Wmv9

[YY1020]

I have a question!!
In fact,XviD...DivX.....WMV9.........three codec,which one is good??
I feel that,WMV9--->DivX----->XviD(Good----->Bad)
Let's to discuss!!

[DAvenger]

Hard to say :rolleyes: Each has its pros and cons .... but hell I am not an expert (moreover I didn't get permission from SeeMoreDigital to post here ;) )

Anyway, do not forget RV9 which seems to give pretty good results at low bitrates ...

[SeeMoreDigital]

Well YY1020,

this kind of topic has been discussed many times in this section of the forum. So there should be many threads for you to find!

I think it's fair to say that the quality produce by any given codec encode will also depend on the bitrate speed you use.

What I mean to say is that many people prefer the look of WMV9 for encodes of 600-950kbps and DivX or Xvid for encodes of 950kbps and above.

There is also much debate on how a codec performs when resizing and cropping the original source file.

Now, shock horror. You did not mention anything about RealMedia Video 9. Which will upset many people here, who monitor this section of the forum!

RMV9 has come along in leaps and bounds during the last month or so!

Maybe it is time to carry out some new comparison tests. Myself, well, I quite like some of the new DivX 'test' codecs!

[SeeMoreDigital]

Dear RadLight boss,

Please feel free to post away Sir!

[DAvenger]

Originally posted by SeeMoreDigital
Please feel free to post away Sir!

Thanks :D You are not so bad after all :p

[Gaia]

Totally useless tread. Just read Doom9's Codec comparisons. Infact most people think totally opposed. 1. XviD 2. DivX 3. WMV 9.

This just turns in to stupid flame war.

[SeeMoreDigital]

Oh dear Gala, that's a bit of a sweeping statement!

I did'nt get that overall impression with Doom9's codec comparision. But maybe the post has been updated since I last read it

Anyway, all the manufacturers codecs are, and will continue to improve, so it really is hard to say.

If you read some of Sirbers posts fore-instance he swears that RMV9 is the 'dogs doo-dars' for animes at low bitrates. But I'll take his word for it, as the closest I've got to an anime is encoding ToyStory 2.

I personally like to generate low bitrate (below 900kbps), full anamorphic encodes and in my opinion (and the opinion of several of my co-workers is that) WMV9 is better than the rest.

But each to there own!

[Nic]

Gaia is right. Forum rule 12:-
Do not ask "what's best" because this question cannot be answered objectively.

First sign of flaming or unconstructive comments this thread will most likely be closed.

-Nic

[SeeMoreDigital]

Yes, agreed Nic.

If YY1020 is interested Doom9's tests are located here: -

http://www.doom9.org/

Just go to the 'search' facility and type in 'codec comp'

Cheers

[karl_lillevold]

Originally posted by Gaia
Just read Doom9's Codec comparisons.
A very detailed and thorough comparison, indeed, but I would like to make sure everyone knows that it did use the 1.5 year old RV9, because RV9-EHQ (http://forum.doom9.org/showthread.php?s=&threadid=55193) was not available at that time. The feedback in the EHQ thread indicates EHQ significantly improves RV9 quality, and should now be the new recommended default. Also see this PSNR comparison (http://forum.doom9.org/showthread.php?s=&threadid=57687) for all the current codecs. I have yet to verify that the DirectShowSource method provides accurate PSNR numbers for RV9, but it should be pretty close, and for sure not higher than they really are.

[Sagittaire]

I also made a comparison of the codecs. Will see here if you want to know some more.

http://jfl1974.free.fr/HTM/32_Test_Codec_Video.htm

And I confirm that in term of PSNR the classification is as follows: RV9 EHQ, Kauehi, WMV9, XviD, DivX SBC.

And here my opinion on visual quality: RV9 EHQ, WMV9, Kauehi, XviD, DivX SBC.

I know perfectly well each one of these codec (XviD really well) and I used best the reglages possible for each one of them.

[bond]

Originally posted by Sagittaire
I also made a comparison of the codecs. Will see here if you want to know some more.
http://jfl1974.free.fr/HTM/32_Test_Codec_Video.htmhm isnt minority report a really grainy movie?

i dont know if it is a good idea to take this movie for a comparison as the codecs who dont smooth out the grain and try to keep most of the details will have problems of course (like xvid)...

[temporance]

Originally posted by bond
i dont know if it is a good idea to take this movie for a comparison as the codecs who dont smooth out the grain and try to keep most of the details will have problems of course (like xvid)... I agree somewhat. In a grainy movie, a codec that does pre-filtering (and we all suspect RM and WMV do this) will have the advantage in PSNR tests. But the prefilter may make some movies look worse -- a poorly designed prefilter will soften fine details and edges.

If a codec does not do prefiltering, then it should look sharper, unless the codec's ME is confused by the grain (and simple ME is very susceptible to just a little noise).

[Sagittaire]

The PNSR is based Harry Potter II Trailer. It's a very clean source. The visual test is based Minority Report ... :devil:

[bond]

i wasnt talking about the psnr results (althought i wonder why you used xvid only with 1 b-frame) as i dont really like psnr results as they dont show the same results for me as if i am comparing the frames visually (with avscompare) and only the visual way can give you the "real" quality of the picture imho...

and from looking at your samples i would say that wmv9 has less details than the other codecs which is the same result as i get from my tests (and that's why i dont judge wmv9 to be better than divx5 or xvid)
would be great to get some rv9 samples too (i really have to test rv9 ehq now :D )...

[SeeMoreDigital]

Well, what a very well laid out and presented report. I'm sure Karl Lillevold will be very pleased with the outcome of EHQ. I enjoyed reading it, many thanks.


But here's a thought! How about we all do some more tests but agree on encoding the same source movie first.

My suggestions would be...
Men in Black 1 (94mins PAL. Don't know NTSC runtime) for 1.85:1 encode test and
GoldMember (91mins PAL. Don't know NTSC runtime) for 2.35:1 encode test.

[temporance]

Originally posted by bond
i wasnt talking about the psnr results (althought i wonder why you used xvid only with 1 b-frame) as i dont really like psnr results as they dont show the same results for me as if i am comparing the frames visually (with avscompare) and only the visual way can give you the "real" quality of the picture imho...Agreed, I find that PSNR is good at comparing results from the same type of codec, e.g. comparing different settings of xvid and DivX which are both MPEG-4 codecs. But PSNR doen't always agree with the eye when comparing between codecs. There are more sophisticated objective quality measures out there, but they are difficult to get hold of. I found one piece of software that you actually have to fax a signed license agreement before you can dl the demo!and from looking at your samples i would say that wmv9 has less details than the other codecs which is the same result as i get from my testsYes, I get this impression in my assorted test encodes, and the same goes for RV9-EHQ. I'm finding that the proprietary codecs (RV9-ENQ and WMV9) have a more processed and artifical look than the MPEG-4 based codecs (DivX and xvid). Yes, they look cleaner and more "perfect" at lower bitrates, but they also look like they're computer generated or something weird. Could be the result of some sophisticated preprocessing. So it makes sense that RV9 should be popular for anime.

[Sagittaire]

@ All XviD Fan ... ;-)

You say that the PSNR is not representatif quality of the codecs but it is a tool which the developers uses (Koepi also).

It is the only objective manner to compare the codecs. And the values in favour of the RV9 are indisputable especially in bitrate or the compressibilite is weak.

Finally visually I also find that RV9 EHQ is largely better. It is a personal opinion and I do not oblige you to have the same one ... ;-)


@ bond

Source Bugsan unite-video


xvid koepi 24-06-03, 2pass: 20000ko (650kbps)
avisynth 2.52
Mpeg2Dec3 1.08
FFdshow 030103
Athlon XP 2000+
---avs------avs------avs------avs------avs------avs---
mpeg2source("E:\Ripping\lotrcutted\lotr.d2v",idct=7)
Crop(4,80,-4,-80)
BicubicResize(640,256,0,0.5)
---avs------avs------avs------avs------avs------avs---
clip1 = mpeg2source("E:\Ripping\lotrcutted\lotr.d2v",idct=7)
.Crop(4,80,-4,-80).BicubicResize(640,256,0,0.5).ConvertToYUY2()
clip2 = directshowsource("xvid_XX.avi",fps=25).ConvertToYUY2()
Compare(clip1,clip2,"YUV","psnr.log")
---avs------avs------avs------avs------avs------avs---
info:
altcc-best = "altcc-h-150-100-50"
exotic = "mpeg vhq4 bf1 cm cmop altcc-best"
pp4 = "ffdshow nic pp4 strength 256"

|---------|---------|
| K140503 | K240603 |
|-----------------------|---------|---------|
| default | 41.9028 | 41.8955 |
| default pp4 simple | 42.2854 | 42.1851 |
| default pp4 walken | - | 42.2744 |
| default pp0 walken | - | 41.8940 |
|-----------------------|---------|---------|
| chroma motion | 41.9698 | 41.9864 |
| global motion comp | 41.8826 | 41.8799 |
| quarter pixel | 41.5587 | 41.5599 |
| lumi masking | 41.8397 | 41.8423 |
| chroma optimiser | 41.8964 | 41.9019 |
| trellis R-D 0 | 41.8684 | 41.8855 |
| mpeg quantization | 42.0297 | 42.0356 |
|-----------------------|---------|---------|
| VHQ 0 bf -1 | 41.9028 | 41.8955 |
| VHQ 1 bf -1 | 42.0766 | 42.0887 |
| VHQ 2 bf -1 | 42.1032 | 42.1782 |
| VHQ 3 bf -1 | 42.1491 | 42.2292 |
| VHQ 4 bf -1 | 42.2199 | 42.3316 |
| | | |
| VHQ 0 bf 0 | | 41.8946 |
| VHQ 1 bf 0 | | 42.1150 |
| VHQ 2 bf 0 | | 42.2359 |
| VHQ 3 bf 0 | | 42.2522 |
| VHQ 4 bf 0 | | 42.3327 |
| | | |
| VHQ 0 bf 1 | | 42.0270 |
| VHQ 1 bf 1 | | 42.2614 |
| VHQ 2 bf 1 | | 42.3151 |
| VHQ 3 bf 1 | | 42.3338 |
| VHQ 4 bf 1 | | 42.4166 |
|-----------------------|---------|---------|
| bf0 default | 41.9028 | 41.8946 |
| bf1 default | 41.8666 | 42.0270 |
| bf2 default | 41.7618 | 41.9992 |
| bf3 default | 41.7480 | 42.0043 |
| bf4 default | 41.7471 | 41.9937 |
|-----------------------|---------|---------|
| altcc-default | 42.1407 | 42.1330 |
| altcc-best | 42.3478 | 42.3417 |
|-----------------------|---------|---------|
| exotic | 42.6715 | 42.9207 |
| exotic pp4 simple | 43.0756 | 43.2172 |
| exotic pp4 walken | - | 43.2652 |
|-----------------------|---------|---------|

Koepi 240603 bframe test:

|---------|---------|---------|---------|---------|
| 1 - 150 | 1 - 190 | 2 - 150 | 2 - 190 | 3 - 150 |
|---------------|---------|---------|---------|---------|---------|
| bf thresh -40 | 41.9154 | 41.9148 | 41.9160 | | |
| bf thresh -20 | 41.9632 | 41.9542 | 41.9632 | | |
| bf thresh 0 | 42.0270 | 42.0494 | 41.9992 | | 42.0043 |
| bf thresh 10 | 42.0545 | 42.0750 | 42.0118 | | |
| bf thresh 20 | 42.0500 | 42.0745 | 41.9598 | | |
| bf thresh 30 | 42.0407 | 42.0689 | 41.9061 | | |
| bf thresh 40 | 42.0286 | 42.0655 | 41.8842 | | |
| bf thresh 60 | 42.0125 | 42.0318 | 41.8165 | | |
| bf thresh 80 | 42.0007 | 42.0225 | 41.7905 | | |
| bf thresh 90 | 41.9973 | 42.0103 | 41.7871 | | |
|---------------|---------|---------|---------|---------|---------|
| bf offset 0 | 41.9844 | | | | |
| bf offset 25 | 41.9887 | | | | |
| bf offset 50 | 42.0181 | | | | |
| bf offset 75 | 42.0270 | 42.0494 | 41.9992 | | 42.0043 |
| bf offset 100 | 42.0115 | | | | |
| bf offset 125 | 42.0104 | | | | |
|---------------|---------|---------|---------|---------|---------|


The defaut reglage for bframe 1/150/75/0 is very good reglage for the bframe for Koepi 24/06/03 and in conclusion for PSNR ... ;-)

[bond]

Originally posted by Sagittaire
You say that the PSNR is not representatif quality of the codecs but it is a tool which the developers uses (Koepi also).yes, they use it but they also write that psnr results arent without problems

and you also wrote in another thread:
the objective in a test PSNR is to obtain the best possible PSNR. But it is true for a better visual quality DivX Labs recommend bframe + Slowest + BM 0.25 + New Psy + 3 Pass. but the PSNR decrease a little.-> better quality with lower psnr ;)

The defaut reglage for bframe 1/150/75/0 is very good reglage for the bframe for Koepi 24/06/03 and in conclusion for PSNRwhat's reglage? if you mean the default settings than they are -1/150/75/0 ;)
most people use 2 b-frames in their encodings in this forum (they always claim divx5 to support only 1 b-frame also :D ) most people also used 3 b-frames some time ago frequently


but let's stop that :)
your test is great and your screenshots show that wmv9 produces less details than divx5 or xvid (which is the same result i get)
i only think that because of that the conclusion that wmv9 is superior than divx5 or xvid shouldnt be drawed
(and minority report is definitely a hard to encode movie for high detail codecs)

[Sagittaire]

XviD is better for the detail than DivX 5.05 and the WMV9 but less good for the blocking and the ringing ...

DivX Kauehi is really better than DivX 5.05, near to XviD for the detail, near to the WMV9 for the blocking and the ringing ...

Next DivX 6.0 will undoubtedly be one of the best codec of the moment ... ;-)

To finish I also find that XviD is an excellent codec: it is thanks to him which the others progress has this speed ...

thus long life in XviD ... :D

[temporance]

@Sagittaire:

Something interesting to try with WMV9... try measuring PSNR (or some other objective measure) at several of its constant quality settings (say 100%, 95%, 90%, 85%....). Also record the filesize for each of these settings.

Now plot a graph of PSNR against the log of filesize (use a spreadsheet).

You see something interesting. As you decrease the quality setting (which has to be like increasing quantizer in MPEG-4), the PSNR doesn't decrease in a straight line. It does decrease but there are "steps", with something kicking in at around 75% and 55% to boost the PSNR measurement.

Anyone got any ideas what this could be? Do we still have our resident Microsoft engineer on the forum?

My theory is that it actually downsamples the image as the encoding gets harder. So, with an input of 640x480, t would only encode at this res for at high quality settings. As the bitrate gets tighter, it might reduce the resolution internally to, say, 320x240. But the clue is in the graph I described... I'll try to make an image and post it here.

Edit: I couldn't replicate my results this time - perhaps I used higher complexity settings before. Damn, don't want to waste all day on this but it was an interesting graph when I had it before.

I could prove my theory for sure if I could make a spectrum analysis of the decoded image -- you would see a "cliff edge" if the theory is true.

Theory or not, is this approach good for incorporation into xvid etc. ?

[bond]

very interesting, a image would be great!

Originally posted by temporance
Theory or not, is this approach good for incorporation into xvid etc.hm,
in xvid you have to do everything for yourself (choosing vhq, qpel, gmc, cm...), i think in wmv (made for non-pros) the codec automatically chooses the settings internally which are set by the developers to produce a better output

for example think about xvid automatically unticking qpel or choosing higher b-frame settings if you are encoding at real low bitrates (without even telling you that these features even exist, like it can be possible in wmv9)...

[karl_lillevold]

Originally posted by temporance
I agree somewhat. In a grainy movie, a codec that does pre-filtering (and we all suspect RM and WMV do this) will have the advantage in PSNR tests. But the prefilter may make some movies look worse -- a poorly designed prefilter will soften fine details and edges.
Sorry, temporance, but this has been in the RV9 Info Thread (http://forum.doom9.org/showthread.php?s=&threadid=40392) since its beginning

To squash some rumors right away, RV9 is not based on wavelets, and not on fractals, and does not include inherent pre-filtering.
Those who have read this should no longer suspect that RV9 does pre-filtering :p

A codec that does pre-filtering will come out lower in PSNR tests, i.e. it is actually a dis-advantage. This is because the PSNR is measured against the original video, and pre-filtering changes the video, and thus lowers the PSNR. If, on the other hand, PSNR is measured after pre-filtering for both compressed and original, you are right about a PSNR advantage, but that would not have been the case in Sagittaire's PSNR test, if RV9 had been doing any pre-filtering.

[temporance]

Apologies, karl, for my error and (even worse ;) ), lumping RV9 and WMV in the same statement!

Edit: I'm not sure that your assertation "preprocessing decreases PSNR" is always true. I have seen preprocessing increase PSNR, on a very old MPEG-2 codec where noise would cause the motion estimation to choose uncorrelated vectors. So although a noise reduction filter actually adds artifacts that will be measured by PSNR, this is more than compensated by the compression codec performing properly on a clean source image. I haven't tried any of this with modern codecs -- I suspect preprocessing will make a subjective improvement, but produce a similar or worse PSNR.

[Sagittaire]

I think that it is not worth while to seek. The PSNR indicates convergence, pixel by pixel between the source and the video. If RV9 EHQ has the best PSNR that means quite simply than it is the encoding which approaches the most source. It is also simlpe which that. RV9 EHQ is thus the best current codec.

The visual aspect is another problem. For example the WMV9 has a PSNR a little weaker than DivX Kauehi but I find that visually the WMV9 is more beautiful.

[temporance]

Originally posted by Sagittaire
I think that it is not worth while to seek. The PSNR indicates convergence, pixel by pixel between the source and the video. If RV9 EHQ has the best PSNR that means quite simply than it is the encoding which approaches the most source. It is also simlpe which that. RV9 EHQ is thus the best current codec.

The visual aspect is another problem. For example the WMV9 has a PSNR a little weaker than DivX Kauehi but I find that visually the WMV9 is more beautiful. You just contradicted yourself, Sagittaire ;) You say it's not worth criticising PSNR, but then admit that your PSNR results don't agree with what is seen by the human eye!

There's a whole industry trying to come up with something better than PSNR. What about us? We just need to understand its weaknesses. For example, try to use the Overall PSNR (now in Avisynth CVS) - it's more reliable than the average PSNR. A codec could quite easily trick average PSNR to get a higher result, just by messing with rate control.

[Sagittaire]

Yes but the difference between Kauehi and WMV9 is 0.1 ~ 0.2 dB for the PSNR thus the discution is possible ...

When adventage PSNR is 1.0 dB the discution is not possible ... the RV9 EHQ is the best in the low bitrate ...

[Sigmatador]

Shift the frame by one pel, without compression, you'll get an ugly PSNR and a very nearly perfect image ;p

PSNR is a very quick and quite good tools for the dev to measure and compare some feature or code, but not to compare codec.
look this (PSNR=10log(255²/MSE)):
http://www.cns.nyu.edu/~zwang/files/research/quality_index/demo_lena.html

no comment is needed ^^



anyway, i assume that with a low bitrate, Rv9 is the best codec. (yet ^^ ;p)

[SeeMoreDigital]

Can someone explain this,

I recently did some 'video only 2.35:1' 2pass VBR 720x576 anamorphic encodes with DivX 5.0.5 and DivX Kaushi at 627kbps and Kaushi looked better.

I then encoded the same 'video only' file with 2pass VBR WMV9 VCM at 615kbps (these bitrates were used so I could achieve the same file sizes) And WMV9 looked better than DivX.

When you do the tests again using a higher bitrate of 963kbps for DivX and 945kbps for WMV9 (Again these bitrates were used so I could achieve the same file sizes). DivX looked better!

But here's the thing. When you perform the same higher bitrate encode tests again but this time crop and resize the source image, DivX drops off and WMV9 comes back. So the finished results look about the same!

It's just a thought but maybe DivX's Bilinear and Bicubic resizers require some tweaking. Could this be one of the revisions with DivX6.

Anyway, have a gander at one of Darrius "Junto" Thompson's (Director R&D DivX) posts about DivX.

Yes, it seems everytime we think we've reached the limits of improving video quality within the boundaries of MP4 someone on the team comes up with a great idea that has the potential for making improvements that are worth our efforts. At the same time we know there are limits and we constantly think we are close and then someone on the team proves this thought wrong However, we know that other algorithms can be developed that would enable us to take an even larger leap than "might" be available within MP4 whether the idea\algorithm is proprietary or based off of newer standards and we are working on these everyday, so at no time are we confining ourselves just to MP4, we have to see what will give us the best jump in quality along with performance and keep in mind what the average configuration of a video PC user is and then also correlate that with hardware. To make a long story short, you can expect that we are looking at many new ideas not within MP4 as our goal is not to be MP4 but to be DivX...high quality\useability across software and hardware. Also I'm not sure I've ever mentioned this publically but our goal was never to be MP4 our goal was always quality\performance\scalability. We then looked in a toolbox to see what we thought were the best video tools to build on top of and it happened that the tools we believed would give us the best video quality also happened to be part of MP4 and this is how the MP4 route was chosen by DivX. It was not....let's be MP4. It was, "How do we create quality".

Interesting eh!

Cheers

[Sagittaire]

Once again the PSNR is a tool which is used by the developers of codec to evaluate their codec ...

http://labs.divx.com/archives/000007.html

http://labs.divx.com/archives/data/TahaneaQualityReport.zip

I have just made the encoding of the film harry Potter II in entirety in RV9 EHQ, XviD, WMV9, Kauehi in 576*240 with 570 Kbps and that is visually or with the PSNR there was no possible comparison: RV9 EHQ was largely the best ... then the WMV9, Kauehi and finally XviD ...

Make the test ... ;-)