Success story

Time and Energy Optimization: some facts!

Japan-based Soza-all explains how real-time rendering became a turning point in his success.

After working for over 12 years as a free-lancer in the VFX industry for movies, TV shows, and commercials, Shinya Hakamata founded an agency in 2020 for 4k CGI stock footage, called Soza-all

Over time the amount of projects that he had to deliver became unsustainable. Shinya Hakamata had always used the native renderer, and the long rendering times that came with it. The tight deadlines forced Shinya Hakamata to make a decision about the number of workstations used on a project. Increasing the number of PCs would have helped to process the required renderings simultaneously – but what about costs?

Vimeo

By loading the video, you agree to Vimeo’s privacy policy.
Learn more

Load video

狼煙 feat DAI JAP&Chikara&流句(極彩色) (Pro.Modeous)Noroshi LyricVideo

An eye on the bills

Typically, very powerful workstations are used when deploying rendering solutions. 

For CPU-based solutions, the focus is on using powerful CPUs with as many processing cores as possible. Systems with multiple CPU sockets (e.g. Intel Xeon System, AMD Epyc) or simple CPUs with many cores (e.g. AMD Ryzen Threadripper with 64-cores) are cost-intensive to purchase. 

GPU-based solutions typically use multiple GPUs at the same time. For this, the power supplies and motherboards have to be designed accordingly. We typically see 2-3 GPUs in workstations designed for GPU rendering, but we also know systems with 7 GPUs. These typically use the most powerful models available.

For Shinya Hakamata electricity costs were also a problem, as electricity in Japan is very expensive.

The average computing time for an image in Full-HD resolution is somewhere between 10 and 30 minutes. With a CPU renderer, it can go well beyond that and the computing time always depends upon the scene complexity. 

We calculated some energy consumption estimates and, for simplicity, we have assumed the power consumption for the different workstations to be 1 kWh uniformly. 

Please note, workstations with high-end CPUs and/or multiple GPUs will consume more power than, for example, moderate hardware such as would be required for U-Render.

A clear breakdown

For an animation with a duration of one minute, you usually need 1,500 images. A minimum of 25 frames per second is required. At a higher frame rate correspondingly more frames would be required, 30 or 60 frames per second, perhaps.

Frames for one minute = 25 frames/sec * 60 = 1,500 frames

Assumptions for calculation duration based on empirical values for GPU and CPU renderer and generous 10 sec/frame for U-render.

  • CPU renderer: 35 min/frame = 1,500 * 35 = 52,500 min = 875 h = 36.46 days
  • GPU renderer: 10 min/frame = 1,500 * 10 = 15,000 min = 250 h = 10.42 days
  • U-Render: 10 sec/frame = 1,500 * 0.17 = 255 min = 4¼ h = 0.18 days

If we assume a power consumption of 1,000 W (1 kWh) and a kWh price of 0.15 € then the following costs result from processing one minute of animation:

  • CPU Renderer: 875 * 0.15 = 131.25 €
  • GPU renderer: 37.50 €
  • U-Render: 0.64 €

If a customer charges 30 minutes of animation per year, the energy costs are as follows:

  • CPU Renderer: 3,937.50 € / year
  • GPU renderer: 1,125.00 € / year
  • U-Render: 19.20 € / year
Vimeo

By loading the video, you agree to Vimeo’s privacy policy.
Learn more

Load video

Food Plate Animation made with U-Render

“I was troubled by the fact that I would not be able to check or deliver my work in time. If I want to deliver in time, I have to increase the number of PCs. More PCs will increase the electricity bill. I was stuck.” – Soza-all.

The breakthrough

However, increasing the number of workstations was not the only option. Shinya discovered U-Render, the real-time renderer, fully integrated with his favourite 3D application, Cinema 4D.

U-Render has comparatively modest hardware requirements and you can use it on a well-equipped notebook with a dedicated graphics card. Significant hardware acquisition costs could be saved.

As a result of being a real-time solution, U-Render also consumes considerably less energy during operation. Due to the additionally shorter computing times, the hardware is then also used for noticeably fewer operating hours. This means that the savings accumulate over the lifetime of the hardware. 

Getting started with U-Render was easy, as there was no need to learn another tool besides the one he was used to working with. Nor the need to change his production process. Excited by the high-quality results, Soza-all’s founder tried out all the features and his favourites are: IBL (Image Based Lighting), Displacement Mapping and Chromatic Aberration.

“Rendering with U-Render is extremely fast. With the standard renderer, scenes that would have taken more than 10 minutes per frame were reduced to 10 to 20 seconds. I think this is especially useful for high resolution work such as 4K and 8K.”

TRY U-RENDER FOR FREE

Sign up for a free trial license to fully unlock U-Render and test it in your own production environment.

BUY U-RENDER NOW

Deliver stunning results in final frame quality within the blink of an eye!