arbitrary starting points for maxSubdivison and minShadingRate. Based on these points, it creates a hyperrectangle in the parameter space, on which it performs a series of calculations. Each vertex of the hyperrectangle represents a different combination of parameters, and the function value at these points is the rendering time.
The algorithm compares the rendering times for all points, then moves towards the point that returned the shortest time, provided that the render quality is not worse. This process is iterative – in each iteration a new hyperrectangle is created around the point with the smallest rendering time and the best solution is sought again.
If the algorithm finds a point where rendering times start to increase (i.e., the rendering quality does not improve), then the size of the hyperrectangle is reduced – this allows for a more precise search for optimal settings in the close vicinity of the current best parameter combination.
This process continues until the difference between successive best rendering times becomes so small that we can assume we have found the optimal settings. The uniqueness of this approach is that it allows you to find the most efficient render settings without having to search the entire parameter space, which would be very time-consuming and inefficient.”