Quicktime 7 MPEG-4 ASP/AVC encoding?

[bond]

Originally posted by Sagittaire
QT7 AVC use 1 bframe adaptative.i didnt see any qt7 avc stream till now which placed b-frames adaptively. do you have one? if yes, i would love to get a sample

1, 2, 3 or more frames don't change really the result (visually and for metric) for other MPEG4 AVC (x264 and NDAVC). oh what a claim...
so using more than 1 b-frame is totally useless in x264 and nd avc in your opinion ("visually and for metric")?

[Sagittaire]

Originally posted by bond
i didnt see any qt7 avc stream till now which placed b-frames adaptively. do you have one? if yes, i would love to get a sample


QT7 AVC HD trailer (Batman Begin for exemple) use one adaptative bframe. Perhabs dev use more advanced codec for these HD encoding.


oh what a claim ...
so using more than 1 b-frame is totally useless in x264 and nd avc in your opinion ("visually and for metric")?

Well make the test with x264 and try to detect visually sample with 1, 2, 3 or 4 bframe. IMO it's impossible simply because in most case encoder use 0 or 1 bframe ...

Here result with metric for x264:
0 bframes: 42.96 dB ( 0000/3075 for bframes/total)
1 bframes: 43.09 dB ( 1144/3075 for bframes/total)
2 bframes: 43.10 dB ( 1354/3075 for bframes/total)
3 bframes: 43.11 dB ( 1438/3075 for bframes/total)
4 bframes: 43.11 dB ( 1471/3075 for bframes/total)

high diff between 0 and 1, little diff between 1 and 2/3/.../N

[babayaga]

Originally posted by Sharktooth
Ok here are the QT7 encodes:
450kbps (http://www.aziendeassociate.com/h264/hp-450.mp4)
900kbps (http://www.aziendeassociate.com/h264/hp-900.mp4)


The link is broken. Could you restore it ?

Thx

[Sharktooth]

Sorry, DNS problems...
I'll fix it asap.

[Sharktooth]

Ok DNS server gone nutz so i moved the files to another host.
QT7 450kbps encode (http://www.webalice.it/f.corriga/h264/hp-450.rar)

QT7 900kbps encode PART1 (http://www.webalice.it/f.corriga/h264/hp-900.part1.rar)
QT7 900kbps encode PART2 (http://www.webalice.it/f.corriga/h264/hp-900.part2.rar)

However look also at this xvid encode (900kbps):
http://www.webalice.it/f.corriga/misc/hp2-xvid-900-ULR.part1.rar
http://www.webalice.it/f.corriga/misc/hp2-xvid-900-ULR.part2.rar

Do you still believe QT AVC is a good codec? Well, i dont think so.

[akupenguin]

Originally posted by Sagittaire
Well make the test with x264 and try to detect visually sample with 1, 2, 3 or 4 bframe. IMO it's impossible simply because in most case encoder use 0 or 1 bframe ...

So encode something other than HPII. Sure, the 1st B-frame makes more difference that the 2nd, but don't call them useless.
720x480x24, 950kbit/s, avg qp=18
bframes=0 PSNR:46.41 fps:7.33 B: 0/129968
bframes=1 PSNR:46.72 fps:7.42 B:57407/129968
bframes=2 PSNR:46.80 fps:7.96 B:75447/129968
bframes=3 PSNR:46.80 fps:8.12 B:83409/129968

[SeeMoreDigital]

Originally posted by Sharktooth
Ok DNS server gone nutz so i moved the files to another host.
QT7 450kbps encode (http://www.webalice.it/f.corriga/h264/hp-450.rar)

Do you still believe QT AVC is a good codec? Well, i dont think so.

Hi Sharktooth,

I've just compared the Harry Potter II QuickTime AVC 450Kbps encode with the Nero AVC 450Kbps encode (ftp://mood.ateme.com/beta/hp2-ateme-450k-hp.mp4).

Apart from Nero's encode being 504KB smaller than QuickTime's... Nero's looks better... to my eyes!

[ac-chan123]

to compare mux files isn't good. demux and compare the raw streams.

[SeeMoreDigital]

Originally posted by ac-chan123
to compare mux files isn't good. demux and compare the raw streams. This should not matter too much as both the QuickTime and Nero encodes have been muxed into the .MP4 container!

That said, even after de-muxing both to RAW .h264 streams, there is still a 492KB difference!


Cheers

[Elias]

Anyone who knows when it'll be out for Windows? Last I heard was something like September in this thread. Not sure if that was official though.

[eb]

Originally posted by SeeMoreDigital
That said, even after de-muxing both to RAW .h264 streams, there is still a 492KB difference!

Is this difference caused by video data volume or structure of headers for video blocks ?

eb

[SeeMoreDigital]

Originally posted by eb
Is this difference caused by video data volume or structure of headers for video blocks ? Your questions are a little over my head eb :(


Cheers

[Elias]

http://fileforum.betanews.com/detail/QuickTime_for_Windows/948389372/1 Finally, a beta. Too bad I can't run Windows any more (Linux only now).

[damrod]

in was reading apple page about h264..

Use Scenario Resolution & Frame Rate Example Data Rates
Mobile Content 176x144, 10-24 fps 50-60 Kbps
Internet/Standard Definition 640x480, 24 fps 1-2 Mbps
High Definition 1280x720, 24p 5-6 Mbps
Full High Definition 1920x1080, 24p 7-8 Mbps

640/480@1-2Mbps ?? i use this with xvid...if i choose to use h264 it's to use 600-700kbps for near the same quality !!

i think apple don't understand video encoding and mpeg4 ;-)


btw quicktime7 basic can play fullscreen video??

[Elias]

in was reading apple page about h264..

Use Scenario Resolution & Frame Rate Example Data Rates
Mobile Content 176x144, 10-24 fps 50-60 Kbps
Internet/Standard Definition 640x480, 24 fps 1-2 Mbps
High Definition 1280x720, 24p 5-6 Mbps
Full High Definition 1920x1080, 24p 7-8 Mbps

640/480@1-2Mbps ?? i use this with xvid...if i choose to use h264 it's to use 600-700kbps for near the same quality !!

i think apple don't understand video encoding and mpeg4 ;-)


btw quicktime7 basic can play fullscreen video??Actually, that depends a lot on the actual source. If it has a lot of dark scenes, and is overall a good compressionable source, h.264 won't even be needed since XviD would do the job pretty much great.

[bond]

640/480@1-2Mbps ?? i use this with xvid...if i choose to use h264 it's to use 600-700kbps for near the same quality !!
well i think big companies never realised the power of xvid and opensource. i assume they simply ignore opensource codecs as its maybe not a professional and serious solution in their eyes

so when talking about mpeg-4 they might refer to encoders like apple's or sorenson's, which cant stand a chance against xvid and therefore need a higher bitrate to get the same quality

[damrod]

off course but divx is a professionnel/commercial codec no?

and it gives better quality than sorenson or apple codec !!!

i think the reason is : "if i pay too much it must be good...because i can complain latter"

stupid reason but common reason...in lot of compagnies

they don't like working and free solution ...it's weird for them lol

[Elias]

Actually, I find XviD, or MPEG-4 Part 2 (Apple's version not included, since it's on par with DivX 3.11) much better than AVC. Even if it's Nero Digital AVC. I don't know why, but I've never gotten good results with AVC. It's way overrated and hyped in my opinion.