[M4IF Technotes] Nancy video codec

[Eileen Liu]

Hi, Perhaps this reply is no longer relevant, but here's our experience with Nancy in the mobile device market. I was closely evaluating multiple solutions in the market for encoding and playing video on mobile devices, for the major wireless company I work for. We had almost decided on licensing Nancy, until hands-on experience with their products left us very disappointed. The application was real-time encoding, streaming and playback of super-QQCIF (i.e. 160x128) video. While MPEG-4 based codecs easily do 10fps encoding concurrent with capture, Nancy seems to capture almost full-res frames in the capture pass, and then encodes post-capture in a second pass. The second pass takes at least as long as the first one, halving encoding performance. Further, this buffering workflow limits Nancy's recording capability on the Nokia 7650 to about 10 seconds. In contrast, an MPEG-4 codec from a Chinese firm we evaluated encoded up to 10 _minutes_ of live video at 10fps. We do believe that any technology that truly achieves Nancy's published technical specifications is worthy of respect and wide market acceptance. However, while Nancy has garnered impressive press, we found their performance numbers to be misrepresented. On a fast X-scale processor, a software Nancy codec does replay at their promised 40 fps, but only for a postage-stamp sized video (about QQQCIF). This would translate, by extrapolation, to about 10fps for QQCIF - a market-average performance. Its 30-fps software encoding performance claim is only valid for high-end PCs, not limited-resource devices. Their mobile device frame rate, even given the two-pass workflow, averages 5fps. Nancy's compression efficiency is much more limited than they claim – baseline MPEG-4 achieves 1/2 to 2/3 bit rate for the same quality. On the limited-color Nokia 7650 display, their decoded video looks terrible even at higher bit rates – it is heavily contoured and very patchy. For delivery, Nancy can stream QQCIF video over a 56.6 Kbps channel only at a maximum of 6 fps. Speaking of press, a report in EE Times from last year stating that Nancy is a leading candidate for standardization by China Mobile (CMCC) seems inaccurate – CMCC has stated unhappiness with that report. On the non-technical front, Nancy is a proprietary codec with only http streaming capabilities and closed licensing terms. MPEG-4, on the other hand, is an open standard with open licensing; offering multiple vendors, open marketplace, true streaming as well as http streaming possibilities, and built-in players for almost every platform. Nancy's porting ability to new platforms appeared sluggish at best. In summary, we found MPEG-4 solutions to be vastly superior to Nancy for most handheld devices and camera-ready cell phones. We also found MPEG-4 to be the superior solution for mobile-to-PC, PC-to-mobile and mobile-to-mobile video e-mail and streaming.  Regards,
Eileen
ramakrishna_kakarala   agilent.com wrote:
> 
> Hi,
> 
> Does anyone know how the Nancy
> video codec from Office Noa compare
> with MPEG-4, Simple Profile?  Nancy
> is supposed to be efficient in computation,
> and has apparently made it into the J-Phone and
> Sharp Zaurus. I'm curious how it compares
> to MPEG4 in coding efficiency.
> 
> Thanks
> Ram
> 
> ---------------------------------
> Ramakrishna Kakarala
> Agilent Technologies
> 3175 Bowers Avenue MS 87H
> Santa Clara  CA  95054
> ramakrishna_kakarala   agilent.com
> (408) 970-2467
> _______________________________________________
> Technotes mailing list
> Technotes   lists.m4if.org
> http://lists.m4if.org/mailman/listinfo/technotes

---------------------------------
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/mp4-tech/attachments/20030421/49f6b281/attachment.html