MPEG4 Modifier v1.3.4 Released (Updated 2005-Dec-14) - Page 19

Elias · 27th March 2006, 09:31

Quote:

Originally Posted by celtic_druid

There is a modified version of ffmpeg that can convert (not re-encode) DivX3 to MPEG-4. After that it will work with MPEG4Modifier.

I've read this before, but never gotten a link to any software that can do this.

squid_80 · 27th March 2006, 10:14

Sigh.

http://forum.doom9.org/showthread.php?t=85229

DFKT · 27th March 2006, 14:14

Great - thanks for your help, people!

I'll try it ASAP.

My DivX3 movies are of some historical/archaeological value to me, and in backing them up on DVD I'd like to have them at least in a correct aspect ratio (can't do anything against the sucky video-quality), so I don't have to re-correct it in MPC/VLC all the time while viewing...

EDIT: It worked fine, those nifty tools that can be found here in this forum are amazing!

Elias · 27th March 2006, 14:16

Quote:

Originally Posted by squid_80

Sigh.

http://forum.doom9.org/showthread.php?t=85229

It's not like I haven't seen that thread before. In fact, I've even posted in it. Still, that DivX3toMPEG4 tool isn't working.

celtic_druid · 27th March 2006, 14:46

Worked when I tried it some time ago. A few artifacts so it could probably do with some work, but overall it worked fine.

Elias · 27th March 2006, 14:46

Quote:

Originally Posted by celtic_druid

Worked when I tried it some time ago. A few artifacts so it could probably do with some work, but overall it worked fine.

I must have used it incorrectly then.

olnima · 10th April 2006, 13:43

Are there any commandline-switches in "MPEG4Modifier 1.34" to set (Pixel/Display-) AR and save changed avi-file? If not, would someone be so kind to add them?

Thanks
Olnima

ReinerSchweinlin · 17th April 2006, 14:18

download-mirror is here:
http://www.hardware-test.de/doom9/ffmpeg-mstomp4.rar

_ck_ · 18th April 2006, 17:27

I have an idea, how about the ability to actually remove n-vops without re-encoding right in the tool?

Perhaps this is possible to do
because the 3ivx directshow filter can do it:
http://www.3ivx.com/support/tbd_451.html
read at the bottom:

Quote:

The 3ivx DirectShow Media Muxer can be used to remove NVops from an already coded bitstream

The reason why n-vops need to be removed is that they cause some
hardware players to get the audio out of sync. Apparently this is true,
it's not a vbr vs cbr issue or a packed bitstream issue. For every
n-vop, the audio will slowly get out of sync. You can read about that here:
http://www.jarnot.com/twiki/bin/view...=print.pattern

Quote:

N-VOPs - This is the rather interesting thing I discovered, that, to my knowledge, no one else has discovered. The Philips 642 seems not to recognize N-VOPs...in other words, it just skips right over them. This leads to audio synch problems. The audio can be brought back in synch by fast forwarding or rewinding a few frames, but then it will go back out of synch the next time it encounters more N-VOPs.

I experience this very problem from time to time on my dvp642
(which is one of the most popular divx/xvid players in the usa)

Perhaps you can make a checkbox to remove n-vops by duplicating the
previous or next frame available?

ReinerSchweinlin · 18th April 2006, 17:30

correct me if I#m wrong, but isn´t that exactly what mpeg4modifier and it´s cli counterpart does? See link above.

celtic_druid · 18th April 2006, 17:58

3ivX would be removing when remuxing to mp4/mov. mp4box can do the same. Doesn't help for avi though since it doesn't support VFR.

No problems with N-VOP's here on my MTK based standalone. Just another reason not to by an ESS based player.

_ck_ · 18th April 2006, 18:16

Quote:

Originally Posted by ReinerSchweinlin

correct me if I#m wrong, but isn´t that exactly what mpeg4modifier and it´s cli counterpart does? See link above.

mpeg4modifier does not remove nvops from already non-packed bitstreams from what I have experiemented with - though it does detect them better than any other tool including gspot, etc.

I may be misunderstanding something but from what I am seeing, it IS possible to have a bitstream that is not packed yet still has nvops and causes this sound sync issue?

Here is a screenshot of the situation existing:

and sure enough the sound slowly goes out of sync on that video

Quote:

Originally Posted by celtic_druid

3ivX would be removing when remuxing to mp4/mov. mp4box can do the same. Doesn't help for avi though since it doesn't support VFR.

No problems with N-VOP's here on my MTK based standalone. Just another reason not to by an ESS based player.

Well some hardware players don't even care about QPEL or GMC but that doesn't mean it's not a good idea to avoid/remove it

I'm still going to hope someone can come up with a way to remove/replace nvops without re-encoding...

ps. where can I find a list of non-ESS hardware players?

SeeMoreDigital · 18th April 2006, 18:21

Quote:

Originally Posted by _ck_

ps. where can I find a list of non-ESS hardware players?

Here's a place where you can start your player search: -

http://forum.doom9.org/showthread.php?t=91616

And try using the forums "search" facility

dukey · 19th April 2006, 04:22

my player has this problem exactly

'N-VOPs - This is the rather interesting thing I discovered, that, to my knowledge, no one else has discovered. The Philips 642 seems not to recognize N-VOPs...in other words, it just skips right over them. This leads to audio synch problems. The audio can be brought back in synch by fast forwarding or rewinding a few frames, but then it will go back out of synch the next time it encounters more N-VOPs. '

Its a yamada player tho, its pretty old. No idea what chipset.

squid_80 · 20th April 2006, 09:06

Quote:

Originally Posted by _ck_

I'm still going to hope someone can come up with a way to remove/replace nvops without re-encoding...

Would you be able to post a small .avi file that exhibits the problem somewhere?

celtic_druid · 20th April 2006, 09:18

I wouldn't call it a problem. Decoders that can't handle N-VOP's are the problem. XviD only drops frames when two are 100% identical.

Here's a clip anyway.
I-VOPs: 1 (1.00%)
P-VOPs: 1 (1.00%)
B-VOPs: 0 (0.00%)
S-VOPs: 0 (0.00%)
N-VOPs: 98 (98.00%)

http://celticdruid.no-ip.com/test/nvop.avi

Without n-vops XviD would have had to encode 99 P-VOP's even though it is just 100 identical frames. Only does it when bframes aren't enabled to, which I guess would be hard to avoid in _ck_'s case since DXN doesn't allow interlaced content and B-VOP's.
edit: I guess not being about to disable N-VOP's even for 100% identical frames could be seen as a problem. ie. qpel isn't a problem since you can disable it if you want.

squid_80 · 20th April 2006, 12:50

Well if you did want to change N-VOPs into coded P-VOPs, you'd have to:
a) change the VOP header from not coded to coded
b) insert a run of binary 1s of length equal to (number of macroblocks)-1 followed by a 0
c) pad with 1s to the next byte boundary (I think)
d) adjust the avi headers and index

Unfortunately I don't have time to write something to do it right now.

_ck_ · 20th April 2006, 17:02

Yeah that example was a mpeg2 interlaced with a quick'n'dirty re-encoding to mpeg4, (attempting to) preserve interlace. But of course the interlace has nothing to do with the nvop problem. The capture source obviously caused the problem by inserting duplicate frames to try to stay in sync (possibly when unavoidably dropping real frames, hardware lag, etc)?

Not sure if it's a good idea to post it as a clip publically but it looks like the problem is at least understood anyway?

You short-circuited my brain on what little I do know about mpeg4 format with your technical solution but I look forward to one day when you do have the time to tackle it

I am very grateful for any tools created to fix this issue without re-encoding, so many thanks whenever you finally get the chance...

SeeMoreDigital · 20th April 2006, 20:24

Quote:

Originally Posted by celtic_druid

I wouldn't call it a problem. Decoders that can't handle N-VOP's are the problem. XviD only drops frames when two are 100% identical.

Here's a clip anyway.
I-VOPs: 1 (1.00%)
P-VOPs: 1 (1.00%)
B-VOPs: 0 (0.00%)
S-VOPs: 0 (0.00%)
N-VOPs: 98 (98.00%)

http://celticdruid.no-ip.com/test/nvop.avi

Without n-vops XviD would have had to encode 99 P-VOP's even though it is just 100 identical frames. Only does it when bframes aren't enabled to, which I guess would be hard to avoid in _ck_'s case since DXN doesn't allow interlaced content and B-VOP's.
edit: I guess not being about to disable N-VOP's even for 100% identical frames could be seen as a problem. ie. qpel isn't a problem since you can disable it if you want.

Hi CD,

I would like to conduct some more "hardware playback" tests with MPEG-4 SP N-VOP encodes.

Could you please generate a 30 second (or so) sample please?

Cheers

_ck_ · 21st April 2006, 06:59

I will attempt to find/create a clip that is suitable for open posting by tomorrow...

27th March 2006, 14:14	#363 \| Link
DFKT Registered User Join Date: Nov 2004 Posts: 12	Great - thanks for your help, people! I'll try it ASAP. My DivX3 movies are of some historical/archaeological value to me, and in backing them up on DVD I'd like to have them at least in a correct aspect ratio (can't do anything against the sucky video-quality), so I don't have to re-correct it in MPC/VLC all the time while viewing... EDIT: It worked fine, those nifty tools that can be found here in this forum are amazing! Last edited by DFKT; 27th March 2006 at 14:46.

10th April 2006, 13:43	#367 \| Link
olnima Registered User Join Date: Dec 2003 Posts: 216	Are there any commandline-switches in "MPEG4Modifier 1.34" to set (Pixel/Display-) AR and save changed avi-file? If not, would someone be so kind to add them? Thanks Olnima Last edited by olnima; 10th April 2006 at 14:05.

20th April 2006, 09:18	#376 \| Link
celtic_druid Registered User Join Date: Oct 2001 Location: Melbourne, Australia Posts: 2,171	I wouldn't call it a problem. Decoders that can't handle N-VOP's are the problem. XviD only drops frames when two are 100% identical. Here's a clip anyway. I-VOPs: 1 (1.00%) P-VOPs: 1 (1.00%) B-VOPs: 0 (0.00%) S-VOPs: 0 (0.00%) N-VOPs: 98 (98.00%) http://celticdruid.no-ip.com/test/nvop.avi Without n-vops XviD would have had to encode 99 P-VOP's even though it is just 100 identical frames. Only does it when bframes aren't enabled to, which I guess would be hard to avoid in _ck_'s case since DXN doesn't allow interlaced content and B-VOP's. edit: I guess not being about to disable N-VOP's even for 100% identical frames could be seen as a problem. ie. qpel isn't a problem since you can disable it if you want. Last edited by celtic_druid; 20th April 2006 at 09:24.

20th April 2006, 17:02	#378 \| Link
_ck_ Registered User Join Date: Apr 2006 Posts: 43	Yeah that example was a mpeg2 interlaced with a quick'n'dirty re-encoding to mpeg4, (attempting to) preserve interlace. But of course the interlace has nothing to do with the nvop problem. The capture source obviously caused the problem by inserting duplicate frames to try to stay in sync (possibly when unavoidably dropping real frames, hardware lag, etc)? Not sure if it's a good idea to post it as a clip publically but it looks like the problem is at least understood anyway? You short-circuited my brain on what little I do know about mpeg4 format with your technical solution but I look forward to one day when you do have the time to tackle it I am very grateful for any tools created to fix this issue without re-encoding, so many thanks whenever you finally get the chance... Last edited by _ck_; 20th April 2006 at 17:08.

27th March 2006, 10:14	#362 \| Link
squid_80 Registered User Join Date: Dec 2004 Location: Melbourne, AU Posts: 1,963	Sigh. http://forum.doom9.org/showthread.php?t=85229

27th March 2006, 14:46	#365 \| Link
celtic_druid Registered User Join Date: Oct 2001 Location: Melbourne, Australia Posts: 2,171	Worked when I tried it some time ago. A few artifacts so it could probably do with some work, but overall it worked fine.

17th April 2006, 14:18	#368 \| Link
ReinerSchweinlin Registered User Join Date: Oct 2001 Posts: 454	download-mirror is here: http://www.hardware-test.de/doom9/ffmpeg-mstomp4.rar

18th April 2006, 17:30	#370 \| Link
ReinerSchweinlin Registered User Join Date: Oct 2001 Posts: 454	correct me if I#m wrong, but isn´t that exactly what mpeg4modifier and it´s cli counterpart does? See link above.

18th April 2006, 17:58	#371 \| Link
celtic_druid Registered User Join Date: Oct 2001 Location: Melbourne, Australia Posts: 2,171	3ivX would be removing when remuxing to mp4/mov. mp4box can do the same. Doesn't help for avi though since it doesn't support VFR. No problems with N-VOP's here on my MTK based standalone. Just another reason not to by an ESS based player.

19th April 2006, 04:22	#374 \| Link
dukey Registered User Join Date: Dec 2005 Posts: 560	my player has this problem exactly 'N-VOPs - This is the rather interesting thing I discovered, that, to my knowledge, no one else has discovered. The Philips 642 seems not to recognize N-VOPs...in other words, it just skips right over them. This leads to audio synch problems. The audio can be brought back in synch by fast forwarding or rewinding a few frames, but then it will go back out of synch the next time it encounters more N-VOPs. ' Its a yamada player tho, its pretty old. No idea what chipset.

20th April 2006, 12:50	#377 \| Link
squid_80 Registered User Join Date: Dec 2004 Location: Melbourne, AU Posts: 1,963	Well if you did want to change N-VOPs into coded P-VOPs, you'd have to: a) change the VOP header from not coded to coded b) insert a run of binary 1s of length equal to (number of macroblocks)-1 followed by a 0 c) pad with 1s to the next byte boundary (I think) d) adjust the avi headers and index Unfortunately I don't have time to write something to do it right now.

21st April 2006, 06:59	#380 \| Link
_ck_ Registered User Join Date: Apr 2006 Posts: 43	I will attempt to find/create a clip that is suitable for open posting by tomorrow...