DreamWorks contributes Lossy Compression to OpenEXR 2.2
August 8, 2014

DreamWorks contributes Lossy Compression to OpenEXR 2.2

GLENDALE, CA — DreamWorks Animation has contributed code and license of their patented, production-quality Lossy Compression Codec to the OpenEXR standard (openexr.org). DreamWorks Lossy Compression strategically removes information from High Dynamic Range (HDR) floating-point images, a format common in Animation and VFX production pipelines. The technique, invented by principal engineer Karl Rasche, preserves visual quality without visible compression artifacts, has excellent decoding performance, and reduces HDR floating-point image file sizes to approximately the size of fixed-range images. The compression quality has been validated by VFX working on DreamWorks Animation feature films. 

OpenEXR creator, Industrial Light & Magic (ILM) has confirmed that DreamWorks Lossy Compression will be released in OpenEXR 2.2 on August 10, 2014 and will be API compatible with current releases. Unlike current lossy OpenEXR compression formats, DreamWorks Lossy Compression allows variable lossiness to balance visual quality and infrastructure resources. Additionally, this contribution includes a significant speedup to the commonly used PIZ compression scheme. 

“The lack of visible compression artifacts at the default compression level is impressive, and repeated application on the same image exhibits no further loss after the first compression. Compared to other lossy methods in the standard, this new technique produces higher-quality images at similar compression ratios, and decompressing the images is very fast,” said Ed Hanway, ILM R&D Supervisor. “We welcome the addition of the technique to OpenEXR.” 

“Images with DreamWorks Lossy Compression are stunningly high quality, and artifact-free. Not dealing with the artifacts common with other compression techniques has saved valuable time in our tight production process,” notes Dave Walvoord, VFX supervisor on How To Train Your Dragon 2. 

DreamWorks Lossy Compression is another instance of the studio’s contribution toward open source software advancing imagery technology.