[Test Codec] H264, VP6, RV10, WMV9, DivX3, XviD, DivX5 .....

[Sagittaire]

Matrix II 720*304

Source DVD PAL 720*576 MPEG2 MP@ML, chapter 20 to 26, ~35 000 frames

Test in french (http://jfl1974.free.fr/HTM/32_Test_Codec_Video.htm) ... but Capture, Sample & Graphs are universal language

Codecs
- H264 Ateme
- XviD 1.0.1
- RV10 Elysean
- VP6 6.2
- DivX 5.2.1
- WMV9 VCM
- DivX 3 ffvfw


Tests
- 500-2000 Kbps and 25-100% compressibility (/quant2 XviD H263)
- SSIM, Average PSNR, Overall PSNR and Frames PSNR test
- Blind test
- Samples
- Diaporama

[bilu]

What's the criteria difference between the two images?

One is called PSNR and the other PSNR Overall... :confused:

EDIT: does that trailer have high-motion parts?

Bilu

[Sirber]

Nice comparison :D

[DAvenger]

Ah, could you please explain the results to us uneducated plebs? :p

Thanks

[TheXung]

Could you use more differing colors for xvid and vp6 or at least make the lines thicker than a pixel.

[Assault]

@ Sagittaire

Did you use XviD with GMC and Trellis? And for your b-frame settings I would recommend 1/150/100. I think this should deliver a little better results but altogether it's a very nice comparison. Thanks :)

@ all

When you look at the two graphs it's very interesting to see that VP6 is a good codec but has a bad 2-pass algorithm. Its results are much better for average psnr than for overall psnr. I think that's because average psnr doesn't take into account how constant the quality is. So I suspect that VP6 has both frames with VERY good quality and frames with VERY bad quality. That should be the reason why the results for overall psnr arent really good. If On2 improves the 2-pass mode VP6 will be a very good codec in my opinion.

Assault

[CruNcher]

@Sagittaire

please post more infos filesize and time needed to encode

[SeeMoreDigital]

Originally posted by Assault
When you look at the two graphs it's very interesting to see that VP6 is a good codec but has a bad 2-pass algorithm. Its results are much better for average psnr than for overall psnr. I think that's because average psnr doesn't take into account how constant the quality is. So I suspect that VP6 has both frames with VERY good quality and frames with VERY bad quality. That should be the reason why the results for overall psnr arent really good. If On2 improves the 2-pass mode VP6 will be a very good codec in my opinion.

Assault It's interesting you should mention that!

I've been trying to generate 2pass VBR encodes with VP6 (using MPEGmediator) but on every occasion the encode looks either the same or only very slightly better than a 1pass CBR VP6 encode.

Personally, I've been thinking that I may be doing something wrong when generating 2pass encodes. But I don't know for sure!

I really would like to get to the bottom of this. Because if VP6 is able to look as good as it does in CBR mode, it should look astonishing in VBR mode. I just wish I could get it right!

Cheers

EDIT: Anyway, what's with the DivX 5.1.2 ident. I thought the latest is DivX 5.1.1 beta2.... Or am I wrong?

[Assault]

@ SeeMoreDigital

I don't think that you're doing something wrong when creating 2pass encodes. I wasn't content with the results I got in my tests either. I also remember C0mPr355 saying that VP6 is their first codec with a pass mode. So perhaps that's the reason for the bad results. :(
Concerning DivX I think you're right. It's probably a typo.

Assault

[SeeMoreDigital]

Thanks for that Assault.

I thought I was begining to flip out there for a second - twice!

Yesterday I did some more low bitrate CBR tests with VP6 using 720x576 image pixel frame sizes. It really is very good!

However, I personally would like to be able to enter kb's instead of KB's per second. And maybe the ability to store the stream in an MP4 container instead of (or as well as) on AVI.

I realize that this codec was originally concieved for use with cellular phone networks. So with that in mind I would have thought being able to shove the video stream in an MP4 container would have been more practical!

Or am I really flipping out now???

Cheers

[Assault]

@ SeeMoreDigital

AFAIK VP6 is no mpeg4 codec. I don't think that it's possible to put non mpeg4 compliant codecs in an MP4 container. And even if it were possible it wouldn't be a good idea in my opinion because the resulting MP4 wouldn't be spec compliant.

Assault

[CruNcher]

@ SeeMoreDigital
i think Christian is allready lurking arround and allready talks with the other Matroska guys howto support it in MKV aswell hehe im sure about that why ??? because i know him :D

[Sagittaire]

Update: VP6 setting and file size

[vinouz]

So please could someone explain (Sagittaire ?) what the difference between PSNR and PSNROverall is ? Do that mean that on PSNR you take the 1324th frame's PSNR ? :p
As I'm the third to ask and no one seemend to answer, I think it would be glad all the others (who surely understood, if they discussed the results without asking what they mean) or Sagittaire make a little step down the stairs to educate at least three ordinary folks.
If the results weren't that different between first and second graph I wouldn't have been much worried about my lack of comprehension...

Thanks in advance,
Vinouz.

[C0mPr355]

@ Sagi

I believe the best results for psnr you can get on vp6 should have the post processing level set to "Best PSNR." This can be done in Start>settings>control panel>vp6 decoder settings.

also could you retry your test with an auto keyframe of 360 rather than 120? I have stated that 360 is a much better number to use because of the way that vp6 generates keyframes.

Please only use Spatial and/or Temporal resampling when you are unable to hit the datarate on much lower bitrate clips. A datarate of 500-1000 really should not have a need to drop frames or dynamically downsize the encoding resolution to hit the datarate.

[Assault]

@ vinouz

I'll try to explain the difference between Overall PSNR and Average PSNR as far a I understand it. The average PSNR (first graph) is calculated by the average PSNR values (as the name already says ;))of the single frames. This means for the final result it doesn't matter how constant the quality of the video is. As an example you can take three frames. Codec A delivers a result of 10dB for the first frame 40dB for the second and 70dB for the third. Codec B delivers 39dB, 40dB and 41dB. The average PSNR for those two codecs will be exactly the same. Nevertheless in "real life" you would certainly prefer the quality of codec B because nobody wants a frame with 10dB in his encode. ;)
Overall PSNR on the other hand takes into account how constant the quality of the video is. But please don't ask for the exact formula of Overall PSNR. It's quite complicated (You can search in the Avisynth forum when you are interested).
So IMHO the Overall PSNR is preferable because it's the aim of the 2pass mode to deliver as constant quality as possible.
I hope that helps.

Assault

[Sagittaire]

Post-Process decrease PSNR and "Best PSNR" too ... for my trailer in 500~1000 Kbps interval

keyframe maxi 360 or 120 frames obtain the same PSNR because the encoder himself choose the keyframe position and frequency for bframe is generaly inferior at 360 frames ...

The VP6 is really bad with the PSNR Overall and really good with¨PSNR because it encode very well the not very complex frames which are in a majority and very badly the complex frames which are minority. Many blocking and ringing in the fast scenes which are cleaned with a very good Post-Process. The VP6 is thus not a good codec to make true VBR ...

[vinouz]

@Assault : thanks for the explanation.

[MfA]

"~~ PSNR, SSIM, VQM Test ~~"

?

[C0mPr355]

@ Sagi

can you please post the filters you used for Avisynth to resize the image as well as where the source is from (the dvd ?).

ty

[superdump]

Vinouz: The PSNR formula involves a logarithm. So if you average the PSNR values across the frames it gives a different and biased result as opposed to averaging all the MSE values and then plugging that into the PSNR formula at the end which gives the overall PSNR. Search for PSNR formula on google or something. You'll see what I'm talking about. :)

[Blueseb]

Well, I watched XviD and VP6 samples and, as far as I can trust my eyes, it seems that is VP6 that have a more constant quality (& generally less detail) against XviD that have a more variable quality (& generally more detail). All this IMHO, correct me if I'm wrong :p

[vinouz]

@superdump : thanks. All I wanted to know is what "Overall" was relating to. For mathematics I'm OK (E[ln(dx)] != ln(E[dx]) of course). What I didn't know is that "PSNROverall" meant PSNR over the average SAD of all the frames, instead of average of PSNRs of all the frames for "PSNR". Is that what you meant ?

[SeeMoreDigital]

PSNR test's are all well and good. But they can be easily fiddled by altering the brightness, contrast or colour of the source prior to encoding!

They are no replacement to a good set of eyes. Even leaning your head to one side can make an image look worse. And if you don't believe me, try watching say, the TV with your head at 90 degrees to the image. The closer you are to the screen the worse the lines look!

Amazing, the tricks your eyes can play on you, in it!

Cheers

[Sharktooth]

Had a look at those avis and there are some things i've noticed.

xvid-600.avi has good details but shows extreme "blockiness" in high motion scenes, some texture motion visible in still backgrounds.

divx-600.avi has good details and no blockiness even in high motion scenes but background texture movement is clearly visible in still backgrounds.

vp6-600.avi has very good details in low motion scenes but blurs out in high motion. No blockiness but shows weird colors when playing very high contrast scenes.

vmw9-600.avi shows variable details from nearly good to very good. No blockiness seen even in high motion scenes.

My opinion is vmw9 and divx are the best encodes.

EDIT: I didnt watch the RV9 clip.

[Sirber]

My opinion is vmw9 and divx are the best encodes.Have you seen that RV9 beats every codec at 600kbps looking at the PNSR scores? I bet you haven't tryed the RV9 clip. How can you give your opinion whitout watching the whole thing?

[ookzDVD]

I think we should add the new 3ivx into the test. :)

[Sharktooth]

Originally posted by Sirber
Have you seen that RV9 beats every codec at 600kbps looking at the PNSR scores? I bet you haven't tryed the RV9 clip. How can you give your opinion whitout watching the whole thing?
If you carefully read my post you see i didnt watch the RV9 encode. The reason is WE ALL KNOW ITS THE BEST.
I just watched the other encodes and draw some conclusion.

[Sirber]

Is it forbidden to talk about the best codec? to review it with others? If someone read your post and don't know much about video compression, what will he think?

[superdump]

We don't all know it's the best. We all know it gets the best PSNRs. I still find it slightly blurry though the high frequency detail retention thing Karl added had made it better. I still prefer the details (and yes Sirber, I can live with some blocking occasionally :P ;)) of XviD myself but, each to their own.

[Sharktooth]

I haven't the time to discuss about silly arguments so I edited my post...
Hope its all clear now.

[iwod]

I asumer when PSNR is being compared Xvid and DIvx don't have pp on........

But i would think RV9 would have it as by default.

So how would PSNR have anything to do with quality?

[deXtoRious]

That's the real problem with these tests. They only test the codecs technically, they test how the video is encoded. They don't really have much to do with real quality because the final movie that we see is always (unless of course you are some kind of a masochist not using pp ;)) irrevokably altered by post-processing in order to increase the relative quality that we perceive. It is rather pointless to try to rationalize this as it does not really give us feasable information that could be used to determine the best codec available (yet some people try to interpret these results as absolute ratings of quality). However, we do it anyway :D

[Sagittaire]

Update ... ;-)

1) SSIM Test
2) VQM Test
3) Script used for encoding and tests
4) 3ivX Demo
5) Vob source: 75 Mo ...

[SeeMoreDigital]

Originally posted by Sagittaire
Update ... ;-)

1) SSIM Test
2) VQM Test
3) Script used for encoding and tests
4) 3ivX Demo
5) Vob source: 75 Mo ... Thanks for posting the 3vix encode.

It doen't look half bad. Better than I've been able to manage so far (mine have been 720x576 anamorphic).

Can you confirm the settings you used please?

Cheers

[Sagittaire]

Defaut setting and FouCC XviD for ffdshow decoding ... or desactive Post-process with 3ivX decodeur ...

PP = :devil:

[Sharktooth]

Sagittaire (if you have some spare time) can you plz add the final divx5.1.1 to your comparison?

[bilu]

Why do the VQM results seem opposed to the PSNR and SSIM ones? :confused:

Bilu

[superdump]

Because a lower VQM value is better. :)

[On2Tech]

Thanks for performing this comparison. It's helped On2 find some bugs in VP6. We'll have a new version out in the next couple of weeks.

Here's a preview of where we stand now: I'll post a description of the bug fixes in my next post.

Clip Size ---- PSNR -----
Overall Avg
======= ======= =======
HP.rmvb 10,387KB 41.4208 43.9264
HPVP6.avi 10,424KB 41.5209 44.3266

Description
===================
HP.rmvb Real clip produced with RMFactory & your settings
HPVP6.avi VP6 clip built with the prealpha new version.

NOTE: When the codec is finally released the results may differ ( we hope they'll actually be better).

[On2Tech]

1) Missed Key Frames

The vp6 encoding missed marking 20 some-odd frames as keyframes. Since VP6 generally encodes the keyframes at a higher quality than surrounding interframes this hurt the overall quality.

2) Keyframes that were way too big.

We realized that the VP6 keyframes were being compressed at datarates that are much higher than they should have been. This meant we effectively wasted a large number of bits making the first frame of a cut scene pristine at the expense of the rest of the frames of the section.

3) Two Pass Quality Control

We were very conservative in our two pass datarate control. More aggressive quantizers seem to help the Overall PSNR number by quite a bit.

4) Exposed Encoder Complexity Number.

We have some control over how much time we spend performing motion searches, we had it disabled because we wanted to make sure it ran relatively quickly by default.

[On2Tech]

One point about these clips. I agree with those people who claim that both PSNR and Overall PSNR are extremely poor measures of a clip's quality. Especially in the case of a clip such as this Harry Potter trailer in which the entire clip is a conglomeration of a large number of different small sections.

In this clip there are a large number of black frames. Getting these black frames exactly right means you get a PSNR score of 100 in the avg PSNR metric. Some of the encodings we did of VP6 produced an exact match of these black frames others produced psnrs near 55. Problem is they all look black. Visually there is no major difference between 55 and 100 db on a black frame but there is a huge difference when it comes to average PSNR.

Likewise Overall PSNR is heavily biased against the odd bad frame. In our clip of 3176 frames a single really awful frame ( let's say with a PSNR of 10) would drag the entire clip's overall PSNR number down by 1.6 db even though its just 1 frame.

I suggest that PSNR is far more relevant on small single scene clips like those in the MPEG-2 test suite.

[CruNcher]

Originally posted by On2Tech
Getting these black frames exactly right means you get a PSNR score of 100 in the avg PSNR metric.

ehm isn't it actually infinite then ?

[Sagittaire]

PSNR isn't a very good video quality mesurement but Overall PSNR, SSIM and VQM are convergent results and not in favour to VP6 ...

[Sirber]

J'té cassé! (Brice de Nice)

(eng: I broke you!)

[On2Tech]

When we made some simple fixes we beat Real 9 's Overall PSNR score on the Harry Potter Clip by .1 db !!! I'll check the other quality metrics tomorrow.

B.t.w. We used rmfactory downloaded from the site yesterday with the exact audience you listed in your first test.

The fact is 2 or 3 bugs in the two pass compressor hurt vp6 badly on that clip. Don't take my word for it. Wait a couple of weeks download the codec and try again ( or pm me and/or Compr355 and we might get you an early copy).

You will have your shot at it soon...

Note: The psnr isn't the only thing that improved. Visually the quality is much better than the vp6 file you posted. I'll test using the other quality metrics as well, but I have no doubt that you will see massive improvements.

P.p.s. With the VP6 post processor set at deblock only ( ie no deringing, no noise filtration or black clamping etc) helped the clip we built by .5 db on overall psnr ( like .3db on avg). I am not sure why your results differed.

[On2Tech]

Originally posted by CruNcher
ehm isn't it actually infinite then ?

You are correct, you do get infinite psnr for perfect reconstruction. Unfortunately I was referring to the results you get when you use the Compare function within avisynth. For whatever the person who wrote that code settled on a perfect reconstruction being 100.

I think 60 might have been a better choice but that's a post for a different forum.

[Stux]

Originally posted by On2Tech
I think 60 might have been a better choice but that's a post for a different forum.

Probably clamping to 60 then, because you can get near perfect reconstruction and actually end up with 70, 80 or 90dB

[Sharktooth]

Glad to see VP6 is getting better.
Keep up the good work guys.

[Sagittaire]

VP6 600 Kbps


Minimum Average Maximum
Mean Absolute Deviation: 0.0007 1.2970 3.7648
Mean Deviation: -0.1555 +0.0432 +0.2624
PSNR: 32.7076 43.6361 79.3822
Overall PSNR: 41.2033


RV9 600 Kbps


Minimum Average Maximum
Mean Absolute Deviation: 0.0006 1.2024 2.6072
Mean Deviation: -0.2449 +0.1311 +0.7298
PSNR: 36.4788 43.7111 80.0075
Overall PSNR: 42.6571


Very good average PSNR but Overall PSNR is very bad (same result with SSIM and VQM). Encoding is very good for slow motion (frames generaly in majority) and very bad for high motion (frames generaly in minority) thus good average PSNR but very bad overall PSNR (More influance for complexe frames in PSNR overall)

SSIM and VQM (Psycho video quality mesurement) are very bad with VP6 I don't understand why ... ??

Best Profil for PSNR and RV9


<?xml version="1.0" encoding="UTF-8"?>
<audience xmlns="http://ns.real.com/tools/audience.2.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://ns.real.com/tools/audience.2.0 http://ns.real.com/tools/audience.2.0.xsd">
<avgBitrate type="uint">580000</avgBitrate>
<maxBitrate type="uint">5800000</maxBitrate>
<streams>
<stream xsi:type="videoStream">
<pluginName type="string">rn-videocodec-realvideo</pluginName>
<codecName type="string">rv9</codecName>
<codecProperties type="bag">
<encoderComplexity type="uint">100</encoderComplexity>
<customPacketSize type="uint">16000</customPacketSize>
<noisyEdgeFilter type="bool">false</noisyEdgeFilter>
<patternAdaptivity type="uint">1</patternAdaptivity>
<scalingFactor type="uint">80</scalingFactor>
<calcPSNR type="bool">true</calcPSNR>
</codecProperties>
<encodingType type="string">vbrBitrate</encodingType>
<maxStartupLatency type="double">60</maxStartupLatency>
<maxFrameRate type="double">25</maxFrameRate>
<maxKeyFrameInterval type="double">10</maxKeyFrameInterval>
<enableLossProtection type="bool">false</enableLossProtection>
</stream>
</streams>
</audience>

[On2Tech]

Please be patient with us. We'll have a new codec for you to try shortly. The overall Quality and SSIM, and VQM measurements you got were bad for VP6 due to a set of bugs in the VP6 two pass compressor. I'm sorry we released it with these bugs. Unfortunately our own tests weren't adequete in the two pass arena. Your compressions revealed what our own tests failed to. Thanks.

The measurements I posted were with an almost fixed codec ( not the one you can download on On2's site). On the Harry Potter clip I got results that are better than real. My audience file in rmfactory only differed from yours in that I shot for 600000 instead of 580000). If you like I'll try again with 580000. ( I copied the audience file from your last compression test), though its kind of a moot point. My numbers won't mean much until everyone can verify them (probably after our thanksgiving break).

Visibly the results with the fixed codec were massively better than the version you produced. I'd like to post the clip, but company policy means we can't distribute anything that we don't have explicit rights to use "even trailers"..

In your version VP6 produced a couple of really terrible sections ( because we missed a keyframe, and didn't clamp q enough). You can check your VP6 clip for yourself. Open it up in Virtual dub and watch for the K that indicates keyframes. You'll find 20 or so missed keyframes.

You'll also see that our two pass datarate control wasn't very two pass. Due to another bug we effectively ended up shooting for a much flatter datarate. This meant that the PSNR was spikey which is bad for overall PSNR.

Note: You would have gotten somewhat better results out of VP6 by using a max quantizer of 15 instead of 40; but you shouldn't have to know that. The two pass compressor should handle things like that. Plus the other bugs (missing keyframes, and keyframes that were 10 db better and 20 times the size of subsequent frames etc) still would have resulted in VP6 encodings that were sub par and noticeably worse than Real.

NOTE: The effect of these bugs was very material dependent. A full movie wouldn't have had nearly as many missed keyframes ( the bug had to do with how close they were together). Plus big Keyframes don't hurt so much when their spread apart.

Again Thank you for including VP6 in your comparison! We now have a better codec because you did.

[Sharktooth]

That explains some weird behaviours of the codec.
Infact i tried VP6 with different sources having completely insonsistent results.
Now that you spotted those bugs i cant wait to test the new codec build:)

[Lefungus]

If my VQM plugin was used in these tests, i'd like to have the results removed as it's very experimental and nothing has been done to ensure results were correlated to video quality.
Everwicked has done a much better job for the vqm test in his video quality studio app.

[Sagittaire]

Thanks for informations
http://www.everwicked.com/vqstudio/

Result for VQM are convergent with Overall PSNR and SSIM thus i think your VQM algo isn't bad ...

But Update for the VQM with vqstudio

[Sagittaire]

Update: Vob Source

Trailer 128 sec with very High Quality MPEG2 source
PAL 25 fps progressives frames
For test yourself http://livrerongeurs.free.fr/Vob/

[bilu]

@Sagittaire

After watching this thread about 3ivx I got quite interested to see you testing it :)

http://forum.doom9.org/showthread.php?s=&threadid=65719

If you ever have the patience, encoding times on the results table would also be interesting.


Bilu

[Sagittaire]

No bframe support for MPEG4 codec is stange and result with PSNR, Overall PSNR, SSIM and VQM are not very good ...

[Sagittaire]

Update with XviD 1.0 beta if you want ... ;-)

PS: XviD Devapi is compatible with ffvfw: we can make XviD Devapi4 with multipass now ...

[Sagittaire]

Update with new VP6 codec ... !!!

Next: Test in HDTV resolution ... ;-)

[bilu]

Wow, VP6 really improved... :eek:

[McoreD]

Appreciated this great information. :)

According to the PSNR Overall results, looks like VP6 beats every other codec at any given bit rate (from 500 Kbps to 1000 Kbps). Well I thought WMV9 was an excellent low bit rate optimized codec, but it looks like RV9 beats WMV9 for bit rates lower than ~900 Kbps. After 900 Kbps WMV9 starts picking up. The graph doesn
t provide information higher than 1000 Kbps but I still guess VP6 may be beating WMV9 for even higher bit rates. Have you guys noticed when you encode in WMV9, for bit rates higher than ~1000 Kbps, Windows Media Player shows the WMV9 codec as Windows Media Video 9 Professional. Well that's off topic. My own conclusion is the best so far is VP6 and then WMV9. (Based on Overall PSNR *only*).
Needless to comment about DivX or XviD.

[Sagittaire]

For compressibility > 60% very good quality for all codec and DivX3 SBC too for exemple. WMV9 is very good codec with low and high bitrate ant compatible with anamorphique resolution in wmv container with WMP9 ... ;-)

Interessing test is the next test in HDTV resolution with 1.5 ~ 3.5 Mbps ...

[SeeMoreDigital]

Originally posted by Sagittaire
For compressibility > 60% very good quality for all codec and DivX3 SBC too for exemple. WMV9 is very good codec with low and high bitrate ant compatible with anamorphique resolution in wmv container with WMP9 ... ;-)

Interessing test is the next test in HDTV resolution with 1.5 ~ 3.5 Mbps ... Yes, this will be interesting. But what image pixel frame size are you using?

Cheers

[trbarry]

The new edited and revised test results show VP6 way better than anything else and also way better than the VP6 results first posted.

I'm impressed.

But to what extent is that because there was an opportunity to go back and refine VP6 code or parameters for this particular series of benchmarks? I like VP6 and I'm not implying anything wrong, but I sometimes worry about such things.

- Tom

[karl_lillevold]

A very nice VP6 improvement in Overall PSNR and VQM for this trailer. I have not worked much with Overall PSNR, nor VQM, generally only average PSNR, but I have some thoughts on how some of the adaptivity in RV9 may not be optimal in certain ways for these measures. It will be interesting to see how a "Harry Potter Edition" of RV9 will behave :) Maybe after New Year's I will have some time to look into it.

[Lefungus]

These results are flawed in my opinion as some codecs are tested with post-processing, and some aren't. A good post-processing filter could increase average dB (in PSNR) by 1 and more. Some experimental post-processing filters could increase it even more in special cases.

EDIT: The usual disclaimer:
I've never tested my vqm implementation, so even if the results seems to be "ok", i wouldn't trust it too much as i know it's not really a good implementation.

[Sagittaire]

@ Karl

Perhaps that by having a configurable Post-Process on Rv9 would have a means of optimizing the tests ...

I makes a test in 1280*720 in 1.5 ~ 3.5 Mbps and the RV9 without PP. With these resolutions the RV9 take really too many resources because of the PP I think ...

Thus a configurable PP would be very good evolution for the RV9

Thank you in advance Karl ... ;-)

@ Lefungus

I make test with PP0,PP2,PP4 and PP6 for XviD and DivX with ffvdshow and DivX dec. PP0 obtains the best results on the interval [500-1000] Kbps with XviD and DivX. For the WMV9 the best results are obtained with PP0 or PP1. With the VP6 it is with PP4. It is perhaps possible to obtain a PSNR with 1 dB in more but not with dec MPEG4 available ...

[Dark-Cracker]

Hi,

i have just a little question. can u set the keyframe interval on VP6 ?

because on RV9 u set it at every 10 sec and on VP6 it's
"Autokeyframe ON" & "Max Frames Btw Keys 120" perhaps sould u take a sample much longuer than 128 sec just to be sure KF doesn't false the result of the PSNR test :)

++

[Sagittaire]

MaxKeyFrame doesn't have an influence because IFrame for VP6 and RV9 are placed by a detection of change of scenes. For this trailer MaxKeyFrame must be rarely used because the scenes are short.

MaxKeyFrame is a limitation which makes it possible easily to navigate in the video but which doesn't have any influance on quality has less to choose a very low value.

[On2Tech]

Originally posted by karl_lillevold
A very nice VP6 improvement in Overall PSNR and VQM for this trailer. I have not worked much with Overall PSNR, nor VQM, generally only average PSNR, but I have some thoughts on how some of the adaptivity in RV9 may not be optimal in certain ways for these measures. It will be interesting to see how a "Harry Potter Edition" of RV9 will behave :) Maybe after New Year's I will have some time to look into it.

A cheap way to get a good avg PSNR boost on this clip is to get 30 db on the 200 or so black frames of the clip by spending 40 or so additional bytes on it. That's why I'd vote for clamping PSNR at somewhere between 50 and 60 db.

Note: I think there's still quite a bit of room for improvement in the VP6 trailer's datarate allocation. It's by no means optimal. By tweaking the internal datarate controls on individual clips I got results that were considerably better (.3 - .5 db) better in overall and / or avg than what sagittaire reported.

[On2Tech]

Originally posted by trbarry
The new edited and revised test results show VP6 way better than anything else and also way better than the VP6 results first posted.

I'm impressed.

But to what extent is that because there was an opportunity to go back and refine VP6 code or parameters for this particular series of benchmarks? I like VP6 and I'm not implying anything wrong, but I sometimes worry about such things.

- Tom

It's always the case that you can bias the codec for a specific clip or type of clip. That isn't what happened here.

The first test sagittaire did exposed a couple of quite silly bugs in the VP6 codec, that we didn't typically see because we use our own tools for compressing and testing clips ( ie not virtual dub). I talked about them in some prior posts.

We also did take to heart that people really didn't care just about avg psnr but also about the worst sections, as a result we made the two pass compressor more aggressively spend bits on tough sections but then ran the new code on a bunch of clips and full length movies.

Your point is well taken however. One clip or one small set of clips doesn't make a good indicator of the overall quality of codec. The more people test codecs and publish the results they get the better.

[Sagittaire]

black frame dropped with avisynth script ...

Video=Mpeg2Source("G:\Mes dossiers\B.A\Harry Potter\azerty.d2v")
Video=Trim(Video,70,3145)
Video=Crop(Video,16,76,688,426)
Video=Convolution3D(Video,0,3,4,3,4,3,0)
Video=LanczosResize(Video,640,272)
Return(Video)

Test with Matrix reloaded movie (+2H00) done same result with 700 Kbps 640*272 (not test with 500, 600, 800, 900 and 1000 Kbps) ...

[Sagittaire]

Update ...

RV9 Encoding

RV9 EHQ VBR High Speed

for best quality with these Codec

for best quality with WMV9 install this reg
http://jfl1974.free.fr/Video/PP-WMV9.reg

for best quality with RV9 EHQ VBR High Speed install this reg
http://jfl1974.free.fr/Video/Enc-RV9.reg
And this dll in producer/codec/
http://www.lillevold.com/files/erv4...Pass_120203.zip

for best quality for test with VP6 install this reg
http://jfl1974.free.fr/Video/PP-VP6.reg

The VP3, DivX3, 3ívX obtained very bad results during these tests ...

In the principal post you will find the setting to obtain the best result with each codec ...

For me the best test is SSIM because it corresponds exactly to my visual impression ...

[karl_lillevold]

I was trying to look into some potential high bitrate RV9 changes, and wanted to compare with the latest VP6 and its nice improvement. The problem is I am having some trouble measuring these numbers correctly, even though my RV9 numbers match the graphs reasonably well.

I did a test at ~ 730 kbps


Filesizes:
RV9 11,480 KB
VP6 11,524 KB

PSNR:
RV9 44.3118
VP6 43.0808

Overall PSNR:
RV9 43.2089
VP6 42.5830

Something must clearly be wrong, perhaps PP settings for VP6?

I installed the latest 6.0.9.2 version of VP6, I used Sagittaire's settings from the 1st post, and I have these reg keys:

[HKEY_LOCAL_MACHINE\SOFTWARE\On2 Technologies\VFW Encoder/Decoder Settings\VP60]
"strPostprocessinglevel"="4"

[HKEY_LOCAL_MACHINE\SOFTWARE\On2 Technologies\VFW Encoder/Decoder Settings\VP61]
"strPostprocessinglevel"="4"

Looking at the VP6 AVI file in VirtualDub(VFW) I see artificial film grain, so I think the problem is the DirectShow settings, where I have disabled Film Grain, do not apply to the VFW decoder used by Avisynth. How do I disable Film Grain and make sure I have the Best PSNR post-filter in VFW VP6 decoder?

Using DirectShowSource for the VP6 AVI file in the Compare script results in RunTime Error in Vdub.

[karl_lillevold]

I found the problem. The reg key strAddNoise needs to be included:

[HKEY_LOCAL_MACHINE\SOFTWARE\On2 Technologies\VFW Encoder/Decoder Settings\VP61]
"strPostprocessinglevel"="4"
"strAddNoise"="0"

Now the numbers are much closer:

PSNR:
RV9 44.3118
VP6 44.3851

Overall PSNR:
RV9 43.2089
VP6 43.4033

nice. very close indeed.

[Sagittaire]

@ Karl

Good year, good health ... ;-)

For me the best test is SSIM because it corresponds exactly to my visual impression. RV9 for very low bitrate is the best for my eyes. For weak bitrate the Post-Process of the RV9 is really powerful but for high bitrates it's too strong ...

[bilu]

Originally posted by Sagittaire
RV9 for very low bitrate is the best for my eyes. For weak bitrate the Post-Process of the RV9 is really powerful but for high bitrates it's too strong ... I agree, it should be possible to make a more adaptive in-loop filtering.

Bilu

[Sirber]

Yep, adaptive in-loop filtering could be a kick in the butt for med and high bitrate. But, changing this require changing the bitstream, so we might only see that in RVX. But, I think with disabling b-frames for high bitrate could improve the quality.

[bilu]

Originally posted by Sirber
But, changing this require changing the bitstream, so we might only see that in RVX. I didn't think it would be that much of a trouble. So in-loop filtering used by RV9 isn't just a fixed algorithm that could be changed into something more adaptive to bitrate and sources?

Bilu

[snowcrash]

Originally posted by Sagittaire
Update ...
for best quality with RV9 EHQ VBR High Speed install this reg
http://jfl1974.free.fr/Video/Enc-RV9.reg


Can you explain what these settings do?

"calcPSNR"=dword:00000001
"firstPassComplexity"=dword:00000041
"encoderComplexity"=dword:00000064
"customPacketSize"=dword:00003e80

In my registry I did not even have an RV9 folder under HKEY_LOCAL_MACHINE\SOFTWARE\RealNetworks.

[karl_lillevold]

@snowcrash: the search button is very useful. In this case, it will take you right to threads explaining these keys, that do not normally exist. Search for each one separately. Note that these keys have their audience file equivalent, and should be used carefully, since they apply to all versions of Producer, and over-ride any other setting, incl. the audience file, so do understand what they do, and don't forget that they have been applied, if you do use them ...

[bilu]

you mean "@snowcrash" not "@bilu" :sly:

[karl_lillevold]

I apologize. post edited.

[Shandra]

Hi Folks,

By the way - either I overlooked that information or it wasn't mentioned before.... After some tests with VP6 I feel inclined to use it in some cases and I will def. look after this codec from now on (as I still wait for Theora) ... now the simple question How do you calculate the aprox. File size ? I would like to compare lets say a divx with a VP6.1 (with various settings) based on same Filesize - but so far I had simply no luck to get my resulting VP6 in range of the DivX size (In this case a 20min 587Kb/s DivX was with MP3 115MB, with the same MP3 I don't even matched that size going as low as 72KB (wich *8 would be around (somewhat less) the targeted Bitrate of my DivX (or am I wrong on this)- I know apple and oranges, but I truly would like to compare results by size of file ;) )... and going below that I fear the quality I now adored in VP6 (my first trial was out of not having read through all relevant specs/threads here with a KB of 512 for VP6 and the result was brilliant - but the resulting 700+MB was not near my wanted ~116MB for my 20min clips...)

P.S.: I hope the adressing of the generated badframes.stt is still not forgoten ;)

[On2Tech]

Originally posted by Shandra
Hi Folks,

P.S.: I hope the adressing of the generated badframes.stt is still not forgoten ;)

We have a fix for that problem. We hoped to address the problem with interlaced material before releasing the fix ( we should have it early this upcoming week)..

[snowcrash]

Originally posted by snowcrash
Can you explain what these settings do?

"calcPSNR"=dword:00000001
"firstPassComplexity"=dword:00000041
"encoderComplexity"=dword:00000064
"customPacketSize"=dword:00003e80


Ok, I will answer my own question, for those who may not know, have only a casual interest in RV9 and don't feel like doing a half hour of reading to find out about the registry hacks that Sagittairre is recommending here:

"firstPassComplexity"=dword:00000041
(speeds up encoding by doing fast first pass)

"customPacketSize"=dword:00003e80
(increases compression efficiency for high action)

"encoderComplexity"=dword:00000064
(sets EHQ level to 100 (?), not sure why it would be 100 and not 80, but hex 64 is indeed equal to 100, I assume this just sets EHQ to "very high")

"calcPSNR"=dword:00000001
(couldn't really find a clear explanation, but I think it computes the PSNR of the output and writes it to a log file)

[karl_lillevold]

@snowcrash: excellent summary. Thanks! Let me then add the following:

customPacketSize : not needed in newer Producers, which will automatically use a large packet size for high bitrate VBR.

encoderComplexity: This setting and how to use it and its equivalents have changed around too much, and will change more, before we are finally done :) 85 is enough. 100 slows down the encoder with only a tiny tiny gain in quality.

firstPassComplexity: a fast 1st pass will be the new default soon, since it appears to work very well, without any loss in quality. In current releases this setting is needed.

calcPSNR: yes, correct. This prints PSNR / Average PSNR and other numbers in rv9log.txt in the 'current' directory. This PSNR is for Y only, Sagittaire's numbers are for YUV and using Avisynth's Compare function.

[Sagittaire]

Final Update ...

Download and compare ... :devil:

~~ Download ~~


http://jfl1974.free.fr/Video/VP6-600.rar
http://jfl1974.free.fr/Video/RV9-600.rar
http://jfl1974.free.fr/Video/WMV9-600.rar
http://jfl1974.free.fr/Video/XviD-600.rar
http://jfl1974.free.fr/Video/DivX-600.rar

[SeeMoreDigital]

Hi Sagittaire,

Can I ask what setting you used for your WMV9 VCM encode?

As when I generated my own WMV9 VCM encode (from the same DVD source) after setting the performance slider to 'full bar' I seem to obtain more stable background images compared to your encode!

Cheers

[Sagittaire]

Not same script perhaps ... very difficult script in my test (Light filter and Lanczos)

Script AviSynth
It's a very simple and usual AviSynth Script
Video=Mpeg2Source("G:\Mes dossiers\B.A\Harry Potter\azerty.d2v")
Video=Trim(Video,70,3145)
Video=Crop(Video,16,76,688,426)
Video=Convolution3D(Video,0,3,4,3,4,3,0)
Video=LanczosResize(Video,640,272)
Return(Video)

WMV9
setting with VirtualDubMod: Maximum quality + Complex + 2 pass

[SeeMoreDigital]

Personally I'm not using any scripts, just MPEGmediator 1.5!

I notice that your source/encode is 5 seconds shorter than mine - you seem to have trimmed some of start and end black fade off.

I only used the 'Main' and not 'Complex' setting too. So I will generate another test using the 'Complex' setting and report my findings.

Cheers

[Sagittaire]

Drop Black Frame Video=Trim(Video,70,3145) is necessary:

1/ Blacks Frames are bad influance for PSNR, OverallPSNR ...

2/ 5 sec for 128 sec:
Real Bitrate for video = Average Bitrate*128/123

[SeeMoreDigital]

Hi Sagittaire,

I generated another encode using 'complex' instead of 'main' and there was no perceivable 'visible' difference between the encodes!

I am however quite surprised and intrigued as to why your 'script' generated does not look as good as my 'non script' crop/resize encode - well to my eyes anyway!

Sagittaire, I have to admit I have not read all your posts in this thread but I seem to remember you talking about generating some HiDef encodes. Are you still thinking about this?

Cheers

[Sagittaire]

Update with XviD-1.0-RC2

XviD 1.0 really better than DivX 5.1.1 for this test now ...

[veryhappyjk]

I tried VP6 6.0.9.x and 6.1.0.2 with some movies, and found the quality in pure VBR is amazing, really ahead divx and xvid.

BUT!!!!!

why On2 didn't develop an interface for 1-pass VBR???

actually to compress this way I have to set:

(1. "one pass - best quality / local file playback (VBR)") - this is OK;

2. ****the ONLY option that affects VBR quality**** is the MAXIMUM parameter under "Advanced\Adjust quantizer".

Other parameters do NOTHING!!!!

why don't you develop a correct inferface???

[Shandra]

Originally posted by veryhappyjk
Other parameters do NOTHING!!!!
why don't you develop a correct inferface??? [/B]

Uh - don't tell me setting the target bitrate doesn't do anything ;) And what form of interface do we need if we have VirtualDub(Mod) or every other program that is addressing direct show codecs the way it was ment to be (not this nancy fancy divx style codec property page) ???

By the way.... even with the latest VP6 I am still getting the Bad Frames file :(

[veryhappyjk]

I'm interested in one-pass VBR, i don't care about a target bitrate! :cool:

I'm not talking of a compressor GUI; the trouble is when you switch to one-pass VBR, a sort of Quantizer should appear and other useless parameters disappear, but this doesn't happen.

Saverio M.
AKA MadActer

[Shandra]

Originally posted by veryhappyjk
the trouble is when you switch to one-pass VBR, a sort of Quantizer should appear and other useless parameters disappar, but this doesn't happen.


Sorry, maybe I don't get the point.... ok, I think I know what you want (just one parameter -> quantizer/quality rating)... Well then... sorry - missunderstanding on my side ;)

(But as the codec desires a target bitrate my feeling to this would be that this approach would be not just a minor adjustment of I/O(UI) Part of the codec but would call for a major adjustment within the code.... but on that I am really to be called "Newbie" as I never thought and therefore researched about such one pass-quality based approach (I never thought about such approaches because I would never trust a codec in as I call it "fire at will"-mode to do as I would like to do - so sorry again for the missunderstanding & playing with the quantizer min/max & undershoot value for now is in combination with aimed bitrate enough testing/playground field for now with me (in getting a feel towards the codec (as always learning-by-doing is what suits me most but also the cause why I am not that ideal person for deep technical semantics))

[veryhappyjk]

@La Sagittaire

Where can I find tools to make my own comparison??? TNX


@Shandra

One-pass vbr is already implemented, and consequently quantizer selection.
The fault is there isn't the right interface implementation, so you have to try all possible parameters combination to find out which factors/options are related to one-pass quality; and in the end you find there is only one, which is the quantizer (graphically= "maximum" parameter.)

as I am a Vp6 supporter (yeah!) I think "one-pass-people" should have consideration :-) though i see today noone cares about one-pass encoding.

[veryhappyjk]

@La Sagittaire

Where can I find tools to make my own comparison??? TNX


@Shandra

One-pass vbr is already implemented, and consequently quantizer selection.
The fault is there isn't the right interface implementation, so you have to try all possible parameters combination to find out which factors/options are related to one-pass quality; and in the end you find there is only one, which is the quantizer (graphically= "maximum" parameter.)

as I am a VP6 supporter (yeah!) I think "one-pass-people" should be considerated (only a little :rolleyes: )
though i see today noone cares about one-pass encoding. :(

[Sagittaire]

Update with RealVideo 10 'Elysian'

[Big_Berny]

Hi Sagittaire,
your test is really great!
Two questions:
1. Do you used the newest VP6-version (6.2.0.10) in this test?
2. Can you post a speedtable of the different codecs? OR can you just tell me the speed which you get with VP6 (aprox.)?

Would be great
Big_Berny

[kilg0r3]

@ Sagitaire

I'd really like to know how xvid scores with the mpeg matrix and probabely even qpel and gmc.

[Sagittaire]

Best results with these seetings.

H263 or HVS better Quality
VHQ4
Ultra High
no Qpel and no GMC

Matrix MPEG isn't best matrix for XviD in this test ... ;-)

[h9903209]

Would you like to do a 1-pass comparison too? seems 2-pass result does not necessarily imply 1-pass the same result from my experience. Just wondering whether you also want to know the difference between 1-pass result of different codec and the difference between 1-pass and 2-pass of different codec... ^^

[Sagittaire]

Update with RealVideo 10 'Elysian'

- RV10 old Rate Control
- RV10 new Rate Control Inloop filtering Q12 & HFE=1

For me old RC is better for low bitrate and new RC is better for high bitrate ...

[kilg0r3]

@ Sagitaire

There is a way to turn of the inloop filtering of RV10 (encoder and decoder!) plus the high-frequency retention setting in the decoder.

BTW I am also a single pass user -> I'd be very interested in such a comparison.

Thank you very much for your efforts.

[eltoder]

Originally posted by Big_Berny
Hi Sagittaire,
your test is really great!
Two questions:
1. Do you used the newest VP6-version (6.2.0.10) in this test?
2. Can you post a speedtable of the different codecs? OR can you just tell me the speed which you get with VP6 (aprox.)?

Would be great
Big_Berny
I'm second for both :)
This test is really interesting. Thanks alot.

-Eugene

[Sagittaire]

VP6 6.2

Sharpeness is good for high bitrate but for not low/medium bitrate: more detail but more artefact too for low bitrate.

-> Best result with sharpeness = 0 for PSNR, OverallPSNR, SSIM in this test

[eltoder]

Originally posted by Sagittaire
VP6 6.2

Sharpeness is good for high bitrate but for not low/medium bitrate: more detail but more artefact too for low bitrate.

-> Best result with sharpeness = 0 for PSNR, OverallPSNR, SSIM in this test
But the first page still says it's using VP 6.1 ;)
And VP6.2 at sharpness = 0 produces somewhat softer than original VP6.1

-Eugene

[kilg0r3]

Some settings-nagging again. of course I assume you treid a lot of different settings. Yet is are my suggestion for Bframes anyway. BFrames 2/1.5/0.

Cheers

[webwonk]

Can One-Pass CBR testing be included? For those of us who do live video streaming this is the most crucial type of test paramter. Which of the popular codecs do it best in the 900kbs-2mbs range?:

VP 6.2.0.10
RV10 Elysian & HFE2
WMV9-Pro/VC9
h.264
XviD 1.0 rc4
DivX Pro 5.1.1

And what is the deal with post-processing? Does it help visual quality? Or does it tax the end-users' system too much (which might offset improvements)?

Any information and results would be most helpful. Thanks in advance.

Web

[karl_lillevold]

I just wanted to mention that I believe PSNR and PSNR-derivatives (SSIM, JND) will become less and less important for video as it is getting closer to transparency, just as has been the case for audio for many years already. Psycho-visual enhancements already exist that make the video appear more like the original, even though PSNR is decreased. How can this happen, you may ask.. The simplest example could be the video being shifted one pixel compared to the original. PSNR and all measurements that do some kind of comparison with the original, pixel for pixel, will be completely ruined, but visually, seeing the one pixel shift is almost impossible.

For instance, HFE2 should not be used for PSNR comparisons. Even the original HFE, like I think Sagittaire is using for his test, is not recommended. Here's a quote from the HFE2 post (http://forum.doom9.org/showthread.php?s=&threadid=75278). "Mathematical analysis between the source and the encoded video has shown that even though PSNR is reduced from HFE2, the visual resemblance is closer to the original."

Then, with regards to RV10 Elysian, in this post (http://forum.doom9.org/showthread.php?s=&threadid=71022&perpage=20&pagenumber=10) I explained how PSNR can be increased 1/2 dB just by setting patternAdaptivity=2, and rcHighBitrateReduce to 60. Still, my experience is that I prefer the default settings for visual quality, maybe even with
<rcPFrameRefQuant type="uint">6</rcPFrameRefQuant>
<rcBFrameRefQuant type="uint">9</rcBFrameRefQuant>
instead of the default 6 and 10. Others might prefer to max out PSNR though.

[CruNcher]

:)

[Sagittaire]

Update with Gladiator Test

[SeeMoreDigital]

Originally posted by Sagittaire
Update with Gladiator Test If your Gladiator source is from an HiDef 1080i 1.77:1 source, why not have a go at generating some 720x576 or even 1024x576 encodes?


Cheers

[Sagittaire]

Originally posted by SeeMoreDigital
If your Gladiator source is from an HiDef 1080i 1.77:1 source, why not have a go at generating some 720x576 or even 1024x576 encodes?


640*352 or 576*320 is the most common resolution used for 16/9 picture. perhaps later with another HDTV source ... Spiderman II HDTV trailer or Van Helsings II HDTV trailer for exemple ...

[SeeMoreDigital]

Originally posted by Sagittaire
640*352 or 576*320 is the most common resolution used for 16/9 picture. Personally I disagree,

Where ever possible it's advisable to keep as much vertical resolution as possible, otherwise why bother capturing and converting from an HiDef source in the first place?

Also remember that an 1080i source is interlaced not progressive like an DVD source.

All the 1080i source material I have looks great when converted to 1280x720. But if that's not possible then the only other option is to use 720x576 anamorphic.


Cheers

[karl_lillevold]

Google-translated conclusion:

"The RV10 "Elysian" with its new rate control and its threshold of activation of the inloop filtering shows from now on as powerful at all the bitrates. It is even with this source the most powerful codec with all bitrates. For the highest bitrates RV10 progressed the most. The excellent surprise comes from H264 ffvfw which is shown already powerful as of its first appearance in ffvfw. The VP6 if it disappoints a little with this source and this resolution remains an excellent choice. Even if the metric tests give a light advantage to the WMV9 compared to XviD 1.0, these two codecs are visually very close all the same. Thank you in Bugsan for its fantastic Java XviD Stats Viewer which allows a very fine analysis of the behavior of the codecs..."

[Sagittaire]

Update:
- Codec DivX 5.2
- Diaporama low and high motion

[Latexxx]

I'd like to see some results from the subjective part of Everwicked's Video Quality Studio (http://www.everwicked.com/vqstudio/).

[Sagittaire]

6 codecs x 4 bitrate = 24 tests and very slow test. Not possible make job with this log ...

I use AviSynth and virtualDub job for PSNR and SSIM test. It's impossible with VQM (LeFungus VQM.dll is not complete)

[Sagittaire]

Test in french (http://jfl1974.free.fr/HTM/32_Test_Codec_Video.htm) ... but Capture, Sample & Graphs are universal language

Update:

- H264 Nero Digital (MPEG4 AVC codec from Ateme)
- XviD 1.0.1 with H263 matrix (visually better with PP4)
- DivX 5.2.1 with best setting for Digital56K
- Bitrate equivalence (with metric)

[IgorC]

The samples of codecs has a quite significal different sizes. For example, VP6.2 500 kbit/s - 5.81 Mb and RV10 500 kbit/s - 5.41 Mb
So difference of size is 7,39%. But may be the differnce on lagrer files will be less.
Waiting for your next test with the upgrated RV10, WMV 9.1, XVID 1.1 (after release of course):)

[Razorblade2000]

How about adding SNOW to your comparison :D
It would be really cool to see how it performs :D

[guada]

It's astonishing Sagittaire,

All my encode visual that it is low or high bitrates gives winning the WMV9.
Besides, the middle bitrate to 1500 kbps that stays the properties of the Vp6.
But my general surprise comes of the codec H-264 SNOW (2pass), that at low bitrate (200kbps) is below the qualities of WMV9.

I would stain to give some pictures at next.

To soon ;)

[Sagittaire]

The samples of codecs has a quite significal different sizes. For example, VP6.2 500 kbit/s - 5.81 Mb and RV10 500 kbit/s - 5.41 Mb
So difference of size is 7,39%. But may be the differnce on lagrer files will be less.
Waiting for your next test with the upgrated RV10, WMV 9.1, XVID 1.1 (after release of course)


Sample extract from complete encoding but I update RV10 sample:

- H264 500 kbps : 5694 Ko
- VP6 500 kbps : 5952 Ko
- RV10 500 kbps : 5750 Ko
- WMV9 500 kbps : 5816 Ko
- XviD 500 kbps : 5824 Ko
- DivX5 500 kbps : 6038 Ko


Sample extract from complete encoding but I update RV10 sample:

- H264 1000 kbps : 6103 Ko
- VP6 1000 kbps : 5964 Ko
- RV10 1000 kbps : 6192 Ko
- WMV9 1000 kbps : 6198 Ko
- XviD 1000 kbps : 6208 Ko
- DivX5 1000 kbps : 6232 Ko

[708145]

But my general surprise comes of the codec H-264 SNOW (2pass), that at low bitrate (200kbps) is below the qualities of WMV9.

Apart from the fact that SNOW is not H.264 but wavelet based, it is still undergoing heavy development.
http://ffmpeg.sourceforge.net/index.php

But now back on topic:
I read all those discussions arguing about Sagittaires test methods, parameter selections, objective measurements, aso.
What I want to know is if you also object about his conclusions regarding codec efficiencies near the bottom of his page. (Like XVID being about 44% less efficient than h.264 @500kbit)

Do you think this is true WYOE (with your own eyes) as well?

bis besser,
Tobias

[guada]

A question: The numbers prove them really the reality ?
I have by habit to say, it is necessary to see to believe.
I don't hold to correct Sagittaire, I do note only that what I see.
But You, can you believe what you don't necessarily see.

It is necessary again that you see it; .:D


it was only a parenthesis.

[Sagittaire]

@ guada

In fact it's very easy: bitrate equivalence with metric graph (Average PSNR, Overall PSNR and SSIM). For example 43dB for H264 at 500 Kbps, 43dB with VP6 at 590 Kbps ...



@ 708145

Originally posted by 708145
Apart from the fact that SNOW is not H.264 but wavelet based, it is still undergoing heavy development.
http://ffmpeg.sourceforge.net/index.php

But now back on topic:
I read all those discussions arguing about Sagittaires test methods, parameter selections, objective measurements, aso.
What I want to know is if you also object about his conclusions regarding codec efficiencies near the bottom of his page. (Like XVID being about 44% less efficient than h.264 @500kbit)

Do you think this is true WYOE (with your own eyes) as well?

bis besser,
Tobias

Read transparent Teegedeck test (~q4 for XviD and 1000 Kbps in my test) in H264 Ateme thread ...

H264 Ateme for very low bitrate (~q8 for XviD and 500 Kbps in my test) is very very powerfull ...

[guada]

Thank you Saigittaire :)

Note: Sorry for this impertinent remark ;)

[708145]

Maybe I should rephrase my question:

Does the majority here believe that your (<- Sagittaire) bitrate equivalences are true for subjective evaluation as well?

(Ignoring the fact that those equivalences were determined using objective methods.)

bis besser,
T0B1A5

P.S.: This is NOT about my own subjective impressions. It's about the validity of objective tests.

[Sagittaire]

it's always the same question ... I don't know but:

On the complete Matrix 3 trailer this Ateme setting produced an output that was 37.05% smaller than XviD @q=4) - which most people will find transparent

[18%-27%] in my test with metric ...


Originally posted by Gej
We have tested several clips produced by the studios and mapped PSNR against bitrate for both codecs

DivX Developper use this technique for bitrate equivalence. One2Tech use certainely this technique too (VP4 25% better than VP3, VP5 15% better than VP4 ...)

[Sagittaire]

x264 seem very good too ...

http://jfl1974.free.fr/Video/x264vsVP6-1000.JPG

Average PSNR
H264 : 44.7337 dB
VP6 : 44.3975 dB
x264 : 44.2624 dB
RV10 : 44.0334 dB
WMV9 : 43.7891 dB
XviD : 43.6912 dB
DivX5 : 43.0575 dB

Overall PSNR
H264 : 44.1604 dB
VP6 : 43.9286 dB
x264 : 43.6708 dB
RV10 : 43.5277 dB
WMV9 : 43.3891 dB
XviD : 43.2531 dB
DivX5 : 42.4873 dB