This story was originally published in the Blender Art Magazine issue #41. You can read it in the magazine, and we are reproducing it here for your convenience.
In this article, I am going to write about the story of RenderStreet, the next generation render farm dedicated to rendering Blender projects. I will touch on the topics of how we came up with the idea, what was like building it and how it turned out in the current stage.
The need
We started thinking about RenderStreet in January 2012. The idea came from a very simple need. The need to render complicated projects in a faster way than a couple of home computer can.
At that time, Oana, one of the founders of RenderStreet and the resident Blenderhead, had a difficult project. She is an architect by formation, but she specialized early in her career on architectural visualization. The project had several still image deliverables (about 10) and each of them took hours to render on our home computers. Since it was drawing close to her deadline, we decided to go shopping for an online rendering service that can alleviate the pressure.
So we went searching, knowing that we are covered financially by the customer who agreed to pay for the renders. We spent a couple days looking at the players in the field and trying to render a few files. We were not happy with what we found, and in the end were stuck with babysitting the renders on our own machines.
The planning
After thinking about our experience, I became aware there is a potential market here. After all, Oana shouldn’t be the only Blender user who uses it for professional services and has a need to render fast. So I co-opted Marius, my longtime partner in other ventures, and started thinking on what it would take to create a better service.
We spent some time planning on what a good service should look like, feel like and what the overall experience for the users should be. We distilled it to 3 major ideas and these ideas have, in time, become our mantra: ease of use, affordable price and great customer service.
Ease of use means, in our opinion, a modern site, with an interface anchored in the modern trends in web design. One that does not ask users to learn anything in order to use it. One that removes the barriers from using such a service for users who are not web wizards. At the same time, it should not be simplistic. Easy to use, yet powerful enough to allow sophisticated users to tweak their projects in more detail.
Affordable, in our opinion does not mean cheap. If we wanted cheap, we would have gone for the worst computers that cost the least money. There’s a fine line between cheap and affordable. We want people to afford to use our service, even if it’s a private person rendering a hobby project. Furthermore, we wanted it to be attractive as well to larger companies who use Blender in a professional way, since they have the largest rendering budgets. In order to achieve this, we would have to get fast servers at decent prices, so we can create an affordable price for the end customer.
Great customer service is the trickiest part to achieve when running a business. Both I and Marius have had experience setting up customer service operations and working with in house and offshore customer service departments. In my opinion, there are two key factors in having great customer service:
- No need for such a service in the first place. If the site is easy to use and the results are delivered on time and on budget, nobody needs to talk to a customer service agent
- When there is a need to talk to such an agent, the person answering needs to know the business inside and out, as well as have experience in using Blender. This way they can understand the customers’ need and provide immediate resolution for his problems.
- After a few weeks of talking about it, planning and investigating the feasibility of the idea, we went on to build the product.
The building phase
Both I and Marius are seasoned programmers, with lots of business experience. We were new to Blender though. So we spent a lot of time learning it, understanding how it works, how the various rendering engines differentiate from each other and how we can apply our programming skills to automate it.
Once we started to get a grasp of how that works, we plunged head on into developing the site. That was the most resource intensive process and took quite a long time to complete to our satisfaction. We had a first draft of the site in mid-August and did a lot of internal testing to make sure we have a stable enough product for a beta launch.
At the same time, we reached out and investigated what it will take to get servers to render the Blender projects on. We had lots of discussions with various providers, and still have at the time of this writing. So far, we have chosen 3 providers in US and Europe to source our servers from. They provide great servers specifications at prices that allow us to fulfill our promise for affordable pricing.
Once we had servers to run on, we spent a lot of time getting to know how to build Blender so it takes full advantage of the hardware it runs on. We choose from the start not to go with stock or third party builds, as we felt those are general enough not to perform at Blender’s peak capabilities on our hardware. With the help of the community, we understood the process quite well and are able to have our own builds that are highly optimized to the platforms they run on. This worked so well, we managed to get the 2.66 version out only 5 hours after the official release went out.
Where we are now
We went online in closed beta in mid-November 2012. We were fortunate enough to be selected as a finalist in the How to Web startup competition, so we got a bit of visibility for the project at that time. When we asked the community to help us test our new render farm, the response was overwhelming. We ended up selecting 100 users for the closed beta, out of which 50 rendered their projects with us and provided feedback on what went wrong and what can be improved. All rendering in the closed beta was free of charge.
The closed beta ended at the beginning of February, when we launched the open beta phase. Even if we call it an open beta, the product is stable enough to be used for every day rendering and so we encourage everybody to use it.
This brings us to the present day. I am happy with the state we are in now, and I think we are on our way to accomplishing the goals we set for ourselves when we started the project. We support 5 versions of blender, from 2.63 to 2.67, render Cycles on GPU, are able to render still images on multiple servers by splitting them into parts and stitching them at the end, and have received many comments about the ease of use of the site and the good customers support provided so far.
That being said, it is still a young project and we have a lot of work ahead of us. To mention just a few items on our list, we still need to add support for VRay, expand our customer service so it can be provided 24/7 and work on integrating more functionality to accommodate the workflow of larger studios.
The conclusion
I think we managed to build a quality product and I want to thank the wonderful Blender community for the support given in shaping RenderStreet and it bringing to market. It has been an interesting journey and still a lot lies ahead to be discovered. I hope we’ll be able to fulfill our promise to the community and keep RenderStreet as the next generation render farm for Blender.
