The 3D volumetric rendering process faster and more efficient

Although Evercoast approached us first in October 2018, one could say that our relationship began much earlier than that. That’s because Ben Nunez, the company’s CEO, has known Merixstudio for over a decade. Having worked with us a couple of times before, he believed we’d be capable of solving the scalability and efficiency-related problems his company was dealing with. To address these issues, we put together a project team with two developers as a core as well as a Project Manager and DevOps engineers lending their skills when needed. 

Merixstudio’s squad acts as a team extension accountable to the client’s Project Manager. This, however, doesn’t curb our technological creativity. To quote Ben Nunez’s review on Clutch, “[Merixstudio’s team] helps us think through things and contribute to what the best approach is. That balance of being able to execute but then also contribute to the technical direction of what they’re working on has been helpful”.

Working with Evercoast, we follow an iterative release process, which means we can distinguish several components of our collaboration. The first would be boosting the pipeline that manages all of the client’s computer vision applications. Initially, the issue was the unbearable length of the sequential processing run on a single computer. Our team addressed it by using AWS – as a result, the 3D rendering pipeline can be completed much faster, using 1,000s of CPU cores. This part of the project went live at the end of 2018.

Then, we proceeded to work on desktop and web applications. The former, created by Evercoast’s team, serve to record videos and transfer them to the cloud. In this case, our scope of work is limited to providing frontend support – one example may be introducing Snapcraft to ensure the quick and smooth installation of the app. Our main focus is the web application, which we’ve built from scratch and are currently engaged in its maintenance. Based on Python, Django, and Angular, this part of the Evercoast’s product is used for managing the assets and the resulting renders. 

Over time, we’ve worked on other improvements as well. At its core, Evercoast’s product relies on the cloud, which means a lot of captured data has to be streamed for processing. Such an approach has its limitations in cases such as trade fairs, where our client can present the system to a wide audience but the Internet connection on-site is far from perfect. We've overcome this issue by preparing a version that can run locally. Now, we’re working on expanding the platform’s reach, power, and flexibility.

Since Evercoast’s product is one of a kind, working on this project is challenging. Evercoast’s 3D rendering process is exceptionally complex and involves processing massive amounts of data. As you may expect, some of our goals are to make it fast, efficient, and thus cheaper to execute. The real challenge is to attain them without having direct access to the cameras – thus not having the opportunity to test the entire rendering process on our own. Luckily, open communication with the client and understanding of their objectives help us deal with this issue.