Once again we've come to the last page of a codec comparison and are asking the dreaded question "which is best?". Actually, if you're a member of my forum you know better than to ask that questions but I'll try to answer it anyway. But be aware that you are not only entitled to see things differently, you might actually see things differently and can in all honesty disagree with my findings (with the proper respect though and some reason.. it's hard not to notice the detail difference between RV9 and XviD but you might value this differently). After all, you don't have my eyes and I don't have yours.
Let's start with 3ivX. The codec has come a long way since the last comparison and the results are a far cry from the devastating results the codec delivered last time it was tested. Nevertheless, I feel the codec could still use some polishing. Maybe b-frames would do it some good (it's the only MPEG-4 codec in this test not to offer this feature), and more accurate bitrate control couldn't hurt either because 3ivX was constantly oversized.
Let's give a last round of applause to DivX3 / SBC as it is leaving the house. Last time the codec found its match, this time it has been beaten in 3 out of 3 tests and that by a fair margin. Consequently, the codec will not be featured in any future comparison (along with every other codec that has not been improved between two comparisons).
I'm afraid to say that the most notable difference between DivX 5.0.5 and DivX 5.1.1 has been the significant drop in speed, with a relatively low gain on quality. True, the codec got a bit more detailed but in that department I feel that RV9 has made more progress. And DivX5 had visible problems with animated content (see the 2nd shot of Futurama, I think this could actually be a bug in the codec).
To me, ffvfw feels like an early XviD version. It certainly had potential but it's not quite there yet. As the underlying libavcodec is more than just an MPEG-4 encoder it remains to be seen which part will be optimized in the future.
NeroDigital has certainly made an impression, mostly in the speed department. Though what I suspected in the beginning has come true: We have a tradeoff between speed and quality and the former was clearly favored. Since NeroDigital is supposed to be used in standalone recorders and hence had to be limited in computational complexity, the results are still good, but ND isn't as detailed as its competition and it is somewhat blocky as well (though those are well hidden in many cases). I'm also no fan of being restricted to one application and the non editability of the Recode output is a major nono to me. I'd like to see this codec freed from the Recode corset.
RV9's EHQ mode manages to make a positive impression, especially in close-ups. But during scenes with a lot of motion you can still notice RV9's tendency to smooth out details. EHQ mostly helped RV9 in the Matrix test whereas I wasn't so impressed with the codec's improvements in the SPR test. It also scored really well in "its" domain, the Futurama test.
VP6 is an interesting newcomer codec. To be used in China's EVD system, that codec might not have been such a bad choice. On2 claims superior performance to H.264, WMV9 and RV9. I can't say anything about H.264 since I have never tested such a codec, but depending on the circumstance they might have a point about WMV9 (which keep in mind hasn't been improved since the last comparison whereas all the other codec have been). As for RV9, I'll leave that up for you to decide. VP6 didn't excel in any domain and could definitely use some polish when it comes to animated features.
Finally, XviD is my winner this time. It came in first in both Matrix and SPR test, and second in Futurama. Also, for the first time in a long testing series, there were no glitches and no problems in the encoding setup (I never had to redo a single encoding session).
So, if you want a fast, detailed and good looking codec, XviD is your codec. If you're mostly encoding animated content or value less blocks over details RV9 seems like a pretty good choice. DivX5, VP6 and NeroDigital are situated somewhere in between, none excelling in any area. I felt that NeroDigital was the least detailed of the trio. 3ivX and ffvfw could use some maturing.
If you're eyeing playback on a DivX compatible DVD player, DivX is obviously your best shot because all players are tested against this codec. However, these players can also handle generic MPEG-4 content, specifically XviD. Get a MediaTek based player and you can enjoy the superior performance of XviD without any problems (don't use GMC.. but since it wasn't a suggested option for the test you probably won't use it anyway). I think that these players should also handle both 3ivX and NeroDigital but it might require a FourCC change (I'll test that when I test standalone players). Ahead is also trying to work together with standalone manufacturers to include NeroDigital not only for playback but also for recording, but seeing how long it took DivXNetworks to have a presentable supply of players this might take a while.
While VP6 is an option in the EVD standard, the entire standard still sounds a little suspect and it remains to be seen if it will ever be available outside of China if it ever takes off.
Today, some codecs no longer come as a simple VFW encoder and DirectShow decoder solution:
3ivX has an entire filter suite. Besides the MPEG-4 video encoder there is an AAC Low complexity audio encoding filter, an MP4 file muxer (so that you can turn AAC and MPEG-4 video into a specs compliant MP4 file), and playback filters to handle MP4, multichannel AAC audio (even High Efficiency AAC) and generic MPEG-4 video.
DivX5 ships with a media player and an application to tweak the bitrate distribution of the video codec, called EKG. DivXNetworks also offers a one click DVD encoding tool called Dr. DivX.
ffvfw will soon be merged with ffdshow, the most versatile video playback filter that can handle almost anything you throw at it. ffvfw also comes with an application to turn AviSynth scripts to AVI frameserving files for applications that can't handle AviSynth input. ffvfw not only supports MPEG-4 encoding, it can also output DivX3, MPEG-1, MPEG-2, H.263, WMV7, WMV9, HuffYUV, MJPEG and a native lossless format FFV1.
NeroDigital is not available as regular VFW codec. Rather, it comes integrated into Ahead's Recode one click DVD backup suite, which not only features NeroDigital output but can also handle DVD-R output. Recode's NeroDigital output comes as an MP4 file with AAC (High Efficiency if you like, both 2ch and 5.1ch is possible), subtitles and chapters (though the latter two require that you use Ahead's Showtime player, you won't get those features in any other player). At this point it is not possible to split or edit those MP4 files or process any input but DVD. Ahead plans to allow other input by including NeroDigital output into their NeroVision video editing tool (which can then also split MP4 files).
RealNetworks not only offers their own video codec, but also has both 2ch and 5.1ch audio codecs. The RealMedia container also supports streaming, subtitles and chapters. Editing is problematic though as RV9 editing isn't a feature of mainstream video editing applications. It is also only possible to use RV9 outside of the RM container when using Matroska.
3ivX, ffvfw, NeroDigital, RV9 and XviD also support aspect ratio flags in the video, so it is possible to encode a movie anamorphically (as it's done with 16:9 movies on DVD). This requires support by either player, container or both.
I've tested an internal alpha version of 3ivX so we can obviously expect at least bugfixed versions in the future.
DivX6 has already been announced but only the future will tell if it's really going to be what was said, and in any case the release is still far off.
NeroDigital output will be added to at least NeroVision which also allows editing and non DVD input (though you can achieve the same by using the NeroDigital DS encoding filter already). Ahead is also trying to get NeroDigital into standalone players, both players and recorders. Due to this, quality improvements in the future might be limited though.
VP6 is still under development and even during the testing a new and improved build has been released. We'll see how that works out next time.
With the release of the XviD betas, there won't be any new features, only bugfixes till the official 1.0 release. But of course there's a future beyond version 1.0.
Other than the existing codecs, H.264 will most likely make an impact in 2004. There are a number of companies already working on codecs, so hopefully we'll be able to compare some of them against existing codecs soon.
Now if you have any questions, please have a look at the FAQ.
This document was last updated on January 20, 2004