About the ever-rising demand of 3D artists as a hobby or a profession, this work devised a cost-efficient render farm using the on-demand AWS services. Using cloud services, the rendering is performed more efficiently than the conventional rendering over the personal computer. This process is further optimized by implementing the concepts of cluster computing where the work is divided over many computers and are completed parallelly to reduce the time taken to complete the whole task. The paper also reports the disadvantages of the conventional process of rendering and throws light on various issues which cannot be unseen by the user when using the conventional process. And it also proves that this render farm is more efficient and cheaper than the existing render farms.
The research performed to find out the optimal type of instance in AWS gave the conclusion that C5a.4xlarge instance from the C-Class gives better output for a lesser price and that fulfils the main motive of this render farm. Also, different number of worker nodes are tabulated along with varying number of frames. It is clearly understood that the percentage of time saving significantly increases as compared to traditional rendering process. The vast amount of rendering can be divided into various smaller parts and can be assigned to the worker nodes by the master nodes combined with the instructions required for the rendering, which are basically user-defined. After the render task is completed, it is retrieved by the master node and also gets stored in backup nodes in case of data mishaps. All these nodes are connected over ansible using SSH. This has proven to be its kind prototypes enabling connection between best of both world of cloud cluster computing along with 3D software. In the proposed idea, the way of copying the files around from the master to the worker nodes and back from worker nodes to the master node is not efficient. So, the current work is extended to develop a newer version of the render farm where one of the AWS services, the S3 buckets are being utilized to set it up as a common disk for all the instances in the render farm. This would lead to lesser network and storage costs. In addition, other AWS services like AWS Cloud Formation, Cloud Watch, Auto Scaling Groups (ASG), Load balancers, etc. can be incorporated which would allow us to automate the deployment of the instances and the termination of the instances. Also, this would let us scale up or scale down the number of instances in the render farm based on the load and the number of frames to be rendered.