sex264 (updated almost daily by now) - Page 6

SeeMoreDigital · 7th January 2005, 22:40

Try this Recode2 encode, generated at just 150Kbps at full DVD resolution!

Cheers

Fl0ppy · 7th January 2005, 23:35

how i can play a .mp4?

I not use nero recode

Blue_MiSfit · 8th January 2005, 00:37

If you have Recode installed you have Ahead's AVC/AAC decoders installed and you should be able to play that MP4 on any decent player like Core, or Media Player Classic

kurt · 8th January 2005, 00:40

or install for example ffdshow (05.01.05) and activate decoding h264 in it....

@SMD: nice vid @ such low bitrate !

bond · 8th January 2005, 01:52

Quote:

Originally posted by Fl0ppy
how i can play a .mp4?
I not use nero recode

this has been answered a thousand times already

ChronoReverse · 8th January 2005, 02:07

I suppose I might be missing something, but the two clips at 2.03MB aren't too far apart in quality.

PlazzTT · 8th January 2005, 02:27

Quote:

Originally posted by ChronoReverse
I suppose I might be missing something, but the two clips at 2.03MB aren't too far apart in quality.

On my PC, if I view them both in full-screen mode (1280x1024) there is very obvious ringing and typical MPEG artifacts around the edges (eg- just over peoples shoulders).

This effect is much less in the X264 encode, IMO.

JoeBG · 8th January 2005, 08:23

Okay I made a comparison between 3, 2 and 1 b-frame
the other settings were:
bitrate=1000, CABAC, deblocking filter -2:-2, Max refrence frames 5, , prediction method= 1 temporal, max key frame interval= 250, subpixel refinement precision=4=always b-frames

For me, Max b-frames=3 brings best results (with celticdruids newest filters), sadly the clips are too big to up them somewhere. I´ll make shorter (=smaller

) ones next time

My problem is, I don´t find a muxer to convert them into MP4 (fourcc is h264). The following don´t work:
- 3ivx: as expectet
- mp4creator60: I really thought he can do this
- mp4box: I was sure this would work
- MP4UI: as expectet

Any new ideas?

The second thing is the question, how to make the first pass faster. The second pass ist as fast as with Ateme. But the first pass with Ateme is much faster. Can´t we have something like "discard first pass" like in Xvid?

plonk420 · 8th January 2005, 09:47

maybe bobolo will share MP4 Toolbox to the general public?

akupenguin · 8th January 2005, 10:10

Quote:

Originally posted by JoeBGermany
My problem is, I don´t find a muxer to convert them into MP4 (fourcc is h264).

ffmpeg -vcodec copy -i x264.avi -y x264.mp4
I don't know how compliant the result is, but I can play it in MPlayer.

Quote:

The second thing is the question, how to make the first pass faster. The second pass ist as fast as with Ateme. But the first pass with Ateme is much faster. Can´t we have something like "discard first pass" like in Xvid?

Feel free to use faster options on the 1st pass. subpixel refinement, # reference frames, and psub8x8 are good candidates to reduce at little cost in the 2nd pass's quality.
I don't have a single option to do the adjustments for you, because the speed/quality tradeoffs are still changing, so any effort I spend now optimizing the decisions will become obsolete soon.

Doom9 · 8th January 2005, 11:17

unfortunately, I'm also getting a 2nd pass error. Interestingly enough, when I closed VDub, opened it and created my job manually (I had set up the 2nd pass job in the job list and wanted to run it this morning), things worked just fine. In absence of a vdub log entry I have no idea what went wrong when trying to start the 2nd pass for the first time.

mezzanine · 8th January 2005, 11:38

My second pass gave an error at ~ 60% (rate.c something error) with b-frames enabled.
With all advanced options switched off everything worked smoothly. I'm gonna try another 2-pass encoding tonight with the default options and see what happens.
Very good quality so far..

IgorC · 8th January 2005, 13:52

sorry for offtopic but i need advise.

Right now i´m encoding a DVD Harry Poter pr. Azkaban( 2h 9m) in recode2 Nero H.264 (too slow options: ref 6, b 3, Extra , MV 511, 2pass.... ) but i want to test sex264 in Vdub. will be there any problems (artefacts) if i´ll do it at the same time?
can i put pause in recode2 encoding?

JoeBG · 8th January 2005, 14:12

Quote:

Originally posted by akupenguin
ffmpeg -vcodec copy -i x264.avi -y x264.mp4
I don't know how compliant the result is, but I can play it in MPlayer.

Works very good for me, thanks.

Another question maybe is what brings best quality in the field motion search precision. I choosed sub-8x8 PFRAME search and 4x4INTRA search together and nothing else. Can I make it better?
Which combination brings best results?

Tommy Carrot · 8th January 2005, 14:52

Quote:

Originally posted by JoeBGermany
Which combination brings best results?

If everything is working correctly, then enabling all of them should give the best quality (and slowest encoding speed too

).

DeathTheSheep · 8th January 2005, 20:28

ALL of them? You gotta be kidding... :-)

Wow, I typically don't use B-frames... With no deblocking and slower encode time, what use are they?

The only things that really help are yer reference frames. I typically stick in 15. Compared to 1, 2, or even 3, this thing packs one heck of a punch in the lamb chops, baa

Cheers, baa

Blue_MiSfit · 8th January 2005, 22:03

hmm well after watching more thoroughly my 95mb family guy x264 episodes I noticed a strange problem with multiple consecutive b-frames.

Occasionally some of the b-frames will recieve absurdly high quantization, blocking very heavily (see my samples), while some of their neighboring b-frames recieve normal quantization, and look nice.

I'm not short on bitrate... considering 95-99% of the frames look incredible.

I'm not complaining, rather inquiring if ppl are aware of this rate control bug.

Samples:
clean frame

b0rked b-frame (previous frame's neighbor)

Leak · 8th January 2005, 22:54

Quote:

Originally posted by Blue_MiSfit
Samples:
clean frame

b0rked b-frame (previous frame's neighbor)

Maybe it's just me, but this doesn't look like it was just overquantized - there's so much residue from the surrounding frames that something else must have gone wrong here...

Not that I'm well-versed in codec development, mind you...

np: Pan Sonic - Käytävä / Corridoor (Kesto (234.48:4) (Disc 3))

akupenguin · 9th January 2005, 00:38

Yes, that's drastically overquantized.
H.264@qp=37 is ~equivalent to mpeg4@qp=18, while H.264@qp=46 would be equivalent to mpeg4@qp=50 if the scale went that high. Any frames that look good at qp=37 are just because they happen to motion compensate very well based on a P-frame at lower quantizer. (unless ffdshow displays some wacky quantizer scale?)
That said, it is a bug and I will try to find it.

Blue_MiSfit · 9th January 2005, 01:28

I am suspicious of ffdshow's OSD... Its "input bitrate" is never correct with h264, so perhaps the quantizer scale is incorrect as well.

I am re-encoding with 1 bvop and will let you know if these frames are better.

[edit]
encode finished. Problems persist with b-frames.

I am using ffdshow to decode, a very recent version from celtric_druid, his jan 5 build.

Could this be a decoder issue and if so how can I try another one?
[/edit]

~misfit

7th January 2005, 22:40	#101 \| Link
SeeMoreDigital Life's clearer in 4K UHD Join Date: Jun 2003 Location: Notts, UK Posts: 12,227	Try this Recode2 encode, generated at just 150Kbps at full DVD resolution! Cheers __________________ \| I've been testing hardware media playback devices and software A/V encoders and decoders since 2001 \| My Network Layout & A/V Gear \|

8th January 2005, 00:37	#103 \| Link
Blue_MiSfit Derek Prestegard IRL Join Date: Nov 2003 Location: Los Angeles Posts: 5,989	If you have Recode installed you have Ahead's AVC/AAC decoders installed and you should be able to play that MP4 on any decent player like Core, or Media Player Classic __________________ These are all my personal statements, not those of my employer :)

8th January 2005, 00:40	#104 \| Link
kurt sidekick Join Date: Apr 2004 Location: old Europe Posts: 610	or install for example ffdshow (05.01.05) and activate decoding h264 in it.... @SMD: nice vid @ such low bitrate ! __________________ greets, kurt. Pioneer PDP-427 XA \| Popcorn Hour NMT C-200 \| Sony STR-DB 840 QS \| Canton Ergo 91 DC

8th January 2005, 02:07	#106 \| Link
ChronoReverse Registered User Join Date: Sep 2002 Posts: 143	I suppose I might be missing something, but the two clips at 2.03MB aren't too far apart in quality. Last edited by ChronoReverse; 8th January 2005 at 02:18.

8th January 2005, 08:23	#108 \| Link
JoeBG stupid Join Date: Sep 2004 Location: Cologne Posts: 638	Okay I made a comparison between 3, 2 and 1 b-frame the other settings were: bitrate=1000, CABAC, deblocking filter -2:-2, Max refrence frames 5, , prediction method= 1 temporal, max key frame interval= 250, subpixel refinement precision=4=always b-frames For me, Max b-frames=3 brings best results (with celticdruids newest filters), sadly the clips are too big to up them somewhere. I´ll make shorter (=smaller ) ones next time My problem is, I don´t find a muxer to convert them into MP4 (fourcc is h264). The following don´t work: - 3ivx: as expectet - mp4creator60: I really thought he can do this - mp4box: I was sure this would work - MP4UI: as expectet Any new ideas? The second thing is the question, how to make the first pass faster. The second pass ist as fast as with Ateme. But the first pass with Ateme is much faster. Can´t we have something like "discard first pass" like in Xvid? __________________ cu Joe ------------------------ freedom is just another word for nothing left to loose Last edited by JoeBG; 8th January 2005 at 09:06.

7th January 2005, 23:35	#102 \| Link
Fl0ppy Registered User Join Date: May 2002 Posts: 26	how i can play a .mp4? I not use nero recode

8th January 2005, 09:47	#109 \| Link
plonk420 amd & h.264 fanboy Join Date: Jun 2002 Location: NTSC Posts: 420	maybe bobolo will share MP4 Toolbox to the general public?

8th January 2005, 11:17	#111 \| Link
Doom9 clueless n00b Join Date: Oct 2001 Location: somewhere over the rainbow Posts: 10,579	unfortunately, I'm also getting a 2nd pass error. Interestingly enough, when I closed VDub, opened it and created my job manually (I had set up the 2nd pass job in the job list and wanted to run it this morning), things worked just fine. In absence of a vdub log entry I have no idea what went wrong when trying to start the 2nd pass for the first time. __________________ For the web's most comprehensive collection of DVD backup guides go to www.doom9.org

8th January 2005, 11:38	#112 \| Link
mezzanine Registered User Join Date: Jun 2002 Posts: 186	My second pass gave an error at ~ 60% (rate.c something error) with b-frames enabled. With all advanced options switched off everything worked smoothly. I'm gonna try another 2-pass encoding tonight with the default options and see what happens. Very good quality so far..

8th January 2005, 13:52	#113 \| Link
IgorC Registered User Join Date: Apr 2004 Posts: 1,315	sorry for offtopic but i need advise. Right now i´m encoding a DVD Harry Poter pr. Azkaban( 2h 9m) in recode2 Nero H.264 (too slow options: ref 6, b 3, Extra , MV 511, 2pass.... ) but i want to test sex264 in Vdub. will be there any problems (artefacts) if i´ll do it at the same time? can i put pause in recode2 encoding? Last edited by IgorC; 8th January 2005 at 13:56.

8th January 2005, 20:28	#116 \| Link
DeathTheSheep <The VFW Sheep of Death> Join Date: Dec 2004 Location: Deathly pasture of VFW Posts: 1,149	ALL of them? You gotta be kidding... :-) Wow, I typically don't use B-frames... With no deblocking and slower encode time, what use are they? The only things that really help are yer reference frames. I typically stick in 15. Compared to 1, 2, or even 3, this thing packs one heck of a punch in the lamb chops, baa Cheers, baa __________________ Recommended all-in-one stop for x264/GCC needs on Windows: Komisar x264 builds!

8th January 2005, 22:03	#117 \| Link
Blue_MiSfit Derek Prestegard IRL Join Date: Nov 2003 Location: Los Angeles Posts: 5,989	hmm well after watching more thoroughly my 95mb family guy x264 episodes I noticed a strange problem with multiple consecutive b-frames. Occasionally some of the b-frames will recieve absurdly high quantization, blocking very heavily (see my samples), while some of their neighboring b-frames recieve normal quantization, and look nice. I'm not short on bitrate... considering 95-99% of the frames look incredible. I'm not complaining, rather inquiring if ppl are aware of this rate control bug. Samples: clean frame b0rked b-frame (previous frame's neighbor) __________________ These are all my personal statements, not those of my employer :)

9th January 2005, 00:38	#119 \| Link
akupenguin x264 developer Join Date: Sep 2004 Posts: 2,392	Yes, that's drastically overquantized. H.264@qp=37 is ~equivalent to mpeg4@qp=18, while H.264@qp=46 would be equivalent to mpeg4@qp=50 if the scale went that high. Any frames that look good at qp=37 are just because they happen to motion compensate very well based on a P-frame at lower quantizer. (unless ffdshow displays some wacky quantizer scale?) That said, it is a bug and I will try to find it.

9th January 2005, 01:28	#120 \| Link
Blue_MiSfit Derek Prestegard IRL Join Date: Nov 2003 Location: Los Angeles Posts: 5,989	I am suspicious of ffdshow's OSD... Its "input bitrate" is never correct with h264, so perhaps the quantizer scale is incorrect as well. I am re-encoding with 1 bvop and will let you know if these frames are better. [edit] encode finished. Problems persist with b-frames. I am using ffdshow to decode, a very recent version from celtric_druid, his jan 5 build. Could this be a decoder issue and if so how can I try another one? [/edit] ~misfit __________________ These are all my personal statements, not those of my employer :) Last edited by Blue_MiSfit; 9th January 2005 at 01:35.