StreamingMedia.com: RealVideo comes out on top

[karl_lillevold]

As posted on helix-client-announce@helixcommunity.org

StreamingMedia.com says RealVideo is the best video codec, followed by Apple’s H.264, Flash and Windows Media.

One of the key findings from StreamingMedia.com’s report was that the quality of the best Flash and H.264 codecs still trailed RealVideo, often by a significant margin. To research the reports, report author Jan Ozer, produced a six-minute test file composed of 38 scenes representing typical business, sports, and entertainment videos, along with several animations and still image pans and zooms. The reports analyzed video quality in up to five configurations—modem, 3GPP, 100 Kbps, 300 Kbps and 500 Kbps—and compared frame quality, temporal and color quality and playback smoothness. Buyers can download all videos and still image files used in the analysis, along with a convenient interface for viewing the video files and still images.

Original StreamingMedia.com article to buy for a hefty $295
http://www.streamingmedia.com/research/codecs.asp
...or the summary (which is free )
http://www.macsimumnews.com/index.php/archive/windows_media_loses_ground_to_apples_h264_on2_flash_codec/

The Helix Community is proud to have RealVideo in its family and hearty congratulations to Karl, Neelesh, Atul and the entire RealVideo team.

Note: RealNetworks did not commission this study, it was a completely independent endeavor.

[Sagittaire]

Yes perhaps good conclusion for this test with these codecs but as you know:
- QT7 AVC is very crap H264 codec
- WMV9 with WME9 is very old codec now
- VP7 is very more powerfull than VP6

Actually RV10 can't fight with good H264 codec (Nero, Ateme, Elecard, x264) for all bitrate and all resolution.

[Sharktooth]

another case of incompetence... and they want even money for a crappy codec comparison...
however i agree RV10 is better than CrapTime(tm) AVC (only in their wet dreams), VMW9 and maybe VP6 (i won't bet on this one though).

[Sirber]

IMO RV10 is more stable in quality than VP7.

[IgorC]

Real just needs some financing to keep development. So marketing does the best it can.


Edit: gram. errors

[^_^x]

i am netral, OK!

if you try to encode hight quality video source(like DVD) to other(x264, rmvb, etc), you'll can't find the true difference among them. all result are good quality.

this test using a bad source video, and just show a little different betwen rv10 and x264


Source history
--------------------------------------------
movie: full metal panic - fumoffu
genre: anime
trace source :
- some one download divx version from internet
- some one decode divx movie to vcd format, for rent
- some one rent that vcd, and re-encode to divx. we call them A


Sample frame
---------------------------------------------
- take 6 second from A with vcdcutter pro save to mpeg2. call them B. size 142,367 KB
- trying to encode B to rmvb with realanime4 rc3. error, only support yuv
- open B with vcdcutter pro and save to avi, xvid maximum quality call them C, this will be the source. size 14,206 KB
- what a bad source

Filtering
----------------------------------------------
Convolution3D(preset="animeHQ")
FieldDeinterlace(full=false, blend=true, chroma=true)
aWarpSharp(16.0, 2, 0.5, 0)
MSharpen(threshold=10, strength=100, highq=true)

Setting
----------------------------------------------
all the test using realanime4 rc3+powerpack
Audio: aac+v2 48 kbps
container: mkv
the result captured from media player classic by pressing "Print Screen" button and paste to photoshop cs. save to jpeg with maximum quality.

RMVB result
----------------------------------------------
codec: rv10
quality: hight
mode: 2 pass VBR
bitrate: 700
rv10 dropdupe: yes
bframe: 3
size: 2,160 KB

http://anane.wyger.nl/img/1_result.rmvb.0700.jpg
http://anane.wyger.nl/img/2_result.rmvb.0700.jpg

X264 result
----------------------------------------------
codec: x264
quality: hight
mode: 2 pass ABR
bitrate: 700
filtering: +1
size: 602 KB (WOW)

http://anane.wyger.nl/img/1_result.x264.0700.jpg
http://anane.wyger.nl/img/2_result.x264.0700.jpg


since the quality and size have a far differences, increase the x264 bitrate
bitrate: 1600
size: 1,349 KB

http://anane.wyger.nl/img/1_result.x264.1600.jpg
http://anane.wyger.nl/img/2_result.x264.1600.jpg

[^_^x]

still have different size
bitrate: 2000
size: 1,673 KB

http://anane.wyger.nl/img/1_result.x264.2000.jpg
http://anane.wyger.nl/img/2_result.x264.2000.jpg


again
bitrate: 2500
size: 2,097 KB

http://anane.wyger.nl/img/1_result.x264.2500.jpg
http://anane.wyger.nl/img/2_result.x264.2500.jpg



note:
- rmvb with bitrate 450 still have good quality with size +/- 1,300 KB
- x264 output size, WOW!!!

[Sirber]

seems RV has troubles reaching the right size... comparison should be made using the same file size :)

[Dark-Cracker]

you can reach a good final filesize using the alternative curve control + the overhead calc formula used in autorv10 ;)

[^_^x]

that's all???, where's the fighting ;)

sirber:
completly right, coz when i encode the full movie, both of them reach filesize 80-83 mb(bitrate 450).

all:
as always, rmvb is better than x264

[Sirber]

as always, rmvb is better than x264kidding? :scared:

[^_^x]

come on, this is realmedia thread, sirber!

btw, can you post your encoding result picture, that you said x264 is better than rmvb.

[Sirber]

I had one long ago, but RV oversized while being inferior to x264. The comp has been removed from my server a long ago.

To make it quick, why do you think I wrote RealAnime LE (original name of RealAnime 4) with only x264 and AAC+ v2 support? Me, The "RV Nerd"? Think about it... :)

[edit]

2000kbps and 2500kbps for anime? Try at 400kbps or lower. Where I encode now (318kbps), RV10 is useless.

[^_^x]

my compy:
p4 2.0ghz, 256mb/2700, seagate 80

i dunno, but when comparing both codec in anime movie, always get better on rmvb. that's why i prefer rmvb than x264.

or should i resetting something to get better x264?

[Sirber]

700kbps?

Upgrade to RealAnime 4.1.0. RC3 used in your test is a bit old. Something wierd is at 318kbps I get higher Q than your 700kbps x264 shot.

[Sagittaire]

x264, nero AVC, Ateme AVC or elecard AVC ar by far really really better than RV10

[^_^x]

700kbps?

Upgrade to RealAnime 4.1.0. RC3 used in your test is a bit old. Something wierd is at 318kbps I get higher Q than your 700kbps x264 shot.

ok2, x264 is the winner. till now i'm trying and trying, but rmvb always get better shot.

so where's my problem?
my 32 mb riva tnt????

note: i just install your application, realanime. should i install x264 codec too?. is it not a packet with your app?.

[karl_lillevold]

x264, nero AVC, Ateme AVC or elecard AVC ar by far really really better than RV10
If you compare H.264 Baseline or Main, it's not really that far at all, if at all. Some devices only support Baseline, like iPod. When the first H.264 DVD players come out, they may only support a subset of Main, maybe not even CABAC. We will see...

[CruNcher]

but if you true rv10 can't be better as H.264 it's based on the principles that have been developed for H.264 only that H.264 was heavily developed and even Real was involved how could something that is developed by dozens of top researchers from arround the world from a dozen different companies and institutes be inferior to something that's developed by 1 of those companies with less research power (except maybe Microsoft they have such a big Research Sector and it grows every year) , it can be as good but not better see VP7, RV10 or VC-1 (could be technicaly better we gonna see soon) and yes maybe Microsoft has the next Einstein somewhere in their Research facilities ;P

[siddharthagandhi]

- QT7 AVC is very crap H264 codec
- WMV9 with WME9 is very old codec now
- VP7 is very more powerfull than VP6

Exactly. I have quicktime, the AVC is a bunch of cow manure. WMV is outdated, VC-1, when it comes out, would be an appropriate comparison. And duh VP7 is more powerfull than VP6. Also, they didnt even test Ateme/Nero Digital codec, which in my opinion is the best codec ever in the history of video.

edit: lol i didnt even see the second paragraph of sagittaire's post...hez like reading my mind i agree 40000 percent with his post