VideoHelp Forum

Try DVDFab and download streaming video, copy, convert or make Blu-rays,DVDs! Download free trial !
Closed Thread
Page 2 of 2
FirstFirst 1 2
Results 31 to 59 of 59
Thread
  1. Originally Posted by poisondeathray View Post
    I still have Mainconcept, Sony, Apple encoders etc... If you want me to whip some new ones up or post old ones, just say the word
    I would love it! Here's the catch, do not use any of the tired "standard" test files, like that park joy or ducks taking off or anything like that.

    Download all the Net FliX uncompressed sources, about 300GB total, and combine them into a single test source or the NetFlix Meridian test sample from Xiph's and create a 500mb test encode using all the encoders you have available and post the settings you used.

  2. Dinosaur Supervisor KarMa's Avatar
    Join Date
    Jul 2015
    Location
    US
    Search Comp PM
    Originally Posted by sophisticles View Post
    I would love it! Here's the catch, do not use any of the tired "standard" test files, like that park joy or ducks taking off or anything like that.
    The catch with you is also going to require 100Mbit encodings so that you can say "see guys X is just as good as Y".

  3. Originally Posted by sophisticles View Post
    Originally Posted by poisondeathray View Post
    I still have Mainconcept, Sony, Apple encoders etc... If you want me to whip some new ones up or post old ones, just say the word
    I would love it! Here's the catch, do not use any of the tired "standard" test files, like that park joy or ducks taking off or anything like that.

    Download all the Net FliX uncompressed sources, about 300GB total, and combine them into a single test source or the NetFlix Meridian test sample from Xiph's and create a 500mb test encode using all the encoders you have available and post the settings you used.


    300GB of all the clips ? LOL I guess don't say the word. It's way too much work and bandwidth. It would take weeks/months to do encoding tests properly with a bunch of metrics and bitrate ranges.

    This Mainconcept AVC/ Apple AVC/ Sony Vegas AVC has been settled a long time ago by tests and posts on several forums - with different types of sources, different genres. I posted some on this forum as well. Go back 8-12 years. I even corrected for some people's encodes using MC because they weren't using ideal settings. Sony Vegas AVC does not even have b-frames - that should tell you something.

    I can't tell if you're being serious . But if you are and really didn't know about this - I'll try to whip something up (without ducks or parks) in another thread . Maybe throw in that updated SIF .

  4. I'll make it easier for you:

    https://www.blackmagicdesign.com/products/blackmagicpocketcinemacamera/workflow

    Download the 4k and 6k ProRes files available, use those as your sources.

  5. Originally Posted by sophisticles View Post
    I'll make it easier for you:

    https://www.blackmagicdesign.com/products/blackmagicpocketcinemacamera/workflow

    Download the 4k and 6k ProRes files available, use those as your sources.


    In some ways this is not "easier", because typical Prores is 10bit422 . And AVC is a pretty "stupid" choice for 4k and 6k

    The algorithm and method used for chroma downsampling, bit depth conversion comes into play (e.g. do you dither? if yes, what algorithm) . Those variables should be controlled and consistent.

    For example, Sony Vegas AVC , and Apple AVC don't have 10bit AVC, nor 4:2:2 . (Newer Vegas has XAVC, but not at typical consumer end delivery bitrates)

    When you do tests, you have to factor in a bunch of things in the test design . Have you even done this before?

    Or do you just want me to make some decisions and go with it ?

  6. Originally Posted by poisondeathray View Post
    Originally Posted by sophisticles View Post
    I'll make it easier for you:

    https://www.blackmagicdesign.com/products/blackmagicpocketcinemacamera/workflow

    Download the 4k and 6k ProRes files available, use those as your sources.
    In some ways this is not "easier", because typical Prores is 10bit422 . And AVC is a pretty "stupid" choice for 4k and 6k

    The algorithm and method used for chroma downsampling, bit depth conversion comes into play (e.g. do you dither? if yes, what algorithm) . Those variables should be controlled and consistent.

    For example, Sony Vegas AVC , and Apple AVC don't have 10bit AVC, nor 4:2:2 . (Newer Vegas has XAVC, but not at typical consumer end delivery bitrates)

    When you do tests, you have to factor in a bunch of things in the test design . Have you even done this before?

    Or do you just want me to make some decisions and go with it ?
    And so the excuses begin.

    I think you have proven my point about those that spew the gospel of x264 blindly.

    To address some of your "points":

    AVC is a pretty "stupid" choice for 4k and 6k
    So all the professional grade cameras from Panasonic, Canon, JVC that record straight to AVC-Intra at resolutions up to 4k are recording to a "stupid" format? I'm glad you are not in charge of these companies.

    As for the idiotic, insipid, purposely obtuse "objections" concerning ProRes, let's assume you actually knew what you were doing and actually made a living as a videographer for weddings, corporate events, maybe shooting a low budget movie ala the Blair Witch Project or shooting a documentary, you would choose one of these cameras and either shoot in Black Magic RAW or if you wanted to save space you would choose one of the ProRes options, and yes the files would be 10-bit 422, but when the pro in question is preparing the deliverable he/she is going to have to encode to 8-bit 420, most likely AVC, because he/she needs the deliverable to be viewable on as many platform as possible.

    You know this but you don't care, you don't give a flip about testing an actual scenario, you care about structuring a test that you can manipulate and massage the results so that you can "prove" that x264 is the "best" option.

    This is why all these "tests" that supposedly show x264's "superiority" are extremely flawed, invariably they take a small 30 second or so clip, usually of some contrived scene that bears little resemblance to anything you will actually see in the big screen, they encode it with that silly CRF setting at ridiculously low bit rate and then encode with the other encoders using the bit rate x264 chose. And to people like you is somehow "scientific", you wouldn't know "scientific" if someone beat you over the head with a chemistry book.

    I have never seen an actual. real life, scenario, with actual footage like you would deliver to a client, where x264 won handily, never. Take any commercial grade footage, something that would be a master, and encode the whole thing using a 2 pass bit rate based setting, for instance NetFlix uses about 15 mb/s for it's 4k (people still complain about the quality) and replicate a deliverable or a consumable end user destined product.

    So if you really want to show you have any intellectual honesty, pretend I hired you to shoot and prepare content for me that I am going to put on a website and sell to subscribers. I ask you to give me samples of your work, post the samples you would show me to convince me to hire you for the job. Give me your best x264, Main Concept, Sony AVC and Apple H264 product.

    Or you can just keep making excuses, which is what I suspect you will choose to do.

  7. Originally Posted by sophisticles View Post
    I think you have proven my point about those that spew the gospel of x264 blindly.
    How so? You have provided zero proof . Where is the proof ?

    AVC is a pretty "stupid" choice for 4k and 6k
    So all the professional grade cameras from Panasonic, Canon, JVC that record straight to AVC-Intra at resolutions up to 4k are recording to a "stupid" format? I'm glad you are not in charge of these companies.
    Do you understand the difference between acquisition and delivery ? Apparently not

    It shows your complete lack of understanding

    What type of bitrates do you think they use?

    Do you think AVC Intra, Ultra, I-frame XAVC is used for end delivery ? LOL




    As for the idiotic, insipid, purposely obtuse "objections" concerning ProRes, let's assume you actually knew what you were doing and actually made a living as a videographer for weddings, corporate events, maybe shooting a low budget movie ala the Blair Witch Project or shooting a documentary, you would choose one of these cameras and either shoot in Black Magic RAW or if you wanted to save space you would choose one of the ProRes options, and yes the files would be 10-bit 422, but when the pro in question is preparing the deliverable he/she is going to have to encode to 8-bit 420, most likely AVC, because he/she needs the deliverable to be viewable on as many platform as possible.
    Exactly! We are not testing acquisition bitrates. We are testing end deliverables, low bitrates. AVC-Intra etc.. isn't used for that


    But how do you get from 10bit 422 to 8bit 420 ? It's a quagmire full of potential issues and differences. That's what I'm talking about.

    If you just plug in prores to Sony Vegas or Apple , you're going to get differences . For example , Apple uses a lower quality chroma downsampling method. It would be unfair

    If one method dithers down with floyd-steinberg, but another truncates bits, there are going to be differences.

    Obviously you don't know the details and have never done proper tests before. You need to control all these factors, otherwise one encoder might get an unfair advantage. This is basic science 101



    I know exactly what I'm doing, and this prooves that you don't know what you're doing

    You don't know even how to use the tools, or the proper workflow , or proper testing technique.

    Remember when you posted wrong color ducks ?

    Do you even know what bit depth and chroma subsampling are ? You don't even know how to process video !

    You "claim" to do a bunch of tests, but they are of limited value because of errors in the procedure and lack of scientific method





    You know this but you don't care, you don't give a flip about testing an actual scenario, you care about structuring a test that you can manipulate and massage the results so that you can "prove" that x264 is the "best" option.

    This is why all these "tests" that supposedly show x264's "superiority" are extremely flawed, invariably they take a small 30 second or so clip, usually of some contrived scene that bears little resemblance to anything you will actually see in the big screen, they encode it with that silly CRF setting at ridiculously low bit rate and then encode with the other encoders using the bit rate x264 chose. And to people like you is somehow "scientific", you wouldn't know "scientific" if someone beat you over the head with a chemistry book.
    How so? I'm asking for clarification

    Do you think a test that shows Sony or MC AVC low quality mush will suddenly get better at 60 seconds ? 1 hours? 3 hours ? Do you think a 1 min test that shows x264 doing well , will suddenly get bad at 2 min, or 30 min ?

    I agree single scene tests are problematic, but they have some value in showing problems of an encoder. Ideally you'd do at least a 10 minute test, but it takes too long to do proper tests . It's not just 1 encode. You need to do a series of encodes per encoder with different bitrates (and perhaps settings) to plot rd cuves


    10bit AVC is an unfair advantage for x264 and MC . If you tested those at 10bit, by other like SIF, Sony Vegas AVC are 8bit, against a 10bit prores as the "source", the later are penalized . 10bit is getting to be quite common now . YT supports it, browsers support 10bit VP9 natively, HEVC hardware chips / TVs support it. An 8bit codec like SIF is going to be a disadvantage .

    If you want to compare at 8bit 4:2;0 , then all those other questions come up . You need to keep a "level playing field", or "apples to apples" . I know how you love those sayings

    I'm also asking because otherwise you're going to say I "chose" this test, or did it this way or that. Isn't that obvious?

    If you choose it, there is no way you can wiggle out of this. I'm giving you the chance to chose the clip(s) and test design, as long as it's scientific and fair. You can even choose a clip where x264 is known to do poorly, I have some, and I purposely collect them .

    Nor can it be an hour long production, I just don't have the time. Has to be reasonable

    And it doesn't have to be 1 clip. I posted a variety of clips and different genres in the past. It's good to see how an encoder does in different situations. It might not be completed in 1 go , if I have time I will add to that thread gradually. It can be a mega thread - anyone can contribute and test, adding encodes from various encoders and settings .

  8. Originally Posted by poisondeathray View Post
    Originally Posted by sophisticles View Post
    I think you have proven my point about those that spew the gospel of x264 blindly.
    How so? You have provided zero proof . Where is the proof ?

    AVC is a pretty "stupid" choice for 4k and 6k
    So all the professional grade cameras from Panasonic, Canon, JVC that record straight to AVC-Intra at resolutions up to 4k are recording to a "stupid" format? I'm glad you are not in charge of these companies.
    Do you understand the difference between acquisition and delivery ? Apparently not

    It shows your complete lack of understanding

    What type of bitrates do you think they use?

    Do you think AVC Intra, Ultra, I-frame XAVC is used for end delivery ? LOL




    As for the idiotic, insipid, purposely obtuse "objections" concerning ProRes, let's assume you actually knew what you were doing and actually made a living as a videographer for weddings, corporate events, maybe shooting a low budget movie ala the Blair Witch Project or shooting a documentary, you would choose one of these cameras and either shoot in Black Magic RAW or if you wanted to save space you would choose one of the ProRes options, and yes the files would be 10-bit 422, but when the pro in question is preparing the deliverable he/she is going to have to encode to 8-bit 420, most likely AVC, because he/she needs the deliverable to be viewable on as many platform as possible.
    Exactly! We are not testing acquisition bitrates. We are testing end deliverables, low bitrates. AVC-Intra etc.. isn't used for that


    But how do you get from 10bit 422 to 8bit 420 ? It's a quagmire full of potential issues and differences. That's what I'm talking about.

    If you just plug in prores to Sony Vegas or Apple , you're going to get differences . For example , Apple uses a lower quality chroma downsampling method. It would be unfair

    If one method dithers down with floyd-steinberg, but another truncates bits, there are going to be differences.

    Obviously you don't know the details and have never done proper tests before. You need to control all these factors, otherwise one encoder might get an unfair advantage. This is basic science 101



    I know exactly what I'm doing, and this prooves that you don't know what you're doing

    You don't know even how to use the tools, or the proper workflow , or proper testing technique.

    Remember when you posted wrong color ducks ?

    Do you even know what bit depth and chroma subsampling are ? You don't even know how to process video !

    You "claim" to do a bunch of tests, but they are of limited value because of errors in the procedure and lack of scientific method





    You know this but you don't care, you don't give a flip about testing an actual scenario, you care about structuring a test that you can manipulate and massage the results so that you can "prove" that x264 is the "best" option.

    This is why all these "tests" that supposedly show x264's "superiority" are extremely flawed, invariably they take a small 30 second or so clip, usually of some contrived scene that bears little resemblance to anything you will actually see in the big screen, they encode it with that silly CRF setting at ridiculously low bit rate and then encode with the other encoders using the bit rate x264 chose. And to people like you is somehow "scientific", you wouldn't know "scientific" if someone beat you over the head with a chemistry book.
    How so? I'm asking for clarification

    Do you think a test that shows Sony or MC AVC low quality mush will suddenly get better at 60 seconds ? 1 hours? 3 hours ? Do you think a 1 min test that shows x264 doing well , will suddenly get bad at 2 min, or 30 min ?

    I agree single scene tests are problematic, but they have some value in showing problems of an encoder. Ideally you'd do at least a 10 minute test, but it takes too long to do proper tests . It's not just 1 encode. You need to do a series of encodes per encoder with different bitrates (and perhaps settings) to plot rd cuves


    10bit AVC is an unfair advantage for x264 and MC . If you tested those at 10bit, by other like SIF, Sony Vegas AVC are 8bit, against a 10bit prores as the "source", the later are penalized . 10bit is getting to be quite common now . YT supports it, browsers support 10bit VP9 natively, HEVC hardware chips / TVs support it. An 8bit codec like SIF is going to be a disadvantage .

    If you want to compare at 8bit 4:2;0 , then all those other questions come up . You need to keep a "level playing field", or "apples to apples" . I know how you love those sayings

    I'm also asking because otherwise you're going to say I "chose" this test, or did it this way or that. Isn't that obvious?

    If you choose it, there is no way you can wiggle out of this. I'm giving you the chance to chose the clip(s) and test design, as long as it's scientific and fair. You can even choose a clip where x264 is known to do poorly, I have some, and I purposely collect them .

    Nor can it be an hour long production, I just don't have the time. Has to be reasonable

    And it doesn't have to be 1 clip. I posted a variety of clips and different genres in the past. It's good to see how an encoder does in different situations. It might not be completed in 1 go , if I have time I will add to that thread gradually. It can be a mega thread - anyone can contribute and test, adding encodes from various encoders and settings .
    I'm done, you have successfully proven everything I have ever said about x264 worshipers.

  9. Originally Posted by sophisticles View Post

    I'm done, you have successfully proven everything I have ever said about x264 worshipers.
    Where's the proof? Do you understand what "proof" means ?

    Giving up so easily ?

    I've posted actual evidence.

    You have done nothing, proven nothing.

    I've posted a bunch of tests and many are posted elsewhere by others. Confirmed many times, many source, different genres


    All you've done time and time again is been proven wrong , and demonstrated your ignorance in video and testing methodology. Get rid of the grudge. You'll feel better

    I've offered to do more tests . But you can't put Sony or Apple at a disadvantage, everything has to be fair. You can even choose a"bad" clip for x264 and stack against it to farther your bias. I don't care


    The first valid, objective thing you've brought but in this thread is the clip length. That is a reasonable objection. But the likelihood of a longer test suddenly improving a bad encoder is low. You might be able to distribute bitrate from an adjacent low complexity section . But it doesn't stand to reason that the other encoders wouldn't do the same. But it's still reasonable and should be tested

  10. If it were up to me I would chose a consistent method and apply it to all

    For example, resize to 1920x1080 using spline36 . Chroma planes are resampled to 960x540 using bicubic. 10bit to 8bit using no dither (truncated) . It doesn't have to be those specifically , if you want another method, tell me. I don't want you complaining later about this or that

    That is the new "original" you test against. That is put through all the encoders .

    There are a bunch of other issues with Apple and Sony vegas. Only certain intermediates will be accepted for pass through. They will convert to RGB otherwise and there will be some quality loss , sometimes shifts in color like your "tests". This is not fair

    However , having done this before, and actually having real knowledge of tools being used, Fourcc IYUV for 8bit 4:2:0 uncompressed will work .




    My complaint would be this is a bit old school. In 2019 - 10bit is common . You shouldn't have to "dumb down" an encoder just to level the playing field. But those tests were done when 8bit 4:2:0 was the only distribution choice. 10bit is a big advantage for things like banding, where x264 is weak . Even MC benefits from it. Even from a 8bit source, 10bit can be beneficial at moderate bitrate ranges (it can be detrimental at low bitrate ranges). This is well known , there are papers, and I also posted tests before as well

    1920x1080 because nobody uses AVC for UHD end delivery today . YT uses VP9, Netflix HEVC . Even 1080 streams on Amazon, Netflix are showing up as HEVC. (Some are still AVC and VC1 on Netflix)

    If you're going to do a test, it would make more sense to test something useful. We're going a bit retro in an attempt to satisfy your grudge



    Is this reasonable ?

  11. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by poisondeathray View Post
    We're going a bit retro in an attempt to satisfy your grudge
    Is this reasonable ?
    That's really why I quit participating in the thread: it's moot.

    For MPEG, there is
    #1 MainConcept
    #2 some other commercials like CCE/Procoder/Cleaner/etc
    #3 all the freewares.

    Quality in that order 1-2-3, where 1 is best. There's really no arguing this on the cusp of the 2020s, long ago settled in 2000s.

    For H.264, it's also simple:
    - spec formats = MainConcept
    - home users, online sites (ie no specs, looser requirements) = x264
    - almost all other commercial apps, especially the Apple version, would dreadful looking

    Again, settled years and years ago in early 2010s.

    A thread like this is mostly letting others know the status quo, because they're either late to the game, or were hiding under a rock. Arguing history is dumb.

    I do agree dumbing down software for a "fair comparison" is often ... well, dumb.

  12. Originally Posted by lordsmurf View Post

    For MPEG, there is
    #1 MainConcept
    #2 some other commercials like CCE/Procoder/Cleaner/etc
    #3 all the freewares.

    Quality in that order 1-2-3, where 1 is best. There's really no arguing this on the cusp of the 2020s, long ago settled in 2000s.
    Not as clear-cut as the AVC case.

    CCE tends to do better than MainConcept MPEG2 for higher frequency details. It tends to preserve grain better (when used correctly), so for the progressive, theatrical DVD scenario I'd give it the edge.



    For H.264, it's also simple:
    - spec formats = MainConcept
    - home users, online sites (ie no specs, looser requirements) = x264
    - almost all other commercial apps, especially the Apple version, would dreadful looking
    x264 > Mainconcept for AVC , all spec formats including BD, and level restricted scenarios

    Plenty of tests demonstrating this too. None showing the opposite. Or if you have some evidence please post it. I've been looking for years for a single outlier test that demonstrates this

    Normally , encoders swap places; on some test one is a little better, for others the other one is a little better. This is not the case for AVC . MC is clearly near the middle of the pack for AVC, it's not recognized a top tier AVC encoder . Ateme looks to be slightly better than x264 in newer tests, but nobody seems interested in AVC anymore. All the focus is on next gen AV1 vs HEVC and upcoming VVC


    In 2010, there were concerns of x264 spec compliance, definitely. Not now in 2019.

    By 2016, x264 AVC streams passed all verification tests , all commercial verifiers. (Otherwise it wouldn't have been offered by companies such as Telestream)
    Last edited by poisondeathray; 10th Oct 2019 at 12:52.

  13. Originally Posted by poisondeathray View Post
    x264 > Mainconcept for AVC , all spec formats including BD, and level restricted scenarios
    Not according to your own tests.

  14. Originally Posted by sophisticles View Post
    Originally Posted by poisondeathray View Post
    x264 > Mainconcept for AVC , all spec formats including BD, and level restricted scenarios
    Not according to your own tests.
    Yes my own tests confirm what other tests show

    If there is deviation or interesting point, I'll mention it

  15. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by poisondeathray View Post
    (Otherwise it wouldn't have been offered by companies such as Telestream)
    That's probably more about licensing costs than anything else. Also, given how it's really just the 1 company (and just for their Episode encoder? which was never great anyway), it could be as simple as x264 fans in the decision-making loop. There are far more entities using MainConcept.

    Telestream also claims "x264 is best in independent tests", but I've never really seen an independent test. Almost all are conducted by x264 fans, or have an obvious pro-x264 bias. Most often trying to justify x264 is "just as good" or "better" than MC. And sometimes it is, sometimes it's not. It depends on use.

    What somewhat amuses me is the fact that, once upon a time, x264 was considered equal to MainConcept. Back in the 2000s, of course, being "best" encoder was all the rage, so "as good as" wasn't good enough. But to achieve "better", x264 had to allow settings that hardware players would balk at, thus out of spec. And yet x264 (more the fanboys, than developers) was suddenly being touted as better when it had to break formation to be better. In theory, that's great, outside the box thinking. But when it involves specs, nope.

    I did a lot of commercial encoding work from about 2008-2014, and x264 had issues. We couldn't use it. Seeing as how H.264 encoding is now passe tech, HEVC is in, I just don't see that as having changed much. So again, MainConcept for spec formats, x264 for spec-less. These days I use x264 far more than anything else, doing whatever settings I want, and can achieve both better size compression while keeping visual quality better. So don't mistake any of my comments as being anti-x264. I'm not.

    It is what is is. It ain't what it ain't.

    Originally Posted by poisondeathray View Post
    In 2010, there were concerns of x264 spec compliance, definitely. Not now in 2019.
    By 2016, x264 AVC streams passed all verification tests , all commercial verifiers.
    Sure, but you're also acting like MainConcept now is like MainConcept of a decade ago. It also was refined, improved. Most of the MC/x264 tests I've ever seen are old, or at least using an older version of MC.

    And if x264 wasn't really mature until 2016, it missed the party.

    poison, if you say x264 is now equal/better than MainConcept, I honestly don't have a problem believing you. But I do shake my head at the pointlessness of it all, seeing as how (1) it can't be "that" much better, and (2) we're talking about a now-legacy format. Again, I use x264 quite a bit for personal and web-streaming needs, and have MainConcept mostly for MPEG-2 and the random BD disc (as nobody wants BD these days, DVD is still in play, and streaming is king). 99% of my work is lossless. I've not run a MPEG2/H264 encoder test in the past 3 years, didn't see a point anymore.

  16. Originally Posted by lordsmurf View Post
    There are far more entities using MainConcept.
    Yes, the low end version, it's often bundled with their other codecs. The bundle pricing is attractive, and it's "safe". Ensured to work, full support.

    Telestream also claims "x264 is best in independent tests", but I've never really seen an independent test. Almost all are conducted by x264 fans, or have an obvious pro-x264 bias. Most often trying to justify x264 is "just as good" or "better" than MC. And sometimes it is, sometimes it's not. It depends on use.

    What somewhat amuses me is the fact that, once upon a time, x264 was considered equal to MainConcept. Back in the 2000s, of course, being "best" encoder was all the rage, so "as good as" wasn't good enough. But to achieve "better", x264 had to allow settings that hardware players would balk at, thus out of spec. And yet x264 (more the fanboys, than developers) was suddenly being touted as better when it had to break formation to be better. In theory, that's great, outside the box thinking. But when it involves specs, nope.
    Not equal, better

    Despite some people claiming MC was/is "better" there are zero tests, independent or otherwise that show this. Not a single one in any scenario, restricted, unrestricted, for mobiles, web, BD, low latency, live stream, etc... If there is evidence past or present - maybe an intelligent discussion could be started. Otherwise they risk becoming that "MC fanboy." Usually there is an outlier somewhere . If it were true it should be easy to find

    I haven't found it. And believe me, I've looked. If you can find problem scenarios, you can report them, and things get improved - for any encoder. In the process you also learn strengths/weaknesses of a particular encoder, and what settings might be appropriate for what scenarios. It's not all black and white, each encoder has strengths and weaknesses in certain areas.

    What typically gets posted are those unrestricted scenarios. But for everything posted, there are many tests not posted. Also there can be posting bias. eg. A test that show something or difference will more likely be posted than a test that show equivalency or nothing interesting. You have to consider all the evidence. When you do that, it's very strong in favor of x264


    And if x264 wasn't really mature until 2016, it missed the party.
    That's a good point, I don't know exactly when, but I'm certain ~2016 everything passed. For BD, one issue turned out to be certain verifier's problem. It passed Panasonic, Eclipse, Scenarist verifiers by 2012. Certain formats like "fake-interlace" "25p" with the Sony verifier are false reports. But people were hesitant to use x264 for commercial purposes and probably still are. And back then there was no LLC entity, no commerical support, or it was very new , in it's infancy .

    poison, if you say x264 is now equal/better than MainConcept, I honestly don't have a problem believing you. But I do shake my head at the pointlessness of it all, seeing as how (1) it can't be "that" much better, and (2) we're talking about a now-legacy format. Again, I use x264 quite a bit for personal and web-streaming needs, and have MainConcept mostly for MPEG-2 and the random BD disc (as nobody wants BD these days, DVD is still in play, and streaming is king). 99% of my work is lossless. I've not run a MPEG2/H264 encoder test in the past 3 years, didn't see a point anymore.
    I don't see a point either, when everything has already been proven. It is what it is.

    And it's exactly those restricted scenarios that matter. That's where you need the best compression because of bandwidth restrictions - be it VOD, Sat, Terrestrial broadcast, Internet delivery, etc... If you can save 20% of bandwidth costs, it makes a difference. That's why all the research into these new formats. UHD is going to be expensive transition USA , the new ATSC 3.0 and moving forward


    So it's partially to help out some anti-x264-dev person understand , and partially so we don't have to do these shenanigans again in the future with AV1/HEVC/VVC etc.... Just look at the tests, use your eyes. Look at the evidence. If there is a problem with the test design, fix it, redo , reevaluate. Be careful of what these metrics report and what they really show.

    Despite being labelled as a "x264 fan", I'm quite the opposite. I go out of my way to show problems. When x264 "won" with those old metrics 8-12 years ago, I cautioned people not to trust them, but to use their eyes.

    I call people out that misrepresent an encoder, x264 or otherwise . Just the other day in fact.
    https://forum.doom9.org/showthread.php?p=1886875#post1886875
    https://forum.doom9.org/showthread.php?p=1886879#post1886879

    I'm not saying there aren't x264 fan boys - I'm just not one of them. If I come across as a "fanboy" that suggests how strong the evidence is. I look at all the evidence, not just 1 test. I look methodically, scientifically, objectively and subjectively.

    Unlike anti-x264 zealots, I'm "anti-everything." I look for problems in everything. If there is a problem , I say there is a problem. If something is blurry, I'll say it's blurry. If x264 has banding in gradients, I'll say it has banding. I'm interested in the truth
    Last edited by poisondeathray; 11th Oct 2019 at 14:21.

  17. Member
    Join Date
    Dec 2012
    Location
    Germany
    Search Comp PM
    Originally Posted by poisondeathray View Post
    To ignore the mountains of scientific evidence of these facts and observational data is simply delusional . Like a "flat earther" or "anti-vaxxer" - it's a fixed false belief.
    there is no scientific evidence of the globe earth. it's a religion - space is a never proven. physically impossible fantasy world

    ignorant assumptions aren't scientific evidence

  18. Member Cornucopia's Avatar
    Join Date
    Oct 2001
    Location
    Deep in the Heart of Texas
    Search PM
    ...and we just went off the deep end!

    If you ever did any user-background reading, @Zero-11, you'd see that pdr does not make ignorant assumptions.

    Scott

  19. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by poisondeathray View Post
    Not equal, better
    No. It started out as "as good as", and then started to be touted as "better". Even some x264 fans were flummoxed when that started to happen. Remember, I've been around a while. I watched, but didn't much care. I wasn't about to use CLI, the x264 GUIs were all crap, and I was doing pro work anyway. It's only been in the past 5 years that I've really dug into x264 more, for personal use. I'm able to do things now that I wasn't then, though I still must be careful so as not to make my own hardware players balk.

    so we don't have to do these shenanigans again in the future
    When I read that, I had a sudden bust of LOL. A video utopia will never exist, too many strong arguing forces (streaming platforms, studios). I used to hate having to re-encode separate versions/encodes because of slight variations on spec lists. It was so pointless, overly anal.

    I'm not saying there aren't x264 fanboys - I'm just not one of them.
    Ah, good, I didn't think you were, even if you did see overly enthusiastic at times.

    Unlike anti-x264 zealots, I'm "anti-everything." I look for problems in everything.
    Then we're not that much different. I look for issues, in order to avoid or correct for them. In video restoration, that skill is a must. (I'm sometimes considered "too negative" by people that need coddling or constant reassurances, but I'm not their boyfriend or therapist -- a comment that generally pisses them off that much more, but that's just the raw truth of the matter, science/fact isn't there to preserve feelings.) What's ironic is that while it seems we disagree on x264, we're probably not that far apart. It really comes back to usage scenarios, and our likely use case differences, our experiences over the years, which have led top a slightly varied opinion.

  20. Originally Posted by lordsmurf View Post
    Originally Posted by poisondeathray View Post
    Not equal, better
    No. It started out as "as good as", and then started to be touted as "better".
    No, it actually started out as a featureless turd. Smoother than rmvb. Xvid was miles better. MPEG2 was better. I was one of those x264 critics. "sophi" would have gleefully recruited me and made me captain of the team then. The turning point was AQ and it's later optimization, that's when x264 suddenly got better than the others. No quality developments really in the last 5-6 years. Some minor tweaks, most code is towards instructions like optimizing for ARM . Project is basically dead, time to look forward



    so we don't have to do these shenanigans again in the future
    When I read that, I had a sudden bust of LOL. A video utopia will never exist, too many strong arguing forces (streaming platforms, studios). I used to hate having to re-encode separate versions/encodes because of slight variations on spec lists. It was so pointless, overly anal.
    But the discussion itself isn't pointless. Very interesting area right now with HEVC/AV1, SVT HEVC/AV1/VP9 variants, VVC coming

    And there will always be disagreements - "quality" is difficult to measure or quantify, perceptions are so varied.

    But you need to start with a decent test design. You can't mix a bunch of variables when want to test an encoder. Yet I get flak from sophi for trying to be scientific about it

    If you can present evidence, make observations, do tests, repeat, and discuss some more then that's a great learning environment. But that's the thing, I like evidence, preferrably not anecdotal or opinion, those are the lowest weight. Metrics aren't that useful either. VMAF is supposed to be better... but there are situations piling up where it's not much better for this round of models (or it's interpretation and intended usage is limited) . I need to examine the videos to see for myself

    Case in point - Turing HEVC video streams are finally being posted around for analysis. When Turing first came out, the Nvidia forum and some gaming sites posted some benchmarks, some screenshots, but not a single video sample . (Still not a single sample of video game source/encode...my theory is they are too busy, addicted to Fortnite). A lot of hoopla about it. Supposedly it was better than x265 slow . I'm not seeing it on the actual samples, it seems to be worse quality wise. Encodes are very clean - but that's just it - they edge on the side of too clean, loss of detail with blurring. Sure that's ok for some scenarios, but not ok for others. Granted x265 can be blurry too but that' s the default settings; you can make it retain high frequency detail too. NVEnc might not be able to, except at high bitrates. Just a few so tests far, so nothing conclusive. But the balloon sort of got deflated when real tests with actual video are trickling out. The quality is good overall and amazingly fast. Probably 20-30x faster than a typical desktop for 10bit HEVC for general use . It's just that achilles heel, that blurring of fine details. Hopefully they can improve that, or give the ability to tune for more detail , even if it's slower.

    And that's they lifecycle of encoder development. They tune for PSNR aka "BLURRY" early on . Only later on , when they mature, do they add perceptual features, psy, AQ , and at least have the ability or option to tune for details. x264 went through this, x265 is in the process right now, nearly graduated (if you recall 1.0 and earlier it was very blurry); hopefully Turing NVENC will make those gains too. Right now AV1 is supposed to be great too. Wins all the benchmarks, sometimes by a lot. IN truth, blurry blurry blurry. Sloooooow as hell. They got a long way to go

    Then we're not that much different. I look for issues, in order to avoid or correct for them. In video restoration, that skill is a must. (I'm sometimes considered "too negative" by people that need coddling or constant reassurances, but I'm not their boyfriend or therapist -- a comment that generally pisses them off that much more, but that's just the raw truth of the matter, science/fact isn't there to preserve feelings.) What's ironic is that while it seems we disagree on x264, we're probably not that far apart. It really comes back to usage scenarios, and our likely use case differences, our experiences over the years, which have led top a slightly varied opinion.
    Thanks for clearing that up , and that's perfectly ok to have different opinions. You can still discuss and compare tests, share user experiences, etc...

    Cheers
    Last edited by poisondeathray; 11th Oct 2019 at 20:50.

  21. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by poisondeathray View Post
    No, it actually started out as a featureless turd.
    Well, yeah, true, but I was only going back to late 2000s (07-09).

    The project changed too many hands, with the strangest being the Jason/Fiona sex-change weirdness (with the weirdness NOT even referring to the change itself, but the seeming meltdown, with things like Doom10 being collateral damage). Many (most? all?) of the devs were college/HS students, with nothing but time on their hands. The first real tout of "best" came from MSU that I can remember, and MSU has never been very reliable as an independent body. They always seem to have ulterior motives, with an extreme anti-payware/commercial slant ... except when they want to sell something.

    I always found this interesting: https://www.hmc.edu/non-wp-sites/old-news/steve-jobs-turns-to-hmc-student-for-tech-analysis.php
    Because my question was always this ... WTF did Steve Jobs know about video? Apple was very anti-video by the 2010s, with their own Apple H.264 codec probably being the worst codec of that era. In fact, what did Jobs know about tech in general in the 2010s? He was the face of Apple, but had no real meaty technical skills to go with it. So you'd read a piece like that, and at first glance, "oh wow". Then upon deeper reflection "hey, wait a second..."

  22. Originally Posted by lordsmurf View Post
    The project changed too many hands,
    Not really "changed hands"; it's an open source project. There are senior developers who "rubber stamp" commits from other people. Fiona didn't start out as senior, nor was she the only one during that time

    There are pros/cons to having same person, vs. many people, vs. rotating door to leading a project. FFMpeg had the lead same guy (Michael Niedermayer) for a long time - but some people have had issues with that. It's one reason Libav exists as separate but almost identical project to FFMpeg. Seems like a big waste of duplicated resources to me; when pooled together could do a lot more. In the end people are people, and some have personalities that don't mix well with others.

    with the strangest being the Jason/Fiona sex-change weirdness (with the weirdness NOT even referring to the change itself, but the seeming meltdown, with things like Doom10 being collateral damage).
    Fiona was always resistant to any constructive criticism of the project. For example, when presented clear evidence of fade issues (many times, many sources), she'd hum and haw and say no it's fine. No, it looks fine, fix your monitor. Or "don't worry when weightp is committed, that will fix it" (it helped to a small extent, but still problems). Finally she succumbed , and wrote a fade compensate patch , but it's not in the main branch (it's available as patch, some unofficial builds). It's helps a bit, but to this day x264 has problems with fades. The point of the story is - one of the main benefits of an open source project is potentially big user base, lots of feedback. (Yes some feedback is garbage, not helpful, and you need to filter it out). But if your personality or ego gets in the way of seeing what's right in front of you, that helps nobody. Still, the project benefited greatly from her contributions, nobody will dispute that. Many developers of various projects tend to be an eccentric bunch on the whole; sometimes you have to tread lightly and handle them delicately


    The first real tout of "best" came from MSU that I can remember, and MSU has never been very reliable as an independent body. They always seem to have ulterior motives, with an extreme anti-payware/commercial slant ... except when they want to sell something.
    Maybe, but not useful that useful, especially the early ones. I think they were PSNR/SSIM only. If people actually bothered to look at encodes instead of a pretty graph or chart they would see this. But it probably wasn't until a few years later that the actual quality could be considered "best." AQ was the largest contributing factor. AQ was being tuned those years , and it took a long time. And AQ - the concept - isn't an x264 "invention". Ateme had both psy and AQ long beforehand, and xvid had variations on the AQ concept with lumimasking. But the actual implementation and tuning specific for x264 is what put it over the top.

  23. Video Restorer lordsmurf's Avatar
    Join Date
    Jun 2003
    Location
    dFAQ.us/lordsmurf
    Search Comp PM
    Originally Posted by Zero-11 View Post
    Originally Posted by poisondeathray View Post
    To ignore the mountains of scientific evidence of these facts and observational data is simply delusional . Like a "flat earther" or "anti-vaxxer" - it's a fixed false belief.
    there is no scientific evidence of the globe earth. it's a religion - space is a never proven. physically impossible fantasy world
    ignorant assumptions aren't scientific evidence
    - If you're just taking the piss, then unfortunately it was a comedy fail.
    - If you're serious, STFU. (I'm always amazed at how some people can fit computer screens up their rectum, seeing as how their head is up their ass.)

    Originally Posted by poisondeathray View Post
    Many developers of various projects tend to be an eccentric bunch on the whole; sometimes you have to tread lightly and handle them delicately
    Well, I think there's eccentric, and then there's "WTF?", and that meltdown was the latter.

    Anyway, fun talk, but back to video...
    Last edited by lordsmurf; 12th Oct 2019 at 01:37.

  24. Member
    Join Date
    Dec 2012
    Location
    Germany
    Search Comp PM
    Originally Posted by lordsmurf View Post
    Originally Posted by Zero-11 View Post
    Originally Posted by poisondeathray View Post
    To ignore the mountains of scientific evidence of these facts and observational data is simply delusional . Like a "flat earther" or "anti-vaxxer" - it's a fixed false belief.
    there is no scientific evidence of the globe earth. it's a religion - space is a never proven. physically impossible fantasy world
    ignorant assumptions aren't scientific evidence
    - If you're just taking the piss, then unfortunately it was a comedy fail.
    - If you're serious, STFU. (I'm always amazed at how some people can fit computer screens up their rectum, seeing as how their head is up their ass.)
    So you can't even tell if I am serious which means you have no idea, you don't know the flat earth model and you don't even know your own model otherwise you wouldn't believe in the globe model.

  25. DECEASED
    Join Date
    Jun 2009
    Location
    Heaven
    Search Comp PM
    Originally Posted by lordsmurf View Post
    Originally Posted by Zero-11 View Post
    censored
    - If you're just taking the piss, then unfortunately it was a comedy fail.
    - If you're serious, STFU. (I'm always amazed at how some people can fit computer screens up their rectum, seeing as how their head is up their ass.)
    That's it, no free speech for trolls.
    users on my IgnoreList: 162 names thus far, but featuring DB83 and manono.

  26. Originally Posted by poisondeathray View Post
    Despite some people claiming MC was/is "better" there are zero tests, independent or otherwise that show this. Not a single one in any scenario, restricted, unrestricted, for mobiles, web, BD, low latency, live stream, etc... If there is evidence past or present - maybe an intelligent discussion could be started. Otherwise they risk becoming that "MC fanboy." Usually there is an outlier somewhere . If it were true it should be easy to find
    Are you not on speaking terms with yourself?

    You conducted and posted a test, using a portion of the Meridian test source, specially prepared for another user. You purposely tried to fix the test so that x264 would win, by using old, outdated, limited versions of Apple's AVC encoder, and old, outdated, versions of MC's encoder, with many of the quality improving options disabled, while enabling x264's "mastering" setting. You still ripped into the x264 encodes and when finally pressed to change the MC settings, you produced encodes that most people agree that Mc beat out x264 visually.

    It's there in that other thread, unless you decide to go back and edit your comments.

    The funniest, in a sad sense, part is that for years you, and people like you, have claimed that objective metrics can not be trusted to determine quality, that only your eyes can be trusted to make that determination. For years I have been saying on these boards that the job of an encoder is to reproduce the source as accurately as possible, not to make it look as good as possible and you and I, and a few other, have argued back and forth about this.

    When your own test results in an scenario where most agree that the MC encode looks better, you then adopt my stance nearly word for word, parroting my viewpoints (does {Polly want a cracker?) and standing on the results of an objective metric (VMAF) rather than what most people would agree is the nicer encode (MC).

    This is especially dishonest of you when you consider that you actually admit the x264 encode has significant problems.

    You are the typical x264 zealot.

  27. Originally Posted by sophisticles View Post
    Originally Posted by poisondeathray View Post
    Despite some people claiming MC was/is "better" there are zero tests, independent or otherwise that show this. Not a single one in any scenario, restricted, unrestricted, for mobiles, web, BD, low latency, live stream, etc... If there is evidence past or present - maybe an intelligent discussion could be started. Otherwise they risk becoming that "MC fanboy." Usually there is an outlier somewhere . If it were true it should be easy to find
    Are you not on speaking terms with yourself?

    You conducted and posted a test, using a portion of the Meridian test source, specially prepared for another user. You purposely tried to fix the test so that x264 would win, by using old, outdated, limited versions of Apple's AVC encoder, and old, outdated, versions of MC's encoder, with many of the quality improving options disabled, while enabling x264's "mastering" setting. You still ripped into the x264 encodes and when finally pressed to change the MC settings, you produced encodes that most people agree that Mc beat out x264 visually.

    It's there in that other thread, unless you decide to go back and edit your comments.

    The funniest, in a sad sense, part is that for years you, and people like you, have claimed that objective metrics can not be trusted to determine quality, that only your eyes can be trusted to make that determination. For years I have been saying on these boards that the job of an encoder is to reproduce the source as accurately as possible, not to make it look as good as possible and you and I, and a few other, have argued back and forth about this.

    When your own test results in an scenario where most agree that the MC encode looks better, you then adopt my stance nearly word for word, parroting my viewpoints (does {Polly want a cracker?) and standing on the results of an objective metric (VMAF) rather than what most people would agree is the nicer encode (MC).

    This is especially dishonest of you when you consider that you actually admit the x264 encode has significant problems.

    You are the typical x264 zealot.
    Thanks for your input. I will respond to this in the other thread since it deals with "Meridian" and testing specifics

  28. Dinosaur Supervisor KarMa's Avatar
    Join Date
    Jul 2015
    Location
    US
    Search Comp PM
    Originally Posted by sophisticles View Post
    You purposely tried to fix the test so that x264 would win, by using old, outdated, limited versions of Apple's AVC encoder, and old, outdated, versions of MC's encoder, with many of the quality improving options disabled, while enabling x264's "mastering" setting. You still ripped into the x264 encodes and when finally pressed to change the MC settings, you produced encodes that most people agree that Mc beat out x264 visually.
    Nothing is stopping you from creating your own encodings and posting them.

    you produced encodes that most people agree that Mc beat out x264 visually.
    Your inability to tell the difference between the samples provided by Poisondeathray comes to mind when reading this.

  29. I'm a Super Moderator johns0's Avatar
    Join Date
    Jun 2002
    Location
    canada
    Search Comp PM
    Time to stop mpeg 2 encoding.
    I think,therefore i am a hamster.




Similar Threads