updated Nero Digital MPEG-4 ASP Codec out, with new quality enhancements [Archive]

SeeMoreDigital

2nd July 2004, 22:23

There's more good news too!

Nero's DSdec filter has been fixed. It now displays anamorphic AVI and MP4 encodes correctly, including ones generated by other codecs.

As mentioned on a previous thread, I was concerned that this version of NVE (like the previous version) might not work on a PC running a clean WinXP install. Well there's no problem here either now!

So to sum up so far. Nero's decoder can now cope with AAC 'audio only' streams in AVI and MP4. Mpeg4 'video only' streams in AVI and MP4. Mpeg4/AAC streams in AVI and MP4 and Mpeg4/mp3 streams in AVI and MP4, using either the ShowTime player or WinMedia Player 9.

So it would seem they have been listening. Thanks Ahead.

I shall generate some encodes and report tomorrow.

Cheers

bobololo

3rd July 2004, 02:10

For those who are interested, here are brief info about this release for the part I know (ie the video codec). So this new version of Recode (2.0.0.18) sports the latest build of our ASP encoder which has been quite improved on several points since the former december release.

Unfortunately, the current version of Recode doesn't completely exploit all these new features. But this is only a way to keep much more for the next major release of Recode expected to be released soon ;)

Firstly, the encoding speed was drastically improved. We couldn't resist to do something to compensate the speed decrease due to "extra" quality mode :) (see below) This feature was *not* enabled in the current Recode (and don't ask me why ;)) so stay tuned for the next major release to enjoy this enhancement.

Then come error resilience and streaming tools which have been updated for integration within Recode 2.1, so you won't be able to benefit from them right now.

And last but not least, the encoding quality gets a significant boost with the introduction of the well known Rate-Distorsion Optimization inherited from our work on H.264 and also further works on the rate control. All these improvements were gathered under the new "extra" quality mode. As one can guess this extra quality has a cost and therefore don't expect to encode at 100 fps with this new quality level ;)

Well, enough speech for now. So if you're curious or simply want to evaluate the extra quality, you're welcome to give it a try and feel free to comment about it.

-- bobololo

SeeMoreDigital

3rd July 2004, 09:58

Hi bobololo, nice to have you back...

I have to say my first impressions are happy ones. I generated some "extra quality" encodes last night/this morning and they do look much, much better.

In fact I'm in the process of generating a much fuller/tougher test using the "extra quality" mode to see how well this new version compares to both DivX and XviD.

I have noticed that although the decoder filter is able to display 720x480/576 anamorphic content correctly, the encoder still has problems with movies that have a aspect ratio of 2.35:1 or 2.40:1. In order for these movies to be encoded and displayed correctly the black mattes have to be cropped away. Which of course is not a problem to do but begs the question...why can't you keep them in if you want to!

The new decoder seems very clever but what exactly is "Use Line21 / Mpeg-4 SP Video Padding"?

I'm sure I could thing of more questions but hey, better let the other forum members get their foot in the door!

Cheers

LostMP4

3rd July 2004, 12:15

Originally posted by bobololo
But this is only a way to keep much more for the next major release of Recode expected to be released soon ;)

Can you translate "soon" in a more understandable and precise way? :p
(and the H264 codec? I guess it'll come out "after" that "soon")

bond

3rd July 2004, 14:00

ok here are my findings:

gui:
- no real changes/still the same: limited to only dvd input, no .avs or other input supported in recode2 itself
- still only 1 audio and 1 subs track possible (it was said that new recode2 will allow more than 1)

encoder:
- it has been said that all modes have been quality tuned ->
- i tested the new extra quality mode, which enables trellis + rdo (as also used by xvid when enabling vhq)
- new fast first analysis pass is already available (which also enables a new faster rate control, which shouldnt hurt the quality compared to the old rc) but when choosing it, the encoder doesnt want to connect to the muxer anymore => first pass still has to be done in full "extra mode" -> slow (slower than xvids fast first pass by far!) but nero's 2nd pass is faster than xvid here with the same featureset enabled (b-frames, qpel, gmc, custom quants...)
- import/export of xvid style matrices

nero mp4 muxer:
- the encoder doesnt want to connect to windows default avi-mux filter anymore (produced buggy output anyways) -> only nero muxer
- nero muxer doesnt output to .avi (so there is no way anymore to output directly to .avi)
- the nero mp4 muxer only connects to the nero encoder or .mp4 files opened via the nero parser filter (no support for the default win source filter nor .avi in any way)
- the muxer supports now the splitting of .mp4 files!!! .mp4 splitting only works with the new "nero file writer" filter (not with the default windows one) and it splits at keyframes (would not be necessary in .mp4)
- the muxer now supports the setting of chapters!!! (milliseconds are not setable tough)
- interleaving is also setable now (in ms)
- the file only gets created once graphedit is closed it seems

output/quality:
- no variable framerate (good if you want to output to avi)
- the output .mp4 has the last frame of the input stream placed 3 times at the beginning
- the output .mp4 has a f*ed up aspect ratio set on the bitstream level. the 640x256 stream get displayed as 640x640 with both neros and 3ivx' decoder

i compared the quality between xvid and nd with the following settings: 2pass (650kbps) with hvs-best, extra/vhq4/tellis, qpel, gmc, default b-frames, no-psy/no-aq, mv range 64
- definitely more details than the old version i tested (still less than xvid)
- overall very dirty picture, more artifacts than xvid, especially around borders (maybe its also due to hvs-best, which nd doesnt seem to like (still xvid doesnt show this "dirtyness"))

other comments:
- after installing nve the nero parser filter gets used for all .mp4 files, which means you cant connect other filters like 3ivx' splitter or decoders, like ffdshow to it anymore. reinstalling 3ivx helps

all in all, a great enhancement, but still we need to wait to get a commercial encoder, which is able to clearly beat xvid ;)

Sharktooth

3rd July 2004, 14:36

I was never able to run Recode.
Even with this new version i still get the same errors
It wants multichannel plugin and dvd plugin...
I tried all the possible fixes with no luck.
The nice part is:
Please pay attention:
The Multichannel Plug-in is already included in the Nero 6 retailpackage and the Nero 6 webversion.
and
Please pay attention:
The DVD-Video Plug-in is already included in the Nero 6 retailpackage and the webversion.
I will go straight for a refund request.

LostMP4

3rd July 2004, 16:14

Originally posted by Sharktooth
I was never able to run Recode.
Even with this new version i still get the same errors
It wants multichannel plugin and dvd plugin...
I tried all the possible fixes with no luck.
The nice part is:

and

I will go straight for a refund request.

These plugins use different serials than the suite: you must open Nero Burning Rom, look under the help menù and add the plugin serial number

LostMP4

3rd July 2004, 16:18

Originally posted by bond
- the output .mp4 has a fucked up aspect ratio set on the bitstream level. the 640x256 stream get displayed as 640x640 with both neros and 3ivx' decoder

Using Recode and testing abnormal crops and/or resizes I always got an Aspect Ratio close to the right one (still waiting for a manual option...)

bond

3rd July 2004, 17:25

Originally posted by bond
i compared the quality between xvid and nd with the following settings: 2pass (650kbps) with hvs-best, extra/vhq4/tellis, qpel, gmc, default b-frames, no-psy/no-aq, mv range 64
- definitely more details than the old version i tested (still less than xvid)
- overall very dirty picture, more artifacts than xvid, especially around borders (maybe its also due to hvs-best, which nd doesnt seem to like (still xvid doesnt show this "dirtyness"))ok after this comparison with the hvs-best custom matrix, where xvid kicked nd's ass i redid the same encode at 650kbps, with h263 quants and the following settings:
xvid (rc4): h263, VHQ4 (RDO), ChromaMotion, QPel, GMC, Trellis, B-vops (2, other default)
nd: h263, Extra Mode (RDO, Trellis), QPel, GMC, 4MV, B-vops (1), no psy, mvrange 128

and as a result i was pretty much surprised but i would say that on the sample i encoded (matrix1, same as doom9 uses), with the settings i used, overall nd can definitely keep up with xvid detailwise!
providing in many scenes a higher amount of details than xvid (!) from my eyes of view (i compared them with the great avscompare tool), and in some scenes (morpheus sitting with the cable attached to his head, neo standing in front of the weapon shelf and agent smith close up interrogating morpheus) less details than xvid
still nd is overall a little bit more dirty than xvid, but that also has to do with the details i assume and i used 2 b-frames in a row in xvid (nd uses only 1), which might also make a difference
also the background of the nd encode seems to be a little bit "swimming", meaning having some movement (not only in plain walls)
and the hvs-best encode didnt really look good (in contrary to the xvid one) as i wrote above already

all in all great work ateme! i would say nd is finally in the race ;)

Sharktooth

3rd July 2004, 20:50

Originally posted by LostMP4
These plugins use different serials than the suite: you must open Nero Burning Rom, look under the help menù and add the plugin serial number
It tells me it's already there...
However those plugins are included in the NVE package and they should not be installed separately.
Once i install NVE and give it MY serial it have to work.

bobololo

3rd July 2004, 22:05

Originally posted by SeeMoreDigital
I have noticed that although the decoder filter is able to display 720x480/576 anamorphic content correctly, the encoder still has problems with movies that have a aspect ratio of 2.35:1 or 2.40:1. In order for these movies to be encoded and displayed correctly the black mattes have to be cropped away. Which of course is not a problem to do but begs the question...why can't you keep them in if you want to!

The new decoder seems very clever but what exactly is "Use Line21 / Mpeg-4 SP Video Padding"?

All these issues are closely related to the applications (Recode and Showtime) that I definitely don't know enough to reply to your questions. Regarding the Line 21, if I'm correct, it's a standard for caption data encoding.

-- bobololo

bobololo

3rd July 2004, 22:20

Originally posted by LostMP4
Can you translate "soon" in a more understandable and precise way? :p
(and the H264 codec? I guess it'll come out "after" that "soon")

When it comes to deal with release date, it's always hard to be accurate you know :) So I'm sorry but I can't be more precise.

-- bobololo

BoNz1

3rd July 2004, 22:41

bond, how does it look with psychovisual enhancements enabled? I forget but wasn't ateme going to revamp the psychovisual enhancements? Anyway, at 650kbps they should do more good than harm and so it might be worth testing :).

SeeMoreDigital

4th July 2004, 11:34

Originally posted by bobololo
All these issues are closely related to the applications (Recode and Showtime) that I definitely don't know enough to reply to your questions. Thanks for the reply. It's a strange one because if you're into generating non cropped and resized encodes (like I am) Recode2 encodes look like this during playback: -

http://img33.exs.cx/img33/8348/SMD_Actual_Output.jpg

When it should really look like this: -

http://img15.exs.cx/img15/2410/SMD_Correct_Output2.jpg

What is interesting though, if you play the Recode2 encode in say the ShowTime player or Media Player Classic and capture a still, you get this: -

http://img36.exs.cx/img36/9075/SMD_Nero_capture.jpg

Which is exactly how an anamorphic capture should look before the player corrects it.

Cheers

SeeMoreDigital

4th July 2004, 12:51

Has anybody else noticed, when playing .MP4 files, Nero's video directshow filter seems to work a lot better with other media players than it does with WMP9.

I've tried NVE (the new version) on a couple of PC's. I've even tried using WMP10 beta but it was the same.

As a matter of interest I noticed that WMP9/10 uses both Nero's audio and video filters but MPC uses only Nero's video filter. I've tried forcing MPC to use Nero's audio filter but I can't make it work.

I wonder if it's Nero's audio filter that making WMP9 unstable!

Any suggestions?

SeeMoreDigital

4th July 2004, 15:18

Also,

I've just noticed that Nero have added some new DSdec filters (NeQTADec.ax and NeQTVDec.ax). Which enable Mpeg4/AAC streams in .MOV to function with the ShowTime player!

Cheers

chilledoutuk

4th July 2004, 19:28

SeeMoreDigital I have looked into the resizing problem you raised and have found that if you disable cropping then the playback aspect is screwed but however if you select custom crop and put 0 in all the cropping amount boxes then the aspect the video is played back is correct.

I suspect this is to do with the program setting a PAR on the stream when you crop and when you don’t it sets a DAR which appears to be wrong.

Apart from that I have to say that I am very impressed with the quality improvement in the encoding part of Nero digital.

However I am having major problems getting the dshow playback filter to work in WMP 8or9 it does work in mpc but that does not recognise the Nero aac playback filter. (Maybe I could try core aac)

Using the new mp4UI tool I was able to extract the avi stream from the video files. To my astonishment the quality seemed to be improved when i played it back with ffdshow with the XViD decoder when compared to Nero Showtime.
I believe that this may be because of the ffdshow filter using a better resizing algorithm than that of Nero digital. It seems obvious to me that the next thing that should be improved is the mp4 directshow playback filters.

SeeMoreDigital

4th July 2004, 20:31

You're a star chilledoutuk!

Yes, using your method does indeed solve the AR problems, for both PAL and NTSC sources. It's still not perfect though!

I've found that if you use mp4UI to strip out the video stream from the .MP4 container and then run it thru' MPEG4modifier you get the following Aspect Ratio information...

For PAL 720x576 you get this: -

http://img7.exs.cx/img7/3923/SMD_Recode2_PAL.gif

And for NTSC 720x480 you get this: -

http://img12.exs.cx/img12/9342/SMD_Recode2_NTSC.gif

Thanks very much for working that out, as that's been pissing me off for months...

Cheers ;)

chilledoutuk

4th July 2004, 23:34

thats very strange when you encode from a anamorphic source such as PAL 16:9 and do not resize the par should be identicle to that of the source stream which would be 16:11 for PAL 16:9.

also i wish you could have more than one audio stream as i think that having directors comentry is a popular feature.

chilledoutuk

5th July 2004, 00:06

Ill asume that the custom PAR's you quoted are for 16:9 aspect sources.

in that case the custom PAR's that nero seems to be setting are very close to that of the MPEG-2 standard

for PAL 16:9 PAR = 16/11 = 1.4545454545454545454545454545455
neros setting = 36/25 = 1.44

for NTSC 16:9 PAR = 40/33 = 1.2121212121212121212121212121212
neros setting = 6/5 = 1.2

orginaly i thought why are nero using custom PAR's thats when i realised it calculates the par so that the video will be resized for apropriate playback resolution. This means that it is possible to take a 720x576 source and crop it, then custom resize to what ever dimension you wish.

I decided to test this use an masively anamorphic resolution of 480x480 (square pixel would have been somit like 640x368).

....and voila video is played back in perfect aspect and it looked alarmingly good quality ok im off to make a few more tests and i must say im very impressed with this nero recode 2 just wish it could open avs files then it would be exceptionaly useful.

GlitchOfDoom

5th July 2004, 03:24

whats the difference between "h263 quantinization" and "h264"?

chilledoutuk

5th July 2004, 06:22

although thats a bit off topic h.263 quantinization simply uses a quntisation method based on the h.263 video standard to make MPEG-4 asp video streams.

H.264 or AVC is the next generation video encoding technology and is not compatible with MPEG-4 asp technologies.

napx

5th July 2004, 09:04

nero's aac encoder my preferred audio encoder, and if the video quality becomes truly comparable to xvid when these new features mentioned are enabled (next month's release?), nero digital could become the next big thing.
P.S: what about updates to the aac encoder? more efficient coding for audio always means more bits to allocate for video!

LostMP4

5th July 2004, 09:11

Originally posted by napx
nero's aac encoder my preferred audio encoder, and if the video quality becomes truly comparable to xvid when these new features mentioned are enabled (next month's release?), nero digital could become the next big thing.
P.S: what about updates to the aac encoder? more efficient coding for audio always means more bits to allocate for video!

Nothing new in this release...

SeeMoreDigital

5th July 2004, 10:21

Originally posted by chilledoutuk
thats very strange when you encode from a anamorphic source such as PAL 16:9 and do not resize the par should be identicle to that of the source stream which would be 16:11 for PAL 16:9.Yes, lets hope they sort this oddity out sooner rather than later - mind you, I've only been banging on about it since early Jan. It really needs to be fixed before the new Kiss (.MP4) player is launched.

Originally posted by chilledoutuk
also i wish you could have more than one audio stream as i think that having directors comentry is a popular feature. Agreed, this would be useful. But I think it will come soon.

After all, Nero's ShowTime player has been able to cope with multiple AAC and MP3 streams since the launch of Recode2 and does so very well IMO. I wonder how well the new Kiss player will cope with this innovation also?

Cheers

Andrey

5th July 2004, 11:10

whats the difference between "h263 quantinization" and "h264"?
To clarify this a bit to you :)
h.263 is an ITU standart (or 'recommendation').
As h.262 (seems that it is mpeg 2 or mpeg 1 video coding scheme), h.264 and so on are.
Example of official titles:
H.263 (02/98) Video coding for low bit rate communication
H.262 (07/95) Generic coding of moving pictures and associated audio information: Video
As for h263 quantinization - h.263 coding scheme uses uncommon (based not on quant matrixes, but on one cooficient - 2) quantization method to other mpeg video coding schemes.
So, becasuse this quantization method lacks officail name (or it is simply not widely used, I don't know) many people refer to it as h263 quantinization...

chilledoutuk

5th July 2004, 17:21

I think that the use of custom PAR's is intentional and as long as the display aspect is maintained then who cares what par they are using.
The main problem is the incorect display aspect when cropping and resizing is disabled but this can be worked around as i have explained earler.

I might hasten to add that i have tested the latest recode encodes on the new XBMC (Xbox Meia Center) and was impressed with the quality and compatibility.

SeeMoreDigital

5th July 2004, 18:34

Originally posted by chilledoutuk
I think that the use of custom PAR's is intentional and as long as the display aspect is maintained then who cares what par they are using.
The main problem is the incorect display aspect when cropping and resizing is disabled but this can be worked around as i have explained earler. Well this problem is getting stranger and stranger. I've just generated another PAL 720x576 encode, using a different source, where the movie has an AR of 2.35:1 and this time (after de-muxing), MPEG4Modifier showed the PAR as being 16:9 PAL. Is that weird or what?

Originally posted by chilledoutuk
I might hasten to add that i have tested the latest recode encodes on the new XBMC (Xbox Meia Center) and was impressed with the quality and compatibility. I really love the fact that the XboX can play Mpeg4/AAC in MP4 :D

Cheers

aliumalik

5th July 2004, 20:25

Any significant changes to the AAC encoder?

chilledoutuk

5th July 2004, 20:32

SeeMoreDigital I think you are a little confused PAL DVDs use 2 anamorphic frame PAR's
PAL 16:9
PAL 4:3

When a DVD movie specifies the aspect ratio on the DVD casing it refers to the actual movie image not including the letterboxing black bars.

As you will know any movie DVD that has a wider aspect ratio than 16:9 will use black letterbox bars to pad the rest of the frame to make the frame size 720x576 so that when the frame is resized proper picture aspect is maintained.

http://www.aoul98.dsl.pipex.com/chilledoutuk/tutorials/aspectratio/images/2.5in1.3framexample.gif

If you do not crop the black borders off and do not resize the video sequence then the par required to display the video with correct aspect would be the same as that of the source video.

Which in your case would have been the PAL 16:9 PAR which is 16:11

I have made a guide for calculating aspect ratio which is

>>Here (http://www.aoul98.dsl.pipex.com/chilledoutuk/tutorials/aspectratio/aspect.htm)<<

SeeMoreDigital

5th July 2004, 20:58

Nope, I'm not confused.

I'll let you into a secret. I actually helped author some of the first test PAL DVD's way back in 1996/97.

This was when some of the manufacturers were debating weather they needed to release DVD titles in two formats. ie: 'Anamorphic (16:9) Widescreen' as we know it now and 'Letterbox (4:3) Version'

Needless to say 'Anamorphic Widescreen' is the only format necessary... regardless of weather the movies aspect ratio is 1.77:1, 1.85:1, 2.35:1, 2.40:1, etc, etc....

EDIT: Here's a link to my website (www.SeeMoreDigital.net/03_Video_Only_Info/Image_Pixel_Frame_Size_Totals.html) which details the popular image pixel frame sizes together with their total pixel counts, with or without matte overlays.

Cheers

chilledoutuk

5th July 2004, 21:27

OK lets back to the point nero is not misbehaving with regards to PAR unless you chose the disable option in cropping.

Well this problem is getting stranger and stranger. I've just generated another PAL 720x576 encode, using a different source, where the movie has an AR of 2.35:1 and this time (after de-muxing), MPEG4Modifier showed the PAR as being 16:9 PAL. Is that weird or what?

Here you seem to be expecting a PAR different to that of the source when you are not resizing the video frame this is wrong.

Nope, I'm not confused.

Ok I dont particlary care whether your confused or not but you are wrong with regard to the first quote.

I'll let you into a secret. I actually helped author some of the first test PAL DVD's way back in 1996/97.

So you should Know better with regard to Pixel Aspect Ratio's

SeeMoreDigital

5th July 2004, 21:45

Oooow harsh!

Well like you now, I'm just trying to get to the bottom of why Recode2's encoder performs like it does.

You yourself have already identified one major piculiarity when attempting to generat encodes from anamorphic 16:9 widescreen DVD's.

Why, when clicking on the 'Crop' tab should selecting the 'Disable' button create encodes with a different PAR than those when selecting the 'Custom Crop' button and manually entering zero under is beyond me... but there we go!

At least XviD and 3ivX make anamorphic encoding more straight forward.

Cheers

Sagittaire

5th July 2004, 22:20

Nero want real Picture Aspect and it's 16/9 for PAL 720*576: 2.35 is the Movie AR or Picture Aspect with crop ...

-16/9 Pal for Pixel AR or 16/11 for Pixel Aspect
-16/9 for Display AR or 16/9 for Picture aspect

SeeMoreDigital

5th July 2004, 22:51

Well the sooner we move away from having to encode using anamorphic image pixel frame sizes the better.

I'm all in favour of using true 16:9 and 4:3 image pixel frame sizes - much of which is already available via some Mpeg2 HDTV platforms.

Hopefully chipsets like Sigma's EM8620L/21L, which can decode Mpeg2 image pixel frame sizes upto 1920x1080 (2,073,600 total), will make it into more and more standalone players.

Maybe we'll even get some HD digital camcorders too!

Cheers

pogo stick

5th July 2004, 23:10

The problem with remuxing (to create chapters) audio and video with Nero parser, muxer and file writer stays in new version. Video is twice slower then audio. :(
Does remuxing existing mp4 (audio+video) with Nero direct show stuff ever worked in Graphedit right? Did anyone try it?

chilledoutuk

6th July 2004, 01:04

I agree that anamorphic encoding is not ideal and I am happy that HDTV keeps pixels square (the way they ahould be).

On the note of HDTV is recode going to support HDTV and video sources?

What concerns me about HD digital camcorders is the data rate of the dv video encoded in HDTV resolutions would result in video thats very difficult to transmit and store unless your using final cut pro HDTV version.

SiXXGuNNZ

6th July 2004, 01:06

Originally posted by bond
nero mp4 muxer:
- the encoder doesnt want to connect to windows default avi-mux filter anymore (produced buggy output anyways) -> only nero muxer
- nero muxer doesnt output to .avi (so there is no way anymore to output directly to .avi)
- the nero mp4 muxer only connects to the nero encoder or .mp4 files opened via the nero parser filter (no support for the default win source filter nor .avi in any way)
- the muxer supports now the splitting of .mp4 files!!! .mp4 splitting only works with the new "nero file writer" filter (not with the default windows one) and it splits at keyframes (would not be necessary in .mp4)
- the muxer now supports the setting of chapters!!! (milliseconds are not setable tough)
- interleaving is also setable now (in ms)
- the file only gets created once graphedit is closed it seems

do I need to use graphedit to split the files?

SeeMoreDigital

6th July 2004, 09:39

Originally posted by chilledoutuk
On the note of HDTV is recode going to support HDTV and video sources? Very nice idea!

But in order to do this I reckon they would have to make their GUI more flexible... perhaps they could include inputting raw Mpeg streams as well as .VOB and .DAT and DV.avi?

Originally posted by chilledoutuk
What concerns me about HD digital camcorders is the data rate of the dv video encoded in HDTV resolutions would result in video thats very difficult to transmit and store unless your using final cut pro HDTV version. Yes, data flow is a problem especially as the current DV.avi format in (SP mode) consumes around 225MB per min. Meaning a typical 60min DV tape stores approx 13,500MB (13.2Gb)!

As we already know, there are quite a few camcorders that encode straight to Mpeg2 but the bit-rate has to be kept horribly slow because the media (MicroMV and MiniDVD) is small on data storage!

So this begs the question, why not manufacture camcorders to allow us to store high bit-rate Mpeg2 data onto DV tape.... or even hard-drives?

Cheers

Morpheus_xx

6th July 2004, 17:03

Originally posted by bond
ok here are my findings:
...
- the nero mp4 muxer only connects to the nero encoder or .mp4 files opened via the nero parser filter (no support for the default win source filter nor .avi in any way)
...

I found out that it is still possible to use avi or avs input to encode within graphedit.

As before I have renamed it to recode.exe then I can use

AVS (Asynch. FileSource) ->ND Video Enc->ND Mux->ND Filewriter
or
AVI->...->ND Video Enc->ND Mux->ND Filewriter

So nice to have the new quality mode :)

SeeMoreDigital

6th July 2004, 17:48

Something in Nero is behaving weirdly!

After changing the 4CC code of some Mpeg4 AVI's to NDIG (which is Nero's 4CC code) I've found that MPC will play the AVI's but with WMP9, it's a very hit and miss affair, ie: some do and some don't!

Can anybody else confirm?

Cheers

chilledoutuk

6th July 2004, 19:09

I have also had major problems using the nero dshow filter with wmp9 as well.
I have managed to get it working in MPC but i had to install core aac as mpc dont acept the nero dshow filter which leads me to think that the audio decoding filter is what is causing the problems.

Thinking about it im not that botherd as MPC can skip the chapters in the video and also gives insight into compatibilty between nero and FAAD 2 which Xbox Media Center uses without having to test on the xbox repetively.

bond

6th July 2004, 20:45

Originally posted by SiXXGuNNZ
do I need to use graphedit to split the files? well recode2 will do the splitting during encoding i assume
to split existing files you need graphedit

Originally posted by pogo stick
The problem with remuxing (to create chapters) audio and video with Nero parser, muxer and file writer stays in new version. Video is twice slower then audio. :(
Does remuxing existing mp4 (audio+video) with Nero direct show stuff ever worked in Graphedit right? Did anyone try it? i didnt try it. can anyone confirm this?
make sure you use all filters nero offers (parser, muxer and the nero filter writer, i assume there is a reason why they have an own file writer (for example splitting only works with the nero file writer))

Originally posted by Morpheus_xx
I found out that it is still possible to use avi or avs input to encode within graphedit.
AVS (Asynch. FileSource) ->ND Video Enc->ND Mux->ND Filewriter
or
AVI->...->ND Video Enc->ND Mux->ND Filewriteryeah it works with the encoder placed in between (meaning reencoding), but not directly to the nero muxer, meaning converting from .avi to .mp4 without reencoding doesnt work with the nero filters

bobololo

6th July 2004, 22:12

@ SMD & chilledoutuk, regarding anamorphic encoding, I had a question that may appear very dumb, but what is the use to keep the black mattes instead of cropping them ?

From a coding point of view, if these mattes aren't 16 pels multiple in height, they can involve very high frequencies at the edges which implies less coding efficiency and ringing at the picture boundaries.

This phenomenon doesn't occur when they are cropped since the internal encoding system prodives some "padding" which insures the continuity of the pixels at picture boundaries. This even works when the cropped picture isn't x16 in height.

-- bobololo

SeeMoreDigital

6th July 2004, 23:29

Originally posted by bobololo
@ SMD & chilledoutuk, regarding anamorphic encoding, I had a question that may appear very dumb, but what is the use to keep the black mattes instead of cropping them ? For me it's simply a standardisation and flexibility issue with Recode2.

In my opinion anamorphic encoding should be done using PAR signalling only, and in this respect Recode2 is nearly there.

*By the way I'm not not saying DAR doesn't have it's place, it can be very useful if you know what your doing. But it's quite a powerful tool. For instance, with DAR it's possible to encode just the video element of an 2.35:1 image using all 576 vertical pixels (instead of just the cropped 432 pixels) and use DAR to introduce the black mattes artificially.

If you use 3ivx and XviD to generate anamorphic encodes both create the same anamorphic values even if the encodes are cropped or not. So to me it would makes sense for Recode2 to follow the same example!

Anyway as I said before, "Why, when clicking on the 'Crop' tab should selecting the 'Disable' button create encodes with a different PAR values than those when selecting the 'Custom Crop' button and manually entering zero under is beyond me... but there we go!".... This isn't very consistent and is very misleading and confusing!

Cheers

chilledoutuk

7th July 2004, 00:15

Recode does something very clever seamore that the 3ivx and xvid dont.
Let me explain what i mean...

when you crop you not changing the PAR required as the amount of distortion on the image aspect is the same as before you cropped therefore the PAR stays the same.

In xvid you just tell it the source is say PAL 4:3 and it puts that flag on the mpeg-4 stream but this is only useful if you do absolutly no resizing of the video frame.

Recode is quite clever in that it alows you to resize the out put to anything upto 720*576 pixels and it will calculate the par required to display this at the correct aspect when played back.

as i mentioned before i was able to take aomething 16:9 aspect and encode it at 480*480 and recode calculated the pixel aspect ratio for perfect playback aspecct.

I believe that this PAR calculation is only turned on when the image is cropped in recode.

hyper84

7th July 2004, 06:05

Originally posted by bond
well recode2 will do the splitting during encoding i assume
to split existing files you need graphedit

i didnt try it. can anyone confirm this?
make sure you use all filters nero offers (parser, muxer and the nero filter writer, i assume there is a reason why they have an own file writer (for example splitting only works with the nero file writer))

yeah it works with the encoder placed in between (meaning reencoding), but not directly to the nero muxer, meaning converting from .avi to .mp4 without reencoding doesnt work with the nero filters

I can confirm the problem with the video being twice as slow as the audio when trying to edit an existing ND .mp4 file. You would think you could edit a nero file with nero filters, eh?:D

BTW, I did used these filters:
Nero parser to split the ND file.
Nero Muxer
Nero file writer

Jvdhorst

7th July 2004, 16:17

Hmm I have another problem with this Nero digital. Tell me if I go to much off topic. But I'm having problems with the audio encoding. It happens when I recode some of the scenes from LOTR Return of the king (region 2/PAL) containing some havy LFE sounds. If I set the audio to AAC 5.1 48 khz (nomatter what bitrate or LC/HE) the mp4 contains very heavy clipping sounds when the low frequency sounds kicks in. Encoding to 2.0 doesn't suffer from this. Mp4ui it tells me the resolution of the sound is allways 24000hz and LC nomatter what settings have been used for encoding, but I don't know how comaptible mp4ui is with nero recode.

SeeMoreDigital

7th July 2004, 16:39

Hi Jvdhorst,

What players have you tried?

With regard to mp4UI it seems to work well for me and is getting better all the time. That said it's vital that you use the newest build. So head over to this thread (http://forum.doom9.org/showthread.php?s=&threadid=77577) and check out the latest RC2 'Private Build' on page5.

Also don't forget, Recode2 displays the full 'sample rate' (ie: 32000, 44100 and 48000Hz) even when using AAC-HE audio. Which will be halved when viewed using file readers/checkers. AAC-LC encodes should always show the full sample rate.

Cheers

chilledoutuk

7th July 2004, 17:34

Jvdhorst i had this problem as well with lotr rotk when i decided to encode it with 5.1 audio using AAC-HE at 256kbps. Its most noticable on the battle scenes with the big elephant things.

what i may try is encoding it with oagmachine with the same settings as what i used with recode. Something tells me its eaither a problem with the channel coupling or more likely recode having difficulty decoding the LFE in the ac3 5.1 stream.

i must say i was mortified as the encode took quite a while and the video quality was fantastic and i cant open the video in MP4UI becuase its too big 3GB (it is anamorphic though).

Does anyone know of mp4 apps that would alow me to extract the video from a .MP4 of this size?

chilledoutuk

7th July 2004, 17:38

one feature i would like to see added in future versions is IVTC for NTSC sources or the abilty to use avisynth or vdub filters to perform varios operations such as ivtc.

SeeMoreDigital

7th July 2004, 17:42

With regard to Recode2 6Ch audio encoding does anybody know if it's VBR?

Personally I use Foobar2000 to generate Nero AAC-HE VBR streams directly from the DVD's AC3 stream.

Originally posted by chilledoutuk
Does anyone know of mp4 apps that would alow me to extract the video from a .MP4 of this size? GraphEdit perhaps?

Cheers

bond

7th July 2004, 18:26

Originally posted by hyper84
I can confirm the problem with the video being twice as slow as the audio when trying to edit an existing ND .mp4 file. You would think you could edit a nero file with nero filters, eh?:D

BTW, I did used these filters:
Nero parser to split the ND file.
Nero Muxer
Nero file writer thats not good :/

SeeMoreDigital

7th July 2004, 18:58

Hi bond,

Given that NeroDigital's filters have changed so much in GraphEdit, it might be helpful to have such information in it's own thread...

How about calling it: -

Updated - Nero Digital Filters with GraphEdit

And just when I was starting to get the hang of things ;)

Cheers

hyper84

7th July 2004, 19:35

Does anyone know of a good method for creating a 2CD encode with nero digital and/or splitting an existing 1400 MB file (xvid or ND)? I can't find a way to split a file while encoding on the recode GUI and the nero/nero problem won't let me split the file later using graphedit. And while I'm asking I'll go ahead and throw this out there :rolleyes: . Is there a way in graphedit to bring in two separate files, say, to merge audio only and video only files?

Jvdhorst

8th July 2004, 00:40

Originally posted by SeeMoreDigital

What players have you tried?

Meanwhile I've tried also a third party decoder plugin for winamp to listen to the audio stream and the soundproblem is still there. I've also tried CoreAAC decoder with graphedt and this also results in the same problem. So I guess it's a nero encoder bug isn't it? I haven't tried another encoder yet, since I'm new to aac encoding and my Besweet needs fixing since it crashes at launch :mad:

SeeMoreDigital

8th July 2004, 08:50

I do seem to remember one of the versions of Recode2 being quite bad at 'audio wise' especially when used with other encoding applications!

If you are experiencing clipping Foobar2000 has a limiter you can apply as well as other useful encoding tools.

Cheers

SeeMoreDigital

9th July 2004, 18:42

Well seeing as though this new version of NVE still isn't perfect 'DS filter wise'. I've re-installed the 3ivX codec and now all is well again via all my software media players!

You've got to hand it to 3ivX ;)

Cheers

bond

11th July 2004, 15:21

ok i now found the time to do another comparison, this time with settings more matching each other (matrix1 clip (~7100 frames), as doom9 uses):

XviD 1.0.1: h263, VHQ4 (RDO, Trellis), QPel, GMC, B-VOPS (1), AQ, ChromaMotion
ND 4.1.6.2: h263, Extra (RDO, Trellis), QPel, GMC, B-VOPS (1), AQ (psy-high), 4MV, MV-Range 64

i measured the following encoding speeds:
XviD:
1st pass: 7:45 -> 15.3 fps
2nd Pass: 27:14 -> 4.36 fps
overall: 34:59 -> 6.78 fps

ND:
1st pass: 19:00 -> 6.24 fps
2nd pass: 18:45 -> 6.33 fps
overall: 37:45 -> 6.28 fps

so ND (promoted as the world's fastest MPEG-4 Encoder) was overall 0.5 fps slower than XviD :D

qualitywise the results didnt varry that much from my prior results with h263 quants (with custom quants (hvs-best) ND produced a totally dirty picture):
- ND provides overall a similar amount of details as XviD with the used testing environment (sometimes more, sometimes less)
- ND shows more artifacts than XviD, making the walls a little less "nice" and the picture appear overall little more dirty
- ND shows a less stable background/more moving pixels than XviD
- ND smooths too much in High-Motion scenes
- ND doesnt really like this very detailed walls in the famous "lobby shooting" scene as it smooths the walls much more than XviD there
- ND seems to produce a slightly brighter image than XviD

all in all i would say that the latest release is the typical "more quality vs. less speed" tradeoff, making ND not "the fastest encoder on earth" anymore, but definitely usable now :cool:
more comments and own tests welcome :)

bobololo

11th July 2004, 16:39

Originally posted by bond
all in all i would say that the latest release is the typical "more quality vs. less speed" tradeoff, making ND not "the fastest encoder on earth" anymore, but definitely usable now :cool:
more comments and own tests welcome :)

Okay, this is your point of view :)

To be in the very same conditions, you should have encoded the first pass un best quality rather than extra.

Anyway, actually you have to consider this version of Recode like an interim version. It wasn't intended to be released in this way especially with regard to the extra quality and the encoding performance.

And also, I checked the aspect ratio issues and I can confirm there are some slight errors in the PAR computing based on the experiments I did. I've reported back to Ahead staff and hopefully this should be definitely solved in the next release for the great pleasure of SMD and every body else ;)

-- bobololo

SeeMoreDigital

11th July 2004, 17:14

At least ND's B-VOP implementation works in hardware which is more that can be said for XviD's ;)

And I thank you bobololo for confirming the PAR situation :cool:

Cheers

Jvdhorst

11th July 2004, 18:14

I must say I like Nero dgital a lot with it's support for subtitles chapters etc. The slightly less picture quality doesn't matter to me if all these features are in the resulting mp4 file.

However I agree on the fact that it deserves no more then the status interim. To summarize on what already has been said:
Audio LFE encoding doesn't work well for 5.1 ac3 source. Directshow encoder filters don't work with encoding via graphedt. Decoding can work as long as only the basic all nero filters are loaded for parsing and video. Coreaac decoder filter needs to be used to decode the audio. On my system mp4 playback with nero filters is a bit stuttery on some media players (including showtime), while using ffdshow is completely fluend. I unregistered the ndparser.ax (with radlight filtermanager) to use 3ivx to parse mp4 files so ffdshow can decode audio and video (there go the chapters and subtitles :( ).

@SMD I managed like before to remux an Xvid bframe avi to mp4 using 3ivx muxer. Then playback is fluent now on Xcard.

SeeMoreDigital

11th July 2004, 18:35

Originally posted by Jvdhorst
@SMD I managed like before to remux an Xvid bframe avi to mp4 using 3ivx muxer. Then playback is fluent now on Xcard. Was this with v1.0.1?

I've just tried re-muxing some short non-packed BS containing 1B-VOP and 2B-VOP and can't get any joy. What settings did you use?

Cheers

Jvdhorst

11th July 2004, 23:57

is off topic. I'll pm you. Posts can be striked.

hyper84

12th July 2004, 02:15

Personally, I still believe that ND is the fastest, at least for me, because when I use xvid(edit:with GK), I have to create a d2v, encode the audio (a two pass process), encode the video, then merge them into an mp4 with mp4box. With ND, it's just a few clicks and it's all done then and there. I can't speak for AutoGK, though. Maybe it makes encoding w/ xvid "faster". Still, I only encode xvid as of now for the extra quality ;) .

hyper84

12th July 2004, 05:30

Originally posted by pogo stick
The problem with remuxing (to create chapters) audio and video with Nero parser, muxer and file writer stays in new version. Video is twice slower then audio. :(
Does remuxing existing mp4 (audio+video) with Nero direct show stuff ever worked in Graphedit right? Did anyone try it?

Hmmm, I tried it again, and it worked! At least, the video is not twice as slow. I did not do anything but split an xvid .mp4 w/ aac audio, though. No creating chapters or anything. Also the SECOND half of the split video seems to be A) out of a/v sync (sometimes) and B)the beginning of the film shows "garbage" (sometimes). Garbage in this case is that the picture either looks grayed out at the beginning (first second or so, not much, as if it corrects itself) or you can see some of the video moving and the rest isn't and it's all blocky (hard to explain :rolleyes: ). This also corrects after a second or so. I have tried with xvid + lcaac and ND + he-aac. Earlier, when I found the video running half as fast I believe that it was xvid + he-aac. All were properly created (trust me, hehe). However, the problem may have been in the pontentially faulty cd that I ripped to my hard drive that contained the mp4. I just don't know! I didn't change anything else. My conclusion is that either I crazy with a screwed system (likely) or the muxer is too buggy to use reliably (equally likely) and that hopefully will be bugfixed for the next nero release.

bond

12th July 2004, 17:32

Originally posted by hyper84
Personally, I still believe that ND is the fastest, at least for me, because when I use xvid(edit:with GK), I have to create a d2v, encode the audio (a two pass process), encode the video, then merge them into an mp4 with mp4box.well ok of course if you take all the stuff to do before encoding nd is surely faster as its a one click solution
still when encoding with the same settings nd was slower than xvid (well ok, 0.5 frames per second is not but much still... ;) )

Originally posted by hyper84
Also the SECOND half of the split video seems to be A) out of a/v sync (sometimes)hm, seems that the splitting function doesnt work 100% correct?

and B)the beginning of the film shows "garbage" (sometimes). Garbage in this case is that the picture either looks grayed out at the beginning (first second or so, not much, as if it corrects itself) or you can see some of the video moving and the rest isn't and it's all blocky (hard to explain :rolleyes: ). This also corrects after a second or so.hm normally thats the typical situation when the keyframe isnt there at the beginning, but
1) when i tried it the muxer splitted exactly at a keyframe
2) .mp4 offers an option to not cut at keyframes, called preroll if i remember it right, meaning being able to cut at any frame you want. still it seems that the nero filters doesnt support it

My conclusion is that either I crazy with a screwed system (likely) or the muxer is too buggy to use reliably (equally likely) and that hopefully will be bugfixed for the next nero release.yeah i hope that ahead will think about investing some times to make the using of their filters possible even outside of recode2, cause atm they are only looking for making them working together with recode

chilledoutuk

12th July 2004, 23:18

Bond one question did you encode with recode without encoding the audio stream?

If you did encode the audio then this would obviously reduce the speed of the encode.

One thing have noticed is recode crashing when I was trying to encode space I think it might have something to do with the decoding of the mpeg-2 stream as changing the settings for encoding the video did not stop the crashes and they always ocured in the same places.

It would be nice to be able to acept other video sources especially .avs files.

hyper84

13th July 2004, 00:46

Originally posted by chilledoutuk

It would be nice to be able to acept other video sources especially .avs files.

I agree. It would be REALLY COOL for Recode to accept .avs files, and also other codecs like xvid :rolleyes: . But seriously, avs input support would ROCK.

I also agree with bond about the moving pixels, at least on blue backgrounds. Bond has the same eyes as me :p . On Ron White's "They call me Tater Salad", Ron stands in front of a blue-lit wall during his show and the wall moves around so much it is distracting. Xvid looks much much more stable. Other than that "problem sample," to me it seems that ND has become quite the competitor to mine eyes.

SeeMoreDigital

13th July 2004, 09:38

Originally posted by hyper84
I agree. It would be REALLY COOL for Recode to accept .avs files, and also other codecs like xvid :rolleyes: . But seriously, avs input support would ROCK. I can't seeing Recode2 offering this level of functionality as it would make DVD copying far too easy.... but we live in hope :)

Cheers

JohnV

13th July 2004, 14:05

I've heard rumors from a pretty reliable source that Recode 2.1 will include AVS input. Of course nothing is certain until it's certain but we will see hopefully soon.

About the multichannel clipping some people experienced with ROTK: I talked to Ivan, and seems it has now been fixed and will be implemented in the next update. Problem was not in the AAC encoder though (if you encode the ac3 multichannel audio with ND audio encoder, there's no clipping).

bond

13th July 2004, 16:58

Originally posted by chilledoutuk
Bond one question did you encode with recode without encoding the audio stream?
If you did encode the audio then this would obviously reduce the speed of the encode.why should i encode the audio if i want to compare the encoding speeds of two video codecs

Originally posted by JohnV
I've heard rumors from a pretty reliable source that Recode 2.1 will include AVS input. Of course nothing is certain until it's certain but we will see hopefully soon.hm from the rumours i heard it should have been already here :D

SeeMoreDigital

13th July 2004, 17:26

One thing is very clear. You can't knock their enthusiasm and their willingness to improve Recode2!

Cheers

JohnV

13th July 2004, 17:53

Originally posted by bond
hm from the rumours i heard it should have been already here :D I have more reliable sources than you do then. :D :D :p
Anyway, if everything goes ok, avs support will be there very soon.

JohnV

13th July 2004, 20:41

Here's a picture from Recode 2.1 beta showing avs input in action:
http://sivut.koti.soon.fi/julaak/Recode21_avs.png

SeeMoreDigital

13th July 2004, 20:56

Nice one JohnV,

So apart from AVS it looks like we'll also be getting chapter marking/making and 2 track audio input.

Any other lovelies?

Cheers

hyper84

13th July 2004, 20:59

Originally posted by JohnV
Here's a picture from Recode 2.1 beta showing avs input in action:
http://sivut.koti.soon.fi/julaak/Recode21_avs.png
:D

chilledoutuk

14th July 2004, 23:26

I suppose we can use avisynth to perform ivtc but i still think that for the majority of ntsc dvds IVTC is needed and those that dont know how to use Avisynth would need to have some form of this built into the application.

SeeMoreDigital

17th July 2004, 14:12

I've been testing using mp4UI and GraphEdit to de-mux the AAC audio stream created by Recode2 and replace it with an Mp3 stream.

But for some reason when I try to play the files using Nero's ShowTime player I can't hear the audio. And the audio stream tag (ie: MP3) is not flagged in the players GUI.

Is it just me or is it a bug... Mpeg4/Mp3 in .MP4 used to work?

Cheers

JohnV

18th July 2004, 09:27

Originally posted by SeeMoreDigital
Nice one JohnV,

So apart from AVS it looks like we'll also be getting chapter marking/making and 2 track audio input.

Any other lovelies?

Cheers
Well, according to bobololo some features may be still changing (for example there could be h264 decoding support in Showtime etc.), but one nice thing which is certainly there is the new very fast 1st pass analysis in 2-pass ND encoding which is available for all modes including "extra high". It also supports at least .avi and .mpg inputs.

LostMP4

18th July 2004, 10:47

Originally posted by JohnV
Well, according to bobololo some features may be still changing (for example there could be h264 decoding support in Showtime etc.), but one nice thing which is certainly there is the new very fast 1st pass analysis in 2-pass ND encoding which is available for all modes including "extra high". It also supports at least .avi and .mpg inputs.
Was he talking about the next August update or an undetermined new version?

bobololo

18th July 2004, 14:15

Originally posted by chilledoutuk
I suppose we can use avisynth to perform ivtc but i still think that for the majority of ntsc dvds IVTC is needed and those that dont know how to use Avisynth would need to have some form of this built into the application.

You should try to encode a NTSC DVD using Recode. If I'm correct, Recode should automatically inverse telecine any NTSC sources. I'm not 100% sure of the details but I guess it's easier for Recode since he can process the input mpeg-2 source as FILM.

-- bobololo

JohnV

18th July 2004, 20:42

Originally posted by LostMP4
Was he talking about the next August update or an undetermined new version? I mean Recode 2.1 which is due "soon".

hyper84

19th July 2004, 03:56

Well I guess "soon" is better than "when its done" ;) . Anyway, it sounds like Nero is really working hard to turn Recode into THE mp4 app (and it sounds like not just for dvd's anymore!). I for one appreciate very much all of Nero team's hard work. I was at first disappointed in Recode, but now I'm really looking forward to 2.1, not to mention h.264 decoding and eventually the h.264 codec! Will that codec be available as a free update or what? Or is it to early to tell? I vote free update :rolleyes: .

chilledoutuk

19th July 2004, 05:19

@ bobololo

yes you are correct on the majority of NTSC DVDs recode does correctly IVTC but there are some sources that it mistakes as just being interlaced when in fact I have determined that it requires IVTC.
For these I suppose the avs input could be used.

One thing that would be useful is a small app for editing the mp4 files i.e. adding new tracks replacing faulty ones etc. Although i have been using mp4ui to do this it is very buggy and cannot open some of the .mp4's that i created with recode due to it being over a certain size.

When recode was first released i saw the potential for the application it's nice to see that potential being filled.
I my opinion recode is the best all in one MPEG-4 encoding application I have ever used kicking the crap out of apps like DrDivx on terms of intuitiveness of the interface and features HE-AAC.

Keep up the good work ill look forward to the next release of this exceptional application.