News

lucille Beta v0.9.6 release

Fixstars Corporation is proud to announce that lucille Beta version 0.9.6 is released.

You can obtain the source code and binaries from download site.

This release contains the following features and bugfixes.

Updates

  • Enable to terminate distributed rendering when you press CTRL + C.
  • Support new shaders:
  1. blend shader
  2. metal flake shader
  3. reflection coat shader
  • Support opacity maps.
  • Support SGI RGB image format for texture maps.
  • Enable to read 4-channel texture images
    • Alpha channel is ignored.
    • You can use translucency with opacity map.

Bug fixes

  • Fixed displacement shaders
  • Fixed a typo in command line argument.
  • Fixed pixel filter bug (“gaussian” pixel filter).

The following images are rendered with the latest lucille.

Maserati in Palace 1

Maserati in Palace 2

[Tech Blog] Designing a Dynamic Simulation for lucille, M³ Platform, and Violin Memory SSD Storage

Scott Frankel, Director of M³ Business Development

Particle dynamics simulations are an essential component of many high-quality Computer Graphics productions. To create complex and realistically rendered natural phenomena, such as smoke, fire, water and atmospherics, tiny 3D objects are created and animated algorithmically by the millions and/or billions. With such large numbers of particles, the resulting scene data size can quickly climb to many gigabytes. With 24 images per second for film (30 images per second for video), the total storage required for sequences of simulation data can easily grow to terabyte scale. Moreover storage I/O can become a dominant factor in aggregate render time efficiency.

In addition, to create an image from particle data, a ray tracing renderer must take into account all the interactions of light between each of the particle objects and all the surrounding objects in the scene. This is non-trivial and huge computational resources are required, especially if the objects are translucent or have reflective surfaces. In addition, to achieve photo-realistic quality, natural soft shadows must be calculated for each object as well. This further increases the computational demands placed on the renderer.

For this scene, we started by creating a polygonal model of downtown New York City. A camera move was key framed to start close-in on the building rooftops, then pull back as a burst of particles stream through the buildings like a torrential wave. After this initial burst, more particles enter from above, enveloping the viewer inside a graceful swirl of shiny metallic balls. Eventually, the camera’s continuous dolly backwards reveals that we’ve been inside a snow dome. The impression is reinforced by setting the objects, now noticeably a miniature, on a wooden tabletop with windows reflecting in the glass dome and a postcard from New York City resting nearby.

To create the illusion of scale, the camera starts low to the rooftops. The building geometry and torrent simulation are parented under a non-renderable Maya locator that animates along the Z-axis during the camera pullback to keep the leading edge of the torrent in view as long as possible and prevent stray particles from entering frame. This animation is invisible from the camera’s perspective.

Rendered metal ball image

The swirling, snow-like, metal ball simulation is created with vortex fields and confined by a collision object made from an invisible inner sphere that’s slightly smaller than the glass outer dome. This prevents geometry that’s instanced to the particles from penetrating beyond the outer glass of the snow dome. An omni-directional emitter with an animated rate parameter is defined to create particles within a region of space specified by minimum and maximum distances. Combined with random speeds, creating particles from a volume allows them to increase in number without drawing attention to their creation, filling the space almost invisibly. The instanced sphere geometry, their radii and surface attributes, and animated opacities are all set directly in ASCII RIB files.

We use Fixstar’s global illumination renderer, lucille, running on the highly-optimized M³ software platform to create photo-realistic images of the scene we created. This yields the most efficient use of multi-core hardware resources, while performing all imaging calculations necessary to create a sequence of HD (1920×1080) output images. With over a billion particles, the scene data is over 50GB per frame. Thus it’s too large to load into RAM, which results in disk I/O becoming a dominant factor in overall render times.

As highly-optimized as Fixstars’ M³ platform is, for the snow dome scene the disk read times represent a bottleneck that’s insurmountable with traditional hardware. Silicon-based storage solutions have shorter latency and much broader bandwidth than conventional HDD-based storage. Violin Memory’s V6616 is a storage array solution configured from Single Level Cell (SLC) Toshiba NAND flash memory and optimized for high IOPS and low latency, while still providing robust RAID protection, ultra-low response times, and high transaction rates. Using the Violin Memory V6616, our aggregate render times are 58 times faster than with conventional HDD storage. Furthermore, having eliminating the long read times, lucille on the M³ platform renders these enormously heavy frames effortlessly, writing frames in approximately 180 seconds on a 64 thread system.

Fixstars’ global illumination renderer, lucille, built atop the M³ multi-core optimization platform, pairs beautifully with silicon-based storage systems like the Violin Memory V6616. This is especially evident with 3D scenes that require significant I/O, such as heavy texture sets, volumetrics, and very large particle dynamics data.

Thank you for visiting our booth at SIGGRAPH 2012!

The Fixstars team would like to thank you for visiting our booth at SIGGRAPH 2012 in LA where we launched our latest GI rendering software, lucille, running on the NAND flash array storage system, the Violin Memory 6616.

We demonstrated a great speed increase in rendering a massive particle scene by utilizing Fixstars’ “lucille” global illumination renderer and the newest Violin 6616 Flash Memory Array. At maximum, it has achieved a 50X speed increase in rendering a massive particle computer graphics (CG) scene.The stunning results are accomplished through the combination of lucille’s superior parallel processing performance and the industry’s fastest I/O throughput of the Violin 6616.

To see more detail of the solution, please check the following brochure and white paper, which were handed out at the booth.

In addition, please check out the particle demo movie rendered by lucille.

Watch larger video

About lucille

lucille is a Global Illumination ray-tracing renderer that combines exceptional quality with incredible speed. From miniature models to complex scenes, lucille’s RenderMan compatibility and efficient design allow it to render photo-realistic images effortlessly.

lucille is now available for download for software developers and CG artists:

About Violin Memory 6616

The Violin Memory 6616 is a storage solution configured from 100% Toshiba NAND flash memory.It’s based on Single Level Cell (SLC) flash memory and optimized for high IOPS and low latency, while still providing robust RAID protection, ultra-low response times, high transaction rates, and real-time queries of large datasets.

Fixstars Achieves 50X Acceleration in Rendering Massive Particle CG Scene Using Violin Memory’s Violin 6616 Flash Memory Array

Solution Takes Advantage of Multi-Core Processing and Multi-Node Development Environments to Deliver Speed and Cost Improvements in Rendering Complex Images

SIGGRAPH 2012, Los Angeles, August 7, 2012 — Fixstars Solutions, the leader in multi-core software solutions, today announced it has achieved a 50X speed increase in rendering a massive particle computer graphics (CG) scene by utilizing Fixstars’ “lucille” global illumination renderer and the newest Violin 6616 Flash Memory Array from Violin Memory. The stunning results were accomplished through the combination of lucille’s superior parallel processing performance and the industry’s fastest I/O throughput provided by the Violin 6616.

Particle dynamics are an essential component of high quality CG production, enabling the execution of complex and realistically rendered natural phenomena, such as flames, explosions, smoke, splashes, mist, and hair. However, rendering particles typically requires large data sets and random storage access. In that case, the I/O speed of data storage represents a bottleneck for traditional storage solutions, which are typically composed of numerous hard disk drives (HDDs).

Watch larger video

Violin 6000 series

Violin Memory’s Violin 6616 Flash Memory Array, which is a storage solution configured entirely from Toshiba NAND flash memory, enables faster data read-and-write times compared to traditional HDD storage systems, since it has no moving parts. Fixstars’ tests confirmed that the Violin 6616 yielded 50 times faster processing speeds than traditional HDD based solutions for rendering large data sets. CG companies can replace traditional racks of multiple HDDs  with Violin Flash Memory Arrays which can yield significant savings in rendering times computation costs and storage footprint.

Fixstars’ lucille has an enormous speed and compatibility advantages over other ray tracing renderers, since it is highly optimized to take advantage of the inherent advanced performance capabilitiesof multi-core processors and multi-node environments. The particle generation simulations were achieved by making maximum use of 4 nodes with 32 cores of 64 threaded to achieve this level of high speed performance. Similarly, the Violin 6616 provides superior performance in multi-core environments, achieving low overhead and high data rates per CPU cycle. During the simulations, the Violin 6616 sustained lucille’s high computation speed, even when comparing single vs.16 compute threads, confirming Fixstars’ view that the combination of lucille and Violin 6616 represents the perfect solution for rendering large data sets, such as dynamic particle simulations.

Don Basile, CEO of Violin Memory commented, “We are pleased to enable Fixstars set a new performance standard with their “lucille” global illumination renderer application by making the move to Violin Flash Memory Arrays that allow the full power of the application to be realized rather than being held back by legacy disk array technologies.”

“With Toshiba’s NAND flash memory chips powering all Violin Flash Memory Arrays, technology companies are able to take full advantage of the capabilities that solid state storage can deliver,” said George Bouchaya, vice president and chief technology officer at the Institute of Strategic Storage Planning and Investment of Toshiba America Electronic Components, Inc. “Twenty-five years after inventing NAND flash, Toshiba continues to drive flash innovation forward, and it’s exciting to see solutions like the Fixstars/Violin rendering system come to life.”

Scott Frankel, lucille Product Manager at Fixstars Corporation commented, “The Violin Memory’s Violin 6616 represents an enormous step forward in performance data storage. Working hand in hand with the parallel processing capabilities of Fixstars’ lucille global illumination renderer, these two innovative products open the door to new horizons in computer graphics rendering.”

Fixstars is demonstrating the high speed particle rendering at SIGGRAPH 2012, August  7th to 9th in Los Angeles, with Violin Memory and Toshiba, who developed the NAND flash memory embedded in the Violin 6616. To see us at the SIGGRAPH, please visit booth #761. If you have any inquiry or would like to make an appointment, please contact us at the following e-mail address or telephone number,.

 

Fixstars Opens the Source Code of lucille

 Fixstars announces today the source code release of their renderer, lucille. Available free of charge, use the source code to build custom render configurations and create additional features. Customizing lucille from source is the most effective way to build special purpose rendering software, without having to code from scratch.

 lucille is a Global Illumination renderer achieving high-speed results without sacrificing quality or physical accuracy. With physically accurate lights and materials, image-based lighting, and ultra-low noise ray tracing, lucille delivers uncompromising results with dramatically reduced computing times. It’s RenderMan compatibility leverages well established workflows, rendering fully ray traced images in a highly-optimized, cross-platform solution.

 lucille has an enormous speed and compatibility advantage over other ray tracing renderers, since it has been built on the M³(M-cubed) Platform. M³ is a software platform which drives multi-core hardware resources to maximum efficiency, accelerating the rendering process dramatically. M³ provides a highly-optimized foundation for multi-core architectures and cluster environments, with performance gains scaling as additional CPUs are added.

 This video processing speed increases with the number of CPU cores. M³ efficiently maximizes resources, spawning render threads across all available cores, without having to change a single line of code.

High Speed Rendering Made Easy with the M³ Platform

How M³ accelerates a renderer

How M³ accelerates a renderer

Maximizing hardware resources, especially with a physically accurate render engine, is critically important in CG production. However, programming for parallelization is very complex and most developers are required to focus on software feature development rather than core efficiency. M³ provides Parallel Framework and Primitives Libraries that yield optimized processing through multi-threading, SIMDization, and Gand cluster execution automatically. Focus on feature development without having to write a single line of parallelization code.

To learn more about M³(M-cubed) Platform

lucille Beta for Artists

 In addition to the lucille source code, we have released lucille Beta for artists, also implemented on the M³  Platform. Try it and get to know lucille’s features, speed, and image quality. Since lucille is still in development, we look forward to your feedback.

Download lucille

The follwings is available for downloaded.

You will need to sign up the Fixstars User Account and send us a download request, in order for the download.

lucille at SIGGRAPH LA 2012

SIGGRAPH 2012

Fixstars is showcasing the lucille, Accelerated Global Illumination Renderer at SIGGRAPH 2012, in Los Angels from August 7-9, booth #761, in the Khronos Pavillion.

 We will be demonstrating the high speed performance of lucille at the booth,  rendering a high resolution scene which is required the vast computational power. lucille has an enormous speed and compatibility advantage over other ray tracing renderers since lucille has been built on M³ (M-cubed)  Platform which drives multi -core hardware resources efficiently, and accelerates rendering process dramatically.Drop by our booth,  experience the performance of M³ (M-cubed) Platform and get lucille Source Code with M³ (M-cubed) Platform .

Also, the high speed particle rendering is demonstrated using Violin Memory which is the industrial fastest flash array. In addition to the high speed performance of lucille, Violin Memory accelerates particle rendering dramatically, achieving high speed I/O performance which is enough to replace multiple racks of traditional disk arrays. Although particle rendering is required  large data sets and random access, Violin Memory resolve the bottlenecks with  petabytes (PB) of capacity and high aggregate bandwidth.

SIGGRAPH is the world’s premier conference and exhibition on computer graphics and interactive techniques. SIGGRAPH ASIA is a sister event of the SIGGRAPH established from 2008. It is currently one of the largest conferences held in Asia which brings together industry experts, researchers, companies and students under one roof to share information and experiences.

SIGGRAPH  2012 Overview

Date Conference: August 5-9Exibition: December 7-9
Hours Tuesday, 7 August 9:30 am – 6 pm
Wednesday, 8 August 9:30 am – 6 pm
Thursday, 9 August 9:30 am – 5 pm
Location Los Angeles Convention Center
1201 South Figueroa Street
Official Website http://s2012.siggraph.org/

About Violin Memory

Violin Memory is the fastest most salable solid state Memory. Built from the ground up to provide scalable high performance memory infrastructure for large dataset applications, Violin’s memory appliances lower the total cost of memory (TCM), providing an 80% savings in space and power.

To learn more about Violin memory, visit: http://www.violin-memory.com/

About M³ (M-cubed) Platform

lucille has been implemented on the M³ (M-cubed), a software platform which works in multiple multi-core environments. M³ will enable application developers to pull out the full performance of multi-core processors, while increasing development efficiency.

To learn more about M³, visit: http://www.fixstars.com/en/lucille/what/m-cubed/

lucille at SIGGRAPH ASIA 2011

Fixstars will be showcasing the lucille, Accelerated Global Illumination Renderer at SIGGRAPH ASIA 2011, in Hong Kong from December 12-15, booth #D07 in the Khronos Pavillion.
Drop by our booth, talk with us, and get the lucille alpha version for free
Our booth will be co-hosted together with the Accelcoder X, the Fast and High Quality Blu-ray Encoding Software of Fixstars.

SIGGRAPH is the world’s premier conference and exhibition on computer graphics and interactive techniques. SIGGRAPH ASIA is a sister event of the SIGGRAPH established from 2008. It is currently one of the largest conferences held in Asia which brings together industry experts, researchers, companies and students under one roof to share information and experiences.

SIGGRAPH ASIA 2011 Overview

Date Conference: December 12-15
Exibition: December 13-15
Hours Tuesday, 13 December 9:30 am – 6 pm
Wednesday, 14 December 9:30 am – 6 pm
Thursday, 15 December 9:30 am – 5 pm
Location Hong Kong Convention and Exhibition Center, Hall 3G
1 Expo Drive, Hong Kong
Tel: 852 2582 8888
Official website http://www.siggraph.org/asia2011