As part of the Mozilla development grant Timothy in collaboration with other xiph hackers has been hard at work on improving the theora encoder. I am happy to share an updated report that Monty has put together. I will jump to the zinger:

“Test versions of Thusnelda are pulling *ahead* of h264 in terms of objective quality as bitrate increases”

These objective measurements are based on peak signal-to-noise ratio (PSNR) values which are not always identical to subjective quality measurements. That being said PSNR is pretty standard objective measurement and good for same content comparisons (1)

Image (1) svntheora_shirt_crop2.png for post 3731

New Theora encoder 1.5x

Image (2) orgtheora_shirt_crop.png for post 3731

Original Theora Encoder 1.5x

Why is this good news for Wikimedia? Wikimedia only supports free file formats. So these improvements mean that every new video uploaded to commons in the near future will be on par with contemporary industry standard high quality proprietary codec. This highly reduces the subjective quality rational for using proprietary codecs.

Why are free codecs important? Wikimedia (and anyone else that wants to switch to free formats) won’t have to pay millions of dollars to in licensing costs to use the h.264 codec and won’t have to sacrifice quality in the process. More importantly it means anyone can encode or decode these files without paying for a license to do so. This means both free and proprietary software can support this format.  Where as previously only controlled free as in beer distributions like adobe flash could support video on the web.   It enables free software projects like firefogg to package the encoder and give it away for free. It helps opens up the video communication platform for distributed two-way communication.

I should also point out the Open video conference is happening mid June for people interested in an open video event.