Smart way to mine with profits. Try Sonm Os!

SONM Partners with POSTKINO FX

SONM Partners with POSTKINO FX to Run Fog Rendering in Post-Production

The era of fog computing usage for real-world tasks is coming. The time will come when fog takes over cloud, presenting a great range of benefits for companies worldwide. The Sonm fog computing marketplace has proven its efficiency in a short period of time by reaching more than 350,000 working hours for the deals running on its platform.

Studio POSTKINO FX deals with the creation of computer graphics for cinematography and TV and is one of the first companies to use Sonm to render their tasks. The MEGAFON provider, which is the second-largest mobile phone operator and the third-largest telecom operator in Russia, operates as the end customer for the graphics created, and as such, the task Sonm directed to was highly sensitive.

“We pay significant attention to the developing trends and to the new technologies that emerge on the market, and usually we do not use independent render farms to run our tasks. We have a very good technical basis in this field and our own render farm. But this time we had a challenge to run two projects at one time, one feature film and one advertising video, and the equipment we had had been busy 24/7 over the previous few weeks. This was a great time to try out what our friends at Sonm offer, to do both projects at the same time and to try out the Sonm platform on this real-world task. The result was inspiring.”

Ivan Lebedev,

The rendering process: challenges and solutions

To render the tasks, the following node configuration was chosen: CPU Celeron + 8Gb RAM + 3 GPU GeForce 1070 +10Mbit IN. There were seven such nodes located in different cities,. The task was rendered using Blender and included nine episodes with a 618-Мbite volume and 522 pictures. We had two days to render the task, including possible changes and test rendering, so the process ran as follows.

First, we rendered a few frames from one episode and then adapted the rendering parameters and hardware configuration. Then we rendered entire episodes with a length of 70 pictures, after which we evaluated the rendering speed and quality. After that, we started rendering the other episodes. In the end, the client changed three initial episodes, two of which had already been rendered, so we re-rendered those.

The settings and the environment for rendering on Sonm and on the client’s render farm were not identical, and therefore we couldn’t use them together to render the same episode. On the whole, the process included a significant amount of manual work by Sonm system architect Eugene Manaev, and we understood that the process stood for automatization.

“When the deadlines are strict and fast approaching and the situation changes all the time, sometimes tasks change right in the middle of the process, and we have to react immediately. Sonm reacted perfectly and took on the new tasks. Great job, nothing to add.”

Ivan Lebedev,

The whole process took 19 hours, and including the re-rendering of two episodes and numerous broken frames, we rendered 632 pictures. The average rendering speed was 33 frames per hour, and the process ran on 21 GTX 1070 video cards.

“When POSTKINO FX turned to us with this task, the only thing we thought was, “Wow, rendering, here we come.” Render farms are usually in the cloud with the equipment located all in a single place. It is a pool of resources that belongs to one single person or to a single group of people. It has limited capacity, and the upscaling of such pools is a subject for discussion. The Sonm platform aggregates resources located all around the world, and the client therefore gets an opportunity to use nodes from a global decentralized pool on demand.

We definitely need to adopt the interface and make it user-friendly. We are in for a lot of work to create a Trusted Platform. But the effort is worth it: The result will be a Fog Rendering Platform that is likely to disrupt the industry.”

Eugene Manaev
SONM System Architect

Alexey Krotov

“The results of the cooperation are very promising. The guys at Sonm have done a lot of work. Of course, there are still things to improve, though. First, we need to automate the process and get a detailed prognosis of the rendering equipment to take on the task. The number of video cards ranged from 20 to 50, and we couldn’t quite determine how long the process would take. Also, CPU usage is needed. And — last but not least — the usability should be improved so that users all around the world can use Sonm for rendering.”

Ivan Lebedev,

Well, what was the result? Here you are!