Recherche avancée

Sur d’autres sites (7341)

• The use cases for a element in HTML

  1er janvier 2014, par silvia

  The W3C HTML WG and the WHATWG are currently discussing the introduction of a <main> element into HTML.

  The <main> element has been proposed by Steve Faulkner and is specified in a draft extension spec which is about to be accepted as a FPWD (first public working draft) by the W3C HTML WG. This implies that the W3C HTML WG will be looking for implementations and for feedback by implementers on this spec.

  I am supportive of the introduction of a <main> element into HTML. However, I believe that the current spec and use case list don’t make a good enough case for its introduction. Here are my thoughts.

  Main use case : accessibility

  In my opinion, the main use case for the introduction of <main> is accessibility.

  Like any other users, when blind users want to perceive a Web page/application, they need to have a quick means of grasping the content of a page. Since they cannot visually scan the layout and thus determine where the main content is, they use accessibility technology (AT) to find what is known as “landmarks”.

  “Landmarks” tell the user what semantic content is on a page : a header (such as a banner), a search box, a navigation menu, some asides (also called complementary content), a footer, …. and the most important part : the main content of the page. It is this main content that a blind user most often wants to skip to directly.

  In the days of HTML4, a hidden “skip to content” link at the beginning of the Web page was used as a means to help blind users access the main content.

  In the days of ARIA, the aria @role=main enables authors to avoid a hidden link and instead mark the element where the main content begins to allow direct access to the main content. This attribute is supported by AT – in particular screen readers – by making it part of the landmarks that AT can directly skip to.

  Both the hidden link and the ARIA @role=main approaches are, however, band aids : they are being used by those of us that make “finished” Web pages accessible by adding specific extra markup.

  A world where ARIA is not necessary and where accessibility developers would be out of a job because the normal markup that everyone writes already creates accessible Web sites/applications would be much preferable over the current world of band-aids.

  Therefore, to me, the primary use case for a <main> element is to achieve exactly this better world and not require specialized markup to tell a user (or a tool) where the main content on a page starts.

  An immediate effect would be that pages that have a <main> element will expose a “main” landmark to blind and vision-impaired users that will enable them to directly access that main content on the page without having to wade through other text on the page. Without a <main> element, this functionality can currently only be provided using heuristics to skip other semantic and structural elements and is for this reason not typically implemented in AT.

  Other use cases

  The <main> element is a semantic element not unlike other new semantic elements such as <header>, <footer>, <aside>, <article>, <nav>, or <section>. Thus, it can also serve other uses where the main content on a Web page/Web application needs to be identified.

  Data mining

  For data mining of Web content, the identification of the main content is one of the key challenges. Many scholarly articles have been published on this topic. This stackoverflow article references and suggests a multitude of approaches, but the accepted answer says “there’s no way to do this that’s guaranteed to work”. This is because Web pages are inherently complex and many <div>, <p>, <iframe> and other elements are used to provide markup for styling, notifications, ads, analytics and other use cases that are necessary to make a Web page complete, but don’t contribute to what a user consumes as semantically rich content. A <main> element will allow authors to pro-actively direct data mining tools to the main content.

  Search engines

  One particularly important “data mining” tool are search engines. They, too, have a hard time to identify which sections of a Web page are more important than others and employ many heuristics to do so, see e.g. this ACM article. Yet, they still disappoint with poor results pointing to findings of keywords in little relevant sections of a page rather than ranking Web pages higher where the keywords turn up in the main content area. A <main> element would be able to help search engines give text in main content areas a higher weight and prefer them over other areas of the Web page. It would be able to rank different Web pages depending on where on the page the search words are found. The <main> element will be an additional hint that search engines will digest.

  Visual focus

  On small devices, the display of Web pages designed for Desktop often causes confusion as to where the main content can be found and read, in particular when the text ends up being too small to be readable. It would be nice if browsers on small devices had a functionality (maybe a default setting) where Web pages would start being displayed as zoomed in on the main content. This could alleviate some of the headaches of responsive Web design, where the recommendation is to show high priority content as the first content. Right now this problem is addressed through stylesheets that re-layout the page differently depending on device, but again this is a band-aid solution. Explicit semantic markup of the main content can solve this problem more elegantly.

  Styling

  Finally, naturally, <main> would also be used to style the main content differently from others. You can e.g. replace a semantically meaningless <div id=”main”> with a semantically meaningful <main> where their position is identical. My analysis below shows, that this is not always the case, since oftentimes <div id=”main”> is used to group everything together that is not the header – in particular where there are multiple columns. Thus, the ease of styling a <main> element is only a positive side effect and not actually a real use case. It does make it easier, however, to adapt the style of the main content e.g. with media queries.

  Proposed alternative solutions

  It has been proposed that existing markup serves to satisfy the use cases that <main> has been proposed for. Let’s analyse these on some of the most popular Web sites. First let’s list the propsed algorithms.

  Proposed solution No 1 : Scooby-Doo

  On Sat, Nov 17, 2012 at 11:01 AM, Ian Hickson <ian@hixie.ch> wrote :
  | The main content is whatever content isn’t
  | marked up as not being main content (anything not marked up with <header>,
  | <aside>, <nav>, etc).
  

  This implies that the first element that is not a <header>, <aside>, <nav>, or <footer> will be the element that we want to give to a blind user as the location where they should start reading. The algorithm is implemented in https://gist.github.com/4032962.

  Proposed solution No 2 : First article element

  On Sat, Nov 17, 2012 at 8:01 AM, Ian Hickson  wrote :
  | On Thu, 15 Nov 2012, Ian Yang wrote :
  | >
  | > That’s a good idea. We really need an element to wrap all the <p>s,
  | > <ul>s, <ol>s, <figure>s, <table>s ... etc of a blog post.
  |
  | That’s called <article>.
  

  This approach identifies the first <article> element on the page as containing the main content. Here’s the algorithm for this approach.

  Proposed solution No 3 : An example heuristic approach

  The readability plugin has been developed to make Web pages readable by essentially removing all the non-main content from a page. An early source of readability is available. This demonstrates what a heuristic approach can perform.

  Analysing alternative solutions

  Comparison

  I’ve picked 4 typical Websites (top on Alexa) to analyse how these three different approaches fare. Ideally, I’d like to simply apply the above three scripts and compare pictures. However, since the semantic HTML5 elements <header>, <aside>, <nav>, and <footer> are not actually used by any of these Web sites, I don’t actually have this choice.

  So, instead, I decided to make some assumptions of where these semantic elements would be used and what the outcome of applying the first two algorithms would be. I can then compare it to the third, which is a product so we can take screenshots.

  Google.com

  http://google.com – search for “Scooby Doo”.

  The search results page would likely be built with :

  • a <nav> menu for the Google bar
  • a <header> for the search bar
  • another <header> for the login section
  • another <nav> menu for the search types
  • a <div> to contain the rest of the page
  • a <div> for the app bar with the search number
  • a few <aside>s for the left and right column
  • a set of <article>s for the search results

  “Scooby Doo” would find the first element after the headers as the “main content”. This is the element before the app bar in this case. Interestingly, there is a <div @id=main> already in the current Google results page, which “Scooby Doo” would likely also pick. However, there are a nav bar and two asides in this div, which clearly should not be part of the “main content”. Google actually placed a @role=main on a different element, namely the one that encapsulates all the search results.

  “First Article” would find the first search result as the “main content”. While not quite the same as what Google intended – namely all search results – it is close enough to be useful.

  The “readability” result is interesting, since it is not able to identify the main text on the page. It is actually aware of this problem and brings a warning before displaying this page :

  

  Facebook.com

  https://facebook.com

  A user page would likely be built with :

  • a <header> bar for the search and login bar
  • a <div> to contain the rest of the page
  • an <aside> for the left column
  • a <div> to contain the center and right column
  • an <aside> for the right column
  • a <header> to contain the center column “megaphone”
  • a <div> for the status posting
  • a set of <article>s for the home stream

  “Scooby Doo” would find the first element after the headers as the “main content”. This is the element that contains all three columns. It’s actually a <div @id=content> already in the current Facebook user page, which “Scooby Doo” would likely also pick. However, Facebook selected a different element to place the @role=main : the center column.

  “First Article” would find the first news item in the home stream. This is clearly not what Facebook intended, since they placed the @role=main on the center column, above the first blog post’s title. “First Article” would miss that title and the status posting.

  The “readability” result again disappoints but warns that it failed :

  

  YouTube.com

  http://youtube.com

  A video page would likely be built with :

  • a <header> bar for the search and login bar
  • a <nav> for the menu
  • a <div> to contain the rest of the page
  • a <header> for the video title and channel links
  • a <div> to contain the video with controls
  • a <div> to contain the center and right column
  • an <aside> for the right column with an <article> per related video
  • an <aside> for the information below the video
  • a <article> per comment below the video

  “Scooby Doo” would find the first element after the headers as the “main content”. This is the element that contains the rest of the page. It’s actually a <div @id=content> already in the current YouTube video page, which “Scooby Doo” would likely also pick. However, YouTube’s related videos and comments are unlikely to be what the user would regard as “main content” – it’s the video they are after, which generously has a <div id=watch-player>.

  “First Article” would find the first related video or comment in the home stream. This is clearly not what YouTube intends.

  The “readability” result is not quite as unusable, but still very bare :

  

  Wikipedia.com

  http://wikipedia.com (“Overscan” page)

  A Wikipedia page would likely be built with :

  • a <header> bar for the search, login and menu items
  • a <div> to contain the rest of the page
  • an &ls ; article> with title and lots of text
  • <article> an <aside> with the table of contents
  • several <aside>s for the left column

  Good news : “Scooby Doo” would find the first element after the headers as the “main content”. This is the element that contains the rest of the page. It’s actually a <div id=”content” role=”main”> element on Wikipedia, which “Scooby Doo” would likely also pick.

  “First Article” would find the title and text of the main element on the page, but it would also include an <aside>.

  The “readability” result is also in agreement.

  

  Results

  In the following table we have summarised the results for the experiments :

  Site	Scooby-Doo	First article	Readability
  Google.com	FAIL	SUCCESS	FAIL
  Facebook.com	FAIL	FAIL	FAIL
  YouTube.com	FAIL	FAIL	FAIL
  Wikipedia.com	SUCCESS	SUCCESS	SUCCESS

  Clearly, Wikipedia is the prime example of a site where even the simple approaches find it easy to determine the main content on the page. WordPress blogs are similarly successful. Almost any other site, including news sites, social networks and search engine sites are petty hopeless with the proposed approaches, because there are too many elements that are used for layout or other purposes (notifications, hidden areas) such that the pre-determined list of semantic elements that are available simply don’t suffice to mark up a Web page/application completely.

  Conclusion

  It seems that in general it is impossible to determine which element(s) on a Web page should be the “main” piece of content that accessibility tools jump to when requested, that a search engine should put their focus on, or that should be highlighted to a general user to read. It would be very useful if the author of the Web page would provide a hint through a <main> element where that main content is to be found.

  I think that the <main> element becomes particularly useful when combined with a default keyboard shortcut in browsers as proposed by Steve : we may actually find that non-accessibility users will also start making use of this shortcut, e.g. to get to videos on YouTube pages directly without having to tab over search boxes and other interactive elements, etc. Worthwhile markup indeed.

• How to cheat on video encoder comparisons

  21 juin 2010, par Dark Shikari — H.264, benchmark, stupidity, test sequences

  Over the past few years, practically everyone and their dog has published some sort of encoder comparison. Sometimes they’re actually intended to be something for the world to rely on, like the old Doom9 comparisons and the MSU comparisons. Other times, they’re just to scratch an itch — someone wants to decide for themselves what is better. And sometimes they’re just there to outright lie in favor of whatever encoder the author likes best. The latter is practically an expected feature on the websites of commercial encoder vendors.

  One thing almost all these comparisons have in common — particularly (but not limited to !) the ones done without consulting experts — is that they are horribly done. They’re usually easy to spot : for example, two videos at totally different bitrates are being compared, or the author complains about one of the videos being “washed out” (i.e. he screwed up his colorspace conversion). Or the results are simply nonsensical. Many of these problems result from the person running the test not “sanity checking” the results to catch mistakes that he made in his test. Others are just outright intentional.

  The result of all these mistakes, both intentional and accidental, is that the results of encoder comparisons tend to be all over the map, to the point of absurdity. For any pair of encoders, it’s practically a given that a comparison exists somewhere that will “prove” any result you want to claim, even if the result would be beyond impossible in any sane situation. This often results in the appearance of a “controversy” even if there isn’t any.

  Keep in mind that every single mistake I mention in this article has actually been done, usually in more than one comparison. And before I offend anyone, keep in mind that when I say “cheating”, I don’t mean to imply that everyone that makes the mistake is doing it intentionally. Especially among amateur comparisons, most of the mistakes are probably honest.

  So, without further ado, we will investigate a wide variety of ways, from the blatant to the subtle, with which you too can cheat on your encoder comparisons.

  Blatant cheating

  1. Screw up your colorspace conversions. A common misconception is that converting from YUV to RGB and back is a simple process where nothing can go wrong. This is quite untrue. There are two primary attributes of YUV : PC range (0-255) vs TV range (16-235) and BT.709 vs BT.601 conversion coefficients. That sums up to a total of 4 possible different types of YUV. When people compare encoders, they often use different frontends, some of which make incorrect assumptions about these attributes.

  Incorrect assumptions are so common that it’s often a matter of luck whether the tool gets it right or not. It doesn’t help that most videos don’t even properly signal which they are to begin with ! Often even the tool that the person running the comparison is using to view the source material gets the conversion wrong.

  Subsampling YUV (aka what everyone uses) adds yet another dimension to the problem : the locations which the chroma data represents (“chroma siting”) isn’t constant. For example, JPEG and MPEG-2 define different positions. This is even worse because almost nobody actually handles this correctly — the best approach is to simply make sure none of your software is doing any conversion. A mistake in chroma siting is what created that infamous PSNR graph showing Theora beating x264, which has been cited for ages since despite the developers themselves retracting it after realizing their mistake.

  Keep in mind that the video encoder is not responsible for colorspace conversion — almost all video encoders operate in the YUV domain (usually subsampled 4:2:0 YUV, aka YV12). Thus any problem in colorspace conversion is usually the fault of the tools used, not the actual encoder.

  How to spot it : “The color is a bit off” or “the contrast of the video is a bit duller”. There were a staggering number of “H.264 vs Theora” encoder comparisons which came out in favor of one or the other solely based on “how well the encoder kept the color” — making the results entirely bogus.

  2. Don’t compare at the same (or nearly the same) bitrate. I saw a VP8 vs x264 comparison the other day that gave VP8 30% more bitrate and then proceeded to demonstrate that it got better PSNR. You would think this is blindingly obvious, but people still make this mistake ! The most common cause of this is assuming that encoders will successfully reach the target bitrate you ask of them — particularly with very broken encoders that don’t. Always check the output filesizes of your encodes.

  How to spot it : The comparison lists perfectly round bitrates for every single test, as opposed to the actual bitrates achieved by the encoders, which will never be exactly matching in any real test.

  3. Use unfair encoding settings. This is a bit of a wide topic : there are many ways to do this. We’ll cover the more blatant ones in this part. Here’s some common ones :

  a. Simply cheat. Intentionally pick awful settings for the encoder you don’t like.

  b. Don’t consider performance. Pick encoding settings without any regard for some particular performance goal. For example, it’s perfectly reasonable to say “use the best settings possible, regardless of speed”. It’s also reasonable to look for a particular encoding speed target. But what isn’t reasonable is to pick extremely fast settings for one encoder and extremely slow settings for another encoder.

  c. Don’t attempt match compatibility options when it’s reasonable to do so. Keyframe interval is a classic one of these : shorter values reduce compression but improve seeking. An easy way to cheat is to simply not set them to the same value, biasing towards whatever encoder has the longer interval. This is most common as an accidental mistake with comparisons involving ffmpeg, where the default keyframe interval is an insanely low 12 frames.

  How to spot it : The comparison doesn’t document its approach regarding choice of encoding settings.

  4. Use ratecontrol methods unfairly. Constant bitrate is not the same as average bitrate — using one instead of the other is a great way to completely ruin a comparison. Another method is to use 1-pass bitrate mode for one encoder and 2-pass or constant quality for another. A good general approach is that, for any given encoder, one should use 2-pass if available and constant quality if not (it may take a few runs to get the bitrate you want, of course).

  Of course, it’s also fine to run a comparison with a particular mode in mind — for example, a comparison targeted at streaming applications might want to test using 1-pass CBR. Of course, in such a case, if CBR is not available in an encoder, you can’t compare to that encoder.

  How to spot it : It’s usually pretty obvious if the encoding settings are given.

  5. Use incredibly old versions of encoders. As it happens, Debian stable is not the best source for the most recent encoding software. Equally, using recent versions known to be buggy.

  6. Don’t distinguish between video formats and the software that encodes them. This is incredibly common : I’ve seen tests that claim to compare “H.264″ against something else while in fact actually comparing “Quicktime” against something else. It’s impossible to compare all H.264 encoders at once, so don’t even try — just call the comparison “Quicktime versus X” instead of “H.264 versus X”. Or better yet, use a good H.264 encoder, like x264 and don’t bother testing awful encoders to begin with.

  Less-obvious cheating

  1. Pick a bitrate that’s way too low. Low bitrate testing is very effective at making differences between encoders obvious, particularly if doing a visual comparison. But past a certain point, it becomes impossible for some encoders to keep up. This is usually an artifact of the video format itself — a scalability limitation. Practically all DCT-based formats have this kind of limitation (wavelets are mostly immune).

  In reality, this is rarely a problem, because one could merely downscale the video to resolve the problem — lower resolutions need fewer bits. But people rarely do this in comparisons (it’s hard to do it fairly), so the best approach is to simply not use absurdly low bitrates. What is “absurdly low” ? That’s a hard question — it ends up being a matter of using one’s best judgement.

  This tends to be less of a problem in larger-scale tests that use many different bitrates.

  How to spot it : At least one of the encoders being compared falls apart completely and utterly in the screenshots.

  Biases towards, a lot : Video formats with completely scalable coding methods (Dirac, Snow, JPEG-2000, SVC).

  Biases towards, a little : Video formats with coding methods that improve scalability, such as arithmetic coding, B-frames, and run-length coding. For example, H.264 and Theora tend to be more scalable than MPEG-4.

  2. Pick a bitrate that’s way too high. This is staggeringly common mistake : pick a bitrate so high that all of the resulting encodes look absolutely perfect. The claim is then made that “there’s no significant difference” between any of the encoders tested. This is surprisingly easy to do inadvertently on sources like Big Buck Bunny, which looks transparent at relatively low bitrates. An equally common but similar mistake is to test at a bitrate that isn’t so high that the videos look perfect, but high enough that they all look very good. The claim is then made that “the difference between these encoders is small”. Well, of course, if you give everything tons of bitrate, the difference between encoders is small.

  How to spot it : You can’t tell which image is the source and which is the encode.

  3. Making invalid comparisons using objective metrics. I explained this earlier in the linked blog post, but in short, if you’re going to measure PSNR, make sure all the encoders are optimized for PSNR. Equally, if you’re going to leave the encoder optimized for visual quality, don’t measure PSNR — post screenshots instead. Same with SSIM or any other objective metric. Furthermore, don’t blindly do metric comparisons — always at least look at the output as a sanity test. Finally, do not claim that PSNR is particularly representative of visual quality, because it isn’t.

  How to spot it : Encoders with psy optimizations, such as x264 or Theora 1.2, do considerably worse than expected in PSNR tests, but look much better in visual comparisons.

  4. Lying with graphs. Using misleading scales on graphs is a great way to make the differences between encoders seem larger or smaller than they actually are. A common mistake is to scale SSIM linearly : in fact, 0.99 is about twice as good as 0.98, not 1% better. One solution for this is to use db to compare SSIM values.

  5. Using lossy screenshots. Posting screenshots as JPEG is a silly, pointless way to worsen an encoder comparison.

  Subtle cheating

  1. Unfairly pick screenshots for comparison. Comparing based on stills is not ideal, but it’s often vastly easier than comparing videos in motion. But it also opens up the door to unfairness. One of the most common mistakes is to pick a frame immediately after (or on) a keyframe for one encoder, but which isn’t for the other encoder. Particularly in the case of encoders that massively boost keyframe quality, this will unfairly bias in favor of the one with the recent keyframe.

  How to spot it : It’s very difficult to tell, if not impossible, unless they provide the video files to inspect.

  2. Cherry-pick source videos. Good source videos are incredibly hard to come by — almost everything is already compressed and what’s left is usually a very poor example of real content. Here’s some common ways to bias unfairly using cherry-picking :

  a. Pick source videos that are already heavily compressed. Pre-compressed source isn’t much of an issue if your target quality level for testing is much lower than that of the source, since any compression artifacts in the source will be a lot smaller than those created by the encoders. But if the source is already very compressed, or you’re testing at a relatively high quality level, this becomes a significant issue.

  Biases towards : Anything that uses a similar transform to the source content. For MPEG-2 source material, this biases towards formats that use the 8x8dct or a very close approximation : MPEG-1/2/4, H.263, and Theora. For H.264 source material, this biases towards formats that use a 4×4 transform : H.264 and VP8.

  b. Pick standard test clips that were not intended for this purpose. There are a wide variety of uncompressed “standard test clips“. Some of these are not intended for general-purpose use, but rather exist to test specific encoder capabilities. For example, Mobile Calendar (“mobcal”) is extremely sharp and low motion, serving to test interpolation capabilities. It will bias incredibly heavily towards whatever encoder uses more B-frames and/or has higher-precision motion compensation. Other test clips are almost completely static, such as the classic “akiyo”. These are also not particularly representative of real content.

  c. Pick very noisy content. Noise is — by definition — not particularly compressible. Both in terms of PSNR and visual quality, a very noisy test clip will tend to reduce the differences between encoders dramatically.

  d. Pick a test clip to exercise a specific encoder feature. I’ve often used short clips from Touhou games to demonstrate the effectiveness of x264′s macroblock-tree algorithm. I’ve sometimes even used it to compare to other encoders as part of such a demonstration. I’ve also used the standard test clip “parkrun” as a demonstration of adaptive quantization. But claiming that either is representative of most real content — and thus can be used as a general determinant of how good encoders are — is of course insane.

  e. Simply encode a bunch of videos and pick the one your favorite encoder does best on.

  3. Preprocessing the source. A encoder test is a test of encoders, not preprocessing. Some encoding apps may add preprocessors to the source, such as noise reduction. This may make the video look better — possibly even better than the source — but it’s not a fair part of comparing the actual encoders.

  4. Screw up decoding. People often forget that in addition to encoding, a test also involves decoding — a step which is equally possible to do wrong. One common error caused by this is in tests of Theora on content whose resolution isn’t divisible by 16. Decoding is often done with ffmpeg — which doesn’t crop the edges properly in some cases. This isn’t really a big deal visually, but in a PSNR comparison, misaligning the entire frame by 4 or 8 pixels is a great way of completely invalidating the results.

  The greatest mistake of all

  Above all, the biggest and most common mistake — and the one that leads to many of the problems mentioned here – is the mistaken belief that one, or even a few tests can really represent all usage fairly. Any comparison has to have some specific goal — to compare something in some particular case, whether it be “maximum offline compression ignoring encoding speed” or “real-time high-speed video streaming” or whatnot. And even then, no comparison can represent all use-cases in that category alone. An encoder comparison can only be honest if it’s aware of its limitations.

• How to cheat on video encoder comparisons

  21 juin 2010, par Dark Shikari — benchmark, H.264, stupidity, test sequences

  Over the past few years, practically everyone and their dog has published some sort of encoder comparison. Sometimes they’re actually intended to be something for the world to rely on, like the old Doom9 comparisons and the MSU comparisons. Other times, they’re just to scratch an itch — someone wants to decide for themselves what is better. And sometimes they’re just there to outright lie in favor of whatever encoder the author likes best. The latter is practically an expected feature on the websites of commercial encoder vendors.

  One thing almost all these comparisons have in common — particularly (but not limited to !) the ones done without consulting experts — is that they are horribly done. They’re usually easy to spot : for example, two videos at totally different bitrates are being compared, or the author complains about one of the videos being “washed out” (i.e. he screwed up his colorspace conversion). Or the results are simply nonsensical. Many of these problems result from the person running the test not “sanity checking” the results to catch mistakes that he made in his test. Others are just outright intentional.

  The result of all these mistakes, both intentional and accidental, is that the results of encoder comparisons tend to be all over the map, to the point of absurdity. For any pair of encoders, it’s practically a given that a comparison exists somewhere that will “prove” any result you want to claim, even if the result would be beyond impossible in any sane situation. This often results in the appearance of a “controversy” even if there isn’t any.

  Keep in mind that every single mistake I mention in this article has actually been done, usually in more than one comparison. And before I offend anyone, keep in mind that when I say “cheating”, I don’t mean to imply that everyone that makes the mistake is doing it intentionally. Especially among amateur comparisons, most of the mistakes are probably honest.

  So, without further ado, we will investigate a wide variety of ways, from the blatant to the subtle, with which you too can cheat on your encoder comparisons.

  Blatant cheating

  1. Screw up your colorspace conversions. A common misconception is that converting from YUV to RGB and back is a simple process where nothing can go wrong. This is quite untrue. There are two primary attributes of YUV : PC range (0-255) vs TV range (16-235) and BT.709 vs BT.601 conversion coefficients. That sums up to a total of 4 possible different types of YUV. When people compare encoders, they often use different frontends, some of which make incorrect assumptions about these attributes.

  Incorrect assumptions are so common that it’s often a matter of luck whether the tool gets it right or not. It doesn’t help that most videos don’t even properly signal which they are to begin with ! Often even the tool that the person running the comparison is using to view the source material gets the conversion wrong.

  Subsampling YUV (aka what everyone uses) adds yet another dimension to the problem : the locations which the chroma data represents (“chroma siting”) isn’t constant. For example, JPEG and MPEG-2 define different positions. This is even worse because almost nobody actually handles this correctly — the best approach is to simply make sure none of your software is doing any conversion. A mistake in chroma siting is what created that infamous PSNR graph showing Theora beating x264, which has been cited for ages since despite the developers themselves retracting it after realizing their mistake.

  Keep in mind that the video encoder is not responsible for colorspace conversion — almost all video encoders operate in the YUV domain (usually subsampled 4:2:0 YUV, aka YV12). Thus any problem in colorspace conversion is usually the fault of the tools used, not the actual encoder.

  How to spot it : “The color is a bit off” or “the contrast of the video is a bit duller”. There were a staggering number of “H.264 vs Theora” encoder comparisons which came out in favor of one or the other solely based on “how well the encoder kept the color” — making the results entirely bogus.

  2. Don’t compare at the same (or nearly the same) bitrate. I saw a VP8 vs x264 comparison the other day that gave VP8 30% more bitrate and then proceeded to demonstrate that it got better PSNR. You would think this is blindingly obvious, but people still make this mistake ! The most common cause of this is assuming that encoders will successfully reach the target bitrate you ask of them — particularly with very broken encoders that don’t. Always check the output filesizes of your encodes.

  How to spot it : The comparison lists perfectly round bitrates for every single test, as opposed to the actual bitrates achieved by the encoders, which will never be exactly matching in any real test.

  3. Use unfair encoding settings. This is a bit of a wide topic : there are many ways to do this. We’ll cover the more blatant ones in this part. Here’s some common ones :

  a. Simply cheat. Intentionally pick awful settings for the encoder you don’t like.

  b. Don’t consider performance. Pick encoding settings without any regard for some particular performance goal. For example, it’s perfectly reasonable to say “use the best settings possible, regardless of speed”. It’s also reasonable to look for a particular encoding speed target. But what isn’t reasonable is to pick extremely fast settings for one encoder and extremely slow settings for another encoder.

  c. Don’t attempt match compatibility options when it’s reasonable to do so. Keyframe interval is a classic one of these : shorter values reduce compression but improve seeking. An easy way to cheat is to simply not set them to the same value, biasing towards whatever encoder has the longer interval. This is most common as an accidental mistake with comparisons involving ffmpeg, where the default keyframe interval is an insanely low 12 frames.

  How to spot it : The comparison doesn’t document its approach regarding choice of encoding settings.

  4. Use ratecontrol methods unfairly. Constant bitrate is not the same as average bitrate — using one instead of the other is a great way to completely ruin a comparison. Another method is to use 1-pass bitrate mode for one encoder and 2-pass or constant quality for another. A good general approach is that, for any given encoder, one should use 2-pass if available and constant quality if not (it may take a few runs to get the bitrate you want, of course).

  Of course, it’s also fine to run a comparison with a particular mode in mind — for example, a comparison targeted at streaming applications might want to test using 1-pass CBR. Of course, in such a case, if CBR is not available in an encoder, you can’t compare to that encoder.

  How to spot it : It’s usually pretty obvious if the encoding settings are given.

  5. Use incredibly old versions of encoders. As it happens, Debian stable is not the best source for the most recent encoding software. Equally, using recent versions known to be buggy.

  6. Don’t distinguish between video formats and the software that encodes them. This is incredibly common : I’ve seen tests that claim to compare “H.264″ against something else while in fact actually comparing “Quicktime” against something else. It’s impossible to compare all H.264 encoders at once, so don’t even try — just call the comparison “Quicktime versus X” instead of “H.264 versus X”. Or better yet, use a good H.264 encoder, like x264 and don’t bother testing awful encoders to begin with.

  Less-obvious cheating

  1. Pick a bitrate that’s way too low. Low bitrate testing is very effective at making differences between encoders obvious, particularly if doing a visual comparison. But past a certain point, it becomes impossible for some encoders to keep up. This is usually an artifact of the video format itself — a scalability limitation. Practically all DCT-based formats have this kind of limitation (wavelets are mostly immune).

  In reality, this is rarely a problem, because one could merely downscale the video to resolve the problem — lower resolutions need fewer bits. But people rarely do this in comparisons (it’s hard to do it fairly), so the best approach is to simply not use absurdly low bitrates. What is “absurdly low” ? That’s a hard question — it ends up being a matter of using one’s best judgement.

  This tends to be less of a problem in larger-scale tests that use many different bitrates.

  How to spot it : At least one of the encoders being compared falls apart completely and utterly in the screenshots.

  Biases towards, a lot : Video formats with completely scalable coding methods (Dirac, Snow, JPEG-2000, SVC).

  Biases towards, a little : Video formats with coding methods that improve scalability, such as arithmetic coding, B-frames, and run-length coding. For example, H.264 and Theora tend to be more scalable than MPEG-4.

  2. Pick a bitrate that’s way too high. This is staggeringly common mistake : pick a bitrate so high that all of the resulting encodes look absolutely perfect. The claim is then made that “there’s no significant difference” between any of the encoders tested. This is surprisingly easy to do inadvertently on sources like Big Buck Bunny, which looks transparent at relatively low bitrates. An equally common but similar mistake is to test at a bitrate that isn’t so high that the videos look perfect, but high enough that they all look very good. The claim is then made that “the difference between these encoders is small”. Well, of course, if you give everything tons of bitrate, the difference between encoders is small.

  How to spot it : You can’t tell which image is the source and which is the encode.

  3. Making invalid comparisons using objective metrics. I explained this earlier in the linked blog post, but in short, if you’re going to measure PSNR, make sure all the encoders are optimized for PSNR. Equally, if you’re going to leave the encoder optimized for visual quality, don’t measure PSNR — post screenshots instead. Same with SSIM or any other objective metric. Furthermore, don’t blindly do metric comparisons — always at least look at the output as a sanity test. Finally, do not claim that PSNR is particularly representative of visual quality, because it isn’t.

  How to spot it : Encoders with psy optimizations, such as x264 or Theora 1.2, do considerably worse than expected in PSNR tests, but look much better in visual comparisons.

  4. Lying with graphs. Using misleading scales on graphs is a great way to make the differences between encoders seem larger or smaller than they actually are. A common mistake is to scale SSIM linearly : in fact, 0.99 is about twice as good as 0.98, not 1% better. One solution for this is to use db to compare SSIM values.

  5. Using lossy screenshots. Posting screenshots as JPEG is a silly, pointless way to worsen an encoder comparison.

  Subtle cheating

  1. Unfairly pick screenshots for comparison. Comparing based on stills is not ideal, but it’s often vastly easier than comparing videos in motion. But it also opens up the door to unfairness. One of the most common mistakes is to pick a frame immediately after (or on) a keyframe for one encoder, but which isn’t for the other encoder. Particularly in the case of encoders that massively boost keyframe quality, this will unfairly bias in favor of the one with the recent keyframe.

  How to spot it : It’s very difficult to tell, if not impossible, unless they provide the video files to inspect.

  2. Cherry-pick source videos. Good source videos are incredibly hard to come by — almost everything is already compressed and what’s left is usually a very poor example of real content. Here’s some common ways to bias unfairly using cherry-picking :

  a. Pick source videos that are already heavily compressed. Pre-compressed source isn’t much of an issue if your target quality level for testing is much lower than that of the source, since any compression artifacts in the source will be a lot smaller than those created by the encoders. But if the source is already very compressed, or you’re testing at a relatively high quality level, this becomes a significant issue.

  Biases towards : Anything that uses a similar transform to the source content. For MPEG-2 source material, this biases towards formats that use the 8x8dct or a very close approximation : MPEG-1/2/4, H.263, and Theora. For H.264 source material, this biases towards formats that use a 4×4 transform : H.264 and VP8.

  b. Pick standard test clips that were not intended for this purpose. There are a wide variety of uncompressed “standard test clips“. Some of these are not intended for general-purpose use, but rather exist to test specific encoder capabilities. For example, Mobile Calendar (“mobcal”) is extremely sharp and low motion, serving to test interpolation capabilities. It will bias incredibly heavily towards whatever encoder uses more B-frames and/or has higher-precision motion compensation. Other test clips are almost completely static, such as the classic “akiyo”. These are also not particularly representative of real content.

  c. Pick very noisy content. Noise is — by definition — not particularly compressible. Both in terms of PSNR and visual quality, a very noisy test clip will tend to reduce the differences between encoders dramatically.

  d. Pick a test clip to exercise a specific encoder feature. I’ve often used short clips from Touhou games to demonstrate the effectiveness of x264′s macroblock-tree algorithm. I’ve sometimes even used it to compare to other encoders as part of such a demonstration. I’ve also used the standard test clip “parkrun” as a demonstration of adaptive quantization. But claiming that either is representative of most real content — and thus can be used as a general determinant of how good encoders are — is of course insane.

  e. Simply encode a bunch of videos and pick the one your favorite encoder does best on.

  3. Preprocessing the source. A encoder test is a test of encoders, not preprocessing. Some encoding apps may add preprocessors to the source, such as noise reduction. This may make the video look better — possibly even better than the source — but it’s not a fair part of comparing the actual encoders.

  4. Screw up decoding. People often forget that in addition to encoding, a test also involves decoding — a step which is equally possible to do wrong. One common error caused by this is in tests of Theora on content whose resolution isn’t divisible by 16. Decoding is often done with ffmpeg — which doesn’t crop the edges properly in some cases. This isn’t really a big deal visually, but in a PSNR comparison, misaligning the entire frame by 4 or 8 pixels is a great way of completely invalidating the results.

  The greatest mistake of all

  Above all, the biggest and most common mistake — and the one that leads to many of the problems mentioned here – is the mistaken belief that one, or even a few tests can really represent all usage fairly. Any comparison has to have some specific goal — to compare something in some particular case, whether it be “maximum offline compression ignoring encoding speed” or “real-time high-speed video streaming” or whatnot. And even then, no comparison can represent all use-cases in that category alone. An encoder comparison can only be honest if it’s aware of its limitations.