VideoHelp Forum
+ Reply to Thread
Page 1 of 2
1 2 LastLast
Results 1 to 30 of 43
Thread
  1. Hi there from Brazil

    So, i guess this will really prove 2 things:
    1) If you know a little bit (dont have to be the master in it) about how to tweak the software encoders, you will get a very satisfatory output with any of this 5 encoders.
    2) The difference on the output from all this 5 encoders will not justificate the price difference the encoders have ($0 to $2000).

    THIS TEST IS SET FOR PROGRESSIVE SOURCE

    Encoders Used:
    Cinema Craft Encoder 2.70.01.05
    Canopus ProCoder 2.01.30.0
    TMPGENC XPRESS 3.0.4.24
    Main Concept MPEG Encoder 1.04.02.00
    QuEnc 0.59 Beta 2

    Source:
    High Quality .avi movie scene
    Codec = XviD
    Avarage bit rate = 3244 kb/s
    Res = 640x352
    FPS = 23.976

    Avisynth Script:
    AviSource("c:\c\teste.avi", false, fourCC="XVID")
    BicubicResize(720,352)
    AddBorders(0,64,0,64)

    ALL THE ENCODERS WERE CONFIGURED (did the best i could) TO OUTPUT THE BEST QUALITY POSSIBLE OUTPUT.

    COMMON SETTINGS FOR ALL THE ENCODERS
    VBR avg=5000 mim=0 max=9500
    GOP SEQ = IBBPBBPBBP... (two B-frames, like commercial DVDs)
    Standard Matrix (the default for every encoder)
    Aspect Ratio = 4:3

    OTHER SETTINGS WERE SET TO THE BEST POSSIBLE FOR ALL THE ENCODERS. SO, FOR EXAMPLE, IF ONE ENCODER HAD THE ADVANTAGE OF DOING MORE PASSES THEN THE OTHER (cce here), BAD LUCK FOR THE OTHERS THEN.

    TIME TAKED TO ENCODE (561 frames): (this test is for max quality, but...)
    Cinema Craft Encoder 2.70.01.05 1'42" (9 pass, real = 10)
    Canopus ProCoder 2.01.30.0 0'37" (2 pass)
    TMPGENC XPRESS 3.0.4.24 1'00" (2 pass)
    Main Concept MPEG Encoder 1.04.02.00 0'42" (2 pass)
    QuEnc 0.59 Beta 2 1'20" (2 pass)

    FRAMES:

    I-Frame

    SOURCE


    Cinema Craft Encoder 2.70.01.05


    TMPGENC XPRESS 3.0.4.24


    Canopus ProCoder 2.01.30.0


    Main Concept MPEG Encoder 1.04.02.00


    QuEnc 0.59 Beta 2




    I-Frame ZOOM

    SOURCE


    Cinema Craft Encoder 2.70.01.05


    TMPGENC XPRESS 3.0.4.24


    Canopus ProCoder 2.01.30.0


    Main Concept MPEG Encoder 1.04.02.00


    QuEnc 0.59 Beta 2




    B-Frame

    SOURCE


    Cinema Craft Encoder 2.70.01.05


    TMPGENC XPRESS 3.0.4.24


    Canopus ProCoder 2.01.30.0


    Main Concept MPEG Encoder 1.04.02.00


    QuEnc 0.59 Beta 2




    B-Frame ZOOM

    SOURCE


    Cinema Craft Encoder 2.70.01.05


    TMPGENC XPRESS 3.0.4.24


    Canopus ProCoder 2.01.30.0


    Main Concept MPEG Encoder 1.04.02.00


    QuEnc 0.59 Beta 2



    B-Frame

    SOURCE


    Cinema Craft Encoder 2.70.01.05


    TMPGENC XPRESS 3.0.4.24


    Canopus ProCoder 2.01.30.0


    Main Concept MPEG Encoder 1.04.02.00


    QuEnc 0.59 Beta 2




    B-Frame ZOOM

    SOURCE


    Cinema Craft Encoder 2.70.01.05


    TMPGENC XPRESS 3.0.4.24


    Canopus ProCoder 2.01.30.0


    Main Concept MPEG Encoder 1.04.02.00


    QuEnc 0.59 Beta 2




    DIRECT LINKS:

    FRAME 000 I

    http://img178.exs.cx/img178/4267/cce270frame000i8ef.jpg
    http://img178.exs.cx/img178/2830/mcframe000i8sz.jpg
    http://img178.exs.cx/img178/1373/procoder2frame000i9po.jpg
    http://img178.exs.cx/img178/9434/quencframe000i8ar.jpg
    http://img178.exs.cx/img178/9595/sourceframe0002fl.jpg
    http://img178.exs.cx/img178/7541/tmpegnc3frame000i4ea.jpg

    FRAME 000 I ZOOM

    http://img178.exs.cx/img178/7009/cce270frame000izoom4rw.jpg
    http://img178.exs.cx/img178/8544/mcframe000izoom9it.jpg
    http://img178.exs.cx/img178/6816/procoder2frame000izoom9vo.jpg
    http://img178.exs.cx/img178/884/quencframe000izoom1qt.jpg
    http://img178.exs.cx/img178/1051/sourceframe000zoom8ve.jpg
    http://img178.exs.cx/img178/4359/tmpegnc3frame000izoom4wm.jpg

    FRAME 160 B

    http://img178.exs.cx/img178/6568/cce270frame160b8na.jpg
    http://img178.exs.cx/img178/63/mcframe160b0dg.jpg
    http://img178.exs.cx/img178/1282/procoder2frame160b9ut.jpg
    http://img178.exs.cx/img178/6883/quencframe160b9hq.jpg
    http://img178.exs.cx/img178/222/sourceframe1605ez.jpg
    http://img178.exs.cx/img178/1452/tmpegnc3frame160b7bt.jpg

    FRAME 160 B ZOOM

    http://img178.exs.cx/img178/2576/cce270frame160bzoom8qv.jpg
    http://img178.exs.cx/img178/6302/mcframe160bzoom5bx.jpg
    http://img178.exs.cx/img178/1885/procoder2frame160bzoom5df.jpg
    http://img178.exs.cx/img178/8178/quencframe160bzoom9yr.jpg
    http://img178.exs.cx/img178/5402/sourceframe160zoom5je.jpg
    http://img178.exs.cx/img178/4998/tmpegnc3frame160bzoom5kz.jpg

    FRAME 512 B

    http://img172.exs.cx/img172/6008/cce270frame512b5vi.jpg
    http://img172.exs.cx/img172/7599/mcframe512b6oh.jpg
    http://img172.exs.cx/img172/7192/procoder2frame512b9xk.jpg
    http://img172.exs.cx/img172/41/quencframe512b5cj.jpg
    http://img172.exs.cx/img172/6446/sourceframe5122ji.jpg
    http://img172.exs.cx/img172/6527/tmpegnc3frame512b7io.jpg

    FRAME 512 B ZOOM

    http://img172.exs.cx/img172/6339/cce270frame512bzoom4au.jpg
    http://img172.exs.cx/img172/404/mcframe512bzoom7yx.jpg
    http://img172.exs.cx/img172/3083/procoder2frame512bzoom7mf.jpg
    http://img172.exs.cx/img172/5096/quencframe512bzoom5to.jpg
    http://img172.exs.cx/img172/6605/sourceframe512zoom2wh.jpg
    http://img172.exs.cx/img172/5527/tmpegnc3frame512bzoom0bj.jpg


    Ok, now is up to you to judge who got better here!


    TIP: Copy all the pictures on the same folder, open ACDSEE and go changing the pictures, so you can really compare then.


    Tkz ppl,
    []'
    Simps
    Quote Quote  
  2. Hmmm, nice work...
    Quote Quote  
  3. Member Soopafresh's Avatar
    Join Date: Jan 2004
    Location: United States
    Search Comp PM
    GREAT ! Thank you very much for doing this. Lots of information there. I'm impressed by Quenc. Not the fastest, but the free-est...
    Quote Quote  
  4. Video Restorer lordsmurf's Avatar
    Join Date: Jun 2003
    Location: Want my advice? PM me.
    Search Comp PM
    Interesting, but...

    1. Still clips don't make for great tests.
    2. XVID doesn't make for great test source.

    Notes:
    - MC is blurry at default. Supposedly some tweaks can fix it, and all the version of MC (Adobe, Sony, Ulead, etc) perform differently.
    - CCE is known to add noise.
    - Procoder has several settings. It is believed by some that MASTERING is sometimes not as good as HIGHEST in the new versions. Unlike the old days of 1.x versions.

    Overall interesting stuff.
    Quote Quote  
  5. Tkz Guys!
    Gave my best!

    In my opinion, procoder chances the luminance/color of the source, and by doing that, sometimes it will miss some details on the frame.

    But still, the playback of the procoder is VERY PRETTY.

    It is my favortite. You can easily see that the procoder has the most pretty color on every frame.

    []'
    Simps
    Quote Quote  
  6. Well, I like Procoder v1.5 best. I've also done some encoding tests
    with other sources (other than DV, cause we know now, that DV is good w/ PC)
    .
    Those tests I've done with PC v1.5 and Movie/Film source has also come out
    pretty good too. I haven't had time to fully work PC, but I'm slowly doing
    so. At the moment, I'm in the "tweaking" phase of PC, and my prefered
    method of encoding w/ PC is through frameserving it from vdub. I have my
    reasons
    .

    I agree w/ LS. Xvid/divX do not do tests justice, of this magnitude. I can
    spot "square" blocks a mile away w/ these codecs. (though I haven't seen
    your's because I'm on dialup, and I just opended this page, and its still D/L'ng
    your pics and things
    ) Anyways. An MPEG is about the only true and non-tainting
    method of demonstrating quality. Space (webspace) is always an issue, and I understand.
    So please expect to here critisism from some, because of this.

    I appreciate your hard work though, cause I know you spent a great deal of time
    with it. Cheers.

    Well, it's very late for me, and I'm pooped. Later peoples.
    -vhelp 3057
    Quote Quote  
  7. Banned
    Join Date: Oct 2003
    Location: Americas
    Search Comp PM
    Notwithstanding the source, if encoder can import the picture it should encode it properly...

    as with ver 1.5 Procoder does apply smoothing and enhances contrast in ver. 2 as well - looks good, some loss of detail in shadows and high contrast areas (loss of detail on glass roof structure), enhances color as well (no surprise), gives photographically "hard look", nice overall, with some tweaking will be the smoothest, best overall looking picture of all (all Canopus products, also hardware based are giving similar "enhanced " look) very good overall, needs tweaking for best - truest output ( that is a plus for "film look" but minus when reencoding a ready film product, seems to be most intrusive of all encoders), love it or hate it

    TMPGEnc, very true to the original, sharp, excellent color balance, some minor artifacting and unfortunately minimal macroblocking, very good picture though (I think best overall - true to the source)

    CCE, close to the TMPGEnc, softer though, less artifacts, on par with TMPG

    MC, v. good balance, minimal "mosquito" artifacts, appears sharper then CCE, v. close to TMPEG

    QEnc, very good indeed, least sharp through but also not much artifacting, slow (but very promising)

    Summarizing, I'm surprised how close the game became. You need to choose what type of distortion you tolerate the best and make your selection. I pick the encoder based on the job requirements but admit that MC is the one I use the most. For DV material I use Prododer though.
    Quote Quote  
  8. I'm a MEGA Super Moderator Baldrick's Avatar
    Join Date: Aug 2000
    Location: Sweden
    Search Comp PM
    Nice comparison...good for all those who converts from xvid/divx sources.

    Why 10passes with CCE?
    Quote Quote  
  9. Member rhegedus's Avatar
    Join Date: Sep 2002
    Location: on the jazz
    Search Comp PM
    How about some motion shots?
    Regards,

    Rob
    Quote Quote  
  10. Originally Posted by Baldrick
    Nice comparison...good for all those who converts from xvid/divx sources.

    Why 10passes with CCE?

    Tkz!
    I was looking for best possible quality, and with cce i could choose up to 99 passes, with the other encoders, only 2 passes. Since cce with 2 pass was too much faster then the others with same 2 pass, i choose to make it a 10 pass, so the encode time would still be reasonable, when compared against the others, and of course, quality would the max possible.



    Originally Posted by rhegedus
    How about some motion shots?

    This actually were motion shots. Quality on the source (aprox 4k avg bitrate for xvid) is very high, as the mpeg2 encoders too, so they dont even seen to be motion at all. But of course, it was not a very high motion scene, like a fast complex car chase thing.
    I will make a very complex motion scene, and update the test here, but you will see, that the results will be pretty much the same.


    And also, i might give procoder 1.5 a shot on this
    Would be cool!

    Tkz!
    []'
    Simps
    Quote Quote  
  11. With 10 passes, your mpeg-2 file will be like more a CBR encode than a VBR encode.
    Quote Quote  
  12. Banned
    Join Date: Oct 2003
    Location: Americas
    Search Comp PM
    How about throwing in Vegas and NeroVision 3 (if possible)? Also, you may post uncompressed original source frames to reference Xvid.
    Quote Quote  
  13. Originally Posted by cd090580
    With 10 passes, your mpeg-2 file will be like more a CBR encode than a VBR encode.
    I dont get it.
    I always thought that the more passes, the better bit distribuction, and so, better image quality.
    At least, thats what my cce user guide say.

    And a CBR encode @ 5000bitrate, will not look as good as VBR encode with avr=5000, max=9500.

    So by doing 10 pass, i thought i would get the best possible output.

    But maybe i am wrong, can someone help me on this?

    Tkz,
    []'
    Simps
    Quote Quote  
  14. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    without motion shots -- doesnt tell me much --


    using xvid as a source is really not saying much ... as all look very soft and full of artifacts ...

    but good job otherwise ... (dont get me wrong -- its a lot of work and you did a good job)

    10 passes for cce is a waste of time -- 3 is ussually enough except at very low bit rates , 4 and 5 maybe ..
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  15. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    9500 is high for dvd on a burned disk ... specially since some encoders spike out ...

    CCE 2.70 has a new GUI and more features ..

    for XVID/DIVX - you might want to try 4:1:1 interpolation


    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  16. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    Originally Posted by cd090580
    With 10 passes, your mpeg-2 file will be like more a CBR encode than a VBR encode.
    nope - not at all
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  17. Ok!
    I hope there are a lot of people out there that 90% of the encodes they do, are from .avi (divx, xvid, etc) to dvd! just like me
    So then, maybe this test can help

    []'s
    Simps
    Quote Quote  
  18. Sorry, double post.
    deleted

    []'
    Simps
    Quote Quote  
  19. Originally Posted by BJ_M
    9500 is high for dvd on a burned disk ... specially since some encoders spike out ...

    CCE 2.70 has a new GUI and more features ..

    for XVID/DIVX - you might want to try 4:1:1 interpolation

    Well, this is what my cce 2.7 manual tells me about interpolation:

    __________________________________________________ ______

    2.18 4:1:1 → 4:2:2 interpolation
    NTSC DV uses the 4:1:1 colorspace. When converting it into 4:2:0
    which is used in MPEG, Cinema Craft Encoder SP will receive 4:2:2
    data from the codec, and convert it into 4:2:0. However, the 4:2:2 data
    is not always interpolated by DV codecs because not all of the DV
    codec will interoplate. If Cinema Craft Encoder SP down-converts
    non-interpolated data into 4:2:0, the actual data will become 4:1:0
    and lose more chrominance information. With this option, Cinema
    Craft Encoder SP will interpolate it to 4:2:2 before converting into
    4:2:0.

    ☞ PAL DV uses the 4:2:0 colorspace, so it does not need 4:2:2
    interpolation.

    ☞ If you apply this option to the codec which interpolates to 4:2:2,
    it may cause a loss of the quality of image.


    __________________________________________________ ______


    I dont quite get it.
    Can someone tell me if checking that 4:1:1 -> 4:2:2 setting, will really help to encode xvid?

    I get the feeling that that shouldnt be checked
    But, im here to learn too!

    Tkz!
    []'
    Simps
    Quote Quote  
  20. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    i suggested to try it -- try it on a high color piece --

    the sample i posted in the cpu speed test section is perfect for this (it is divx)

    http://www.videohelp.com/forum/images/guides/p1166252/part1.exe

    http://www.videohelp.com/forum/images/guides/p1166252/part2.rar
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  21. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    what i mean about the above clips looking soft -- is that they look as they were high compression from dvd and then re-compressed back to mpeg, but he source and or compression wasnt that good to start with ..

    here is a shot i made that came from hdcam video - i compressed it to dvd mpeg2 , took that and compressed to xvid and then re-encoded back to mpeg2 for dvd with procoder 2 at default settings mastering ...

    this is a B frame in both cases - zoom in around the clowns mouth and you can see the diff. but still it doesnt go all soft and flat ..

    mpeg2 before:




    ----------------------------------------------------------------------------
    converted to xvid (4000 bit rate, mpg quant.)
    then encoded to mpeg2 dvd byprocoder 2



    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  22. Good looking image there!
    Very high quality source indeed.

    But what can i say, encoders are not only to be used on sources of the highest possible quality.

    You will see that many many people, dont really make the source video. They just get it from here and there!

    And then, u wont really have the option to start with the best possible source. But still, encoders need to be as close as possible to the source, no matter is the source is very high quality, or good quality, or avarage, etc.

    So, like i said, there are a lot of people encoding .avi (already compressed) over here! So, their sources wont be diferent from the ones i have posted.

    So, for this people (including me), that loves to encode .avi to dvd, i guess this thread will help very much on the encoder of choice!

    Tkz!
    []'
    Simps
    Quote Quote  
  23. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    yea - you are right ...

    a lot of people seem to be making dvd's from xvid / divx sources instead of source material or even the original dvd ....


    for whatever reason ......
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  24. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    but in that case -- an important factor is two (of several things)

    encoding to mpeg4 crushes blacks --

    resizing becomes a important issue -- as in how to ...
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  25. Originally Posted by BJ_M
    but in that case -- an important factor is two (of several things)

    encoding to mpeg4 crushes blacks --

    resizing becomes a important issue -- as in how to ...

    Yeah, you are right.

    The resize method im using is the bicubic avisynth thingy.

    Is there a better way to do that? Is lanczos3/4 better?
    I always wanted to know what is the best quality resize settings.
    I dont even know if there is any new resize filter that can top out the ones in avisynth 2.5
    Maybe you can help me.

    What are the best way to do it in your opinion?

    Tkz,
    []'
    Simps
    Quote Quote  
  26. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    lanczos3 is MUCH better
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  27. Originally Posted by BJ_M
    lanczos3 is MUCH better
    in avisynth it's the Lanczos4resize(int width,int height)
    My AVI -> Any Format Guide is available here.
    My Frame Resize Calculator (enhanced for Virtualdub) is available here
    Quote Quote  
  28. Член BJ_M's Avatar
    Join Date: Jul 2002
    Location: Canada
    Search Comp PM
    Lanczos 3 or 4 -- really not much diff. Mitchell is also very good .. both of these much better than bicubic ..

    BUT - for text or hard edges, bilin. and bicubic can often look better, specially downsizing
    "Each problem that I solved became a rule which served afterwards to solve other problems." - Rene Descartes (1596-1650)
    Quote Quote  
  29. Originally Posted by BJ_M
    Lanczos 3 or 4 -- really not much diff. Mitchell is also very good .. both of these much better than bicubic ..

    BUT - for text or hard edges, bilin. and bicubic can often look better, specially downsizing
    I thought for downsizing bilinear was the way to go? And that bilinear was great if only stretching/skewing one dimension? I could be wrong (it's happened before )
    My AVI -> Any Format Guide is available here.
    My Frame Resize Calculator (enhanced for Virtualdub) is available here
    Quote Quote  
  30. Member monzie's Avatar
    Join Date: Nov 2003
    Location: The Village
    Search Comp PM
    For resizing filters (lanczos, bicubic etc) why not just preview the .avs file and LOOK to see if it looks OK?

    Virtualdub opens .avs files (scripts) and so does many software video players (i use Zoomplayer).

    Personally I find lanczos is slightly better suited to resizing (upwards) xvids (etc) as (in my eyes) it tends to 'sharpen' the picture (which is normally required when upsizing) whilst bicubic is slightly softer (in comparison)....but, to be perfectly honest I cant see THAT much difference..so i tend to stick with lanczos....and dont worry about it.

    As for divx/xvids, my own findings are that these compression methods tend to destroy fine detail in areas of similar shades/colours...for example video footage with 'sky' as the backdrop, or areas of video swathed in shadows (but these areas are usually NOT the focal point of the footage so your 'eyes and brain' tend to overlook the missing detail..unless looking for it)...or in the original posters case the leaves on the trees and the 'rings' on the tree trunk (presuming of course that that detail was present in the original source).

    Because of that missing detail (in say a 'typical' xvid movie back-up of say sub 1000kbs) the amount of bitrate required to convert the file to mpeg2 can be reduced by quite a lot....typically to around 3X the original files bitrate, so a 1000kbs xvid MAY only require a 3000kbs ave MPEG2, and (again my own findings) pumping too much bitrate (for the mpeg2 encoding) into a xvid/divx achieves ZERO/ZILCH/NOTHING as a PLATEAU will have been reached at some point......which unfortunately for SIMPS is the probable reason that all your test results look so similar....those bitrates you set are probably way more than is required from that source.

    Heres a link to my OPTIMAL guide, wether or not you agree with what I say is up to you, but I'd like to point you to the % samples found within the guide....... encoding a 1% or 2% sample will not only save you hours (to view your output) when encoding x/d's video back to DVD (or for TMPG possibly days....) but will give you a damn good idea of the quality of the OVERALL output at any given bitrate on any encoder at any settings, in around 5 minutes (or less)...pretty useful eh?

    http://www.videohelp.com/forum/viewtopic.php?t=255103

    PS. Also, dont forget that a straight DVD to x/d (avi) conversion (NTSC and most PAL..very few people go UP to the correct avi v TV pixel ratio) will BY DEFAULT lose you around 10% of the horizonatal resolution (and therefore quality as your squeezing detail into LESS pixel space..ie 640W from source)...and source to samller res conversions (eg DVD to 512 X 384 lose even more of the original detail BEFORE the ACTUAL compression)...in other words...any 'typical' avi conversion will degrade quality by default...then some more.
    Quote Quote  



Similar Threads