Expectations destroy results

Later, experimenters at Caltech felt that Feynman’s very presence exerted a sort of moral pressure on their findings and methods. He was mercilessly skeptical. He loved to talk about the famous oil-drop experiment of Caltech’s first great physicist Robert Millikan, which revealed the indivisible unit charge of the electron by isolating it in tiny, floating oil drops. The experiment was right but some of the numbers were wrong–and the record of subsequent experimenters stood as a permanent embarrassment to physics. They did not cluster around the correct result; rather, they slowly closed in on it. Millikan’s error exerted a psychological pull, like a distant magnet forcing their observations off center. If a Caltech experimenter told Feynman about a result reached after a complex process of correcting data, Feynman was sure to ask how the experimenter had decided when to stop correcting, and whether the decision had been made before the experimenter could see what effect it would have on the outcome. It was all too easy to fall into the trap of correcting until the answer looked right. To avoid it required an intimate acquaintanceship with the rules of the scientist’s game. It also required not just honesty, but a sense that honesty required exertion.

James Gleick: Genius: The Life and Science of Richard Feynman 

Even something as seemingly free of bias as measuring stuff in physics can be biased in reality by expectations. Who would have thought that the great Millikan may have actually made a mistake?

So it isn’t really a surprise when you read something like ‘The Dishonesty of Sighted Listening Tests‘, which relates how virtually indistinguishable loudspeakers (as far as blind tests went) became very different when they could be seen, and how the ranking orders of them changed.

Posted in Mysticism, Testing | Leave a comment

Audio subjectivism

My brother has drawn my attention to an episode of ‘Skeptoid’ in which (third item) the magical benefits of expensive audio cables are dismissed, along with the value of face masks in combating swine flu, the radioactive dangers of depleted uranium, and some weird medical treatments.

I pretty much agree with the response*. What interested me more was one of the comments, which referenced Ethan Winer’s website. That has some very interesting material on it. Winer knows computers, audio recording, and actually playing music very well. He has a bunch of very sensible thoughts on the myths of some audiophiles. A good starting point is his ‘Audio Magic’ section under ‘Ethan’s Magazine Articles‘.

Too many articles to discuss in detail, but I would suggest you download the five musical samples at Is 24-bit Recording Really Better? and try yourself out on some blind testing. Kind of goes with my own ‘What is Dither?‘ article, in which I explain (with the assistance of pictures), how dither actually works to, at least theoretically, improve sound.

I for one am going to be going through these articles for some time, learning more. Oh, one more: a plausible explanation for some differences in the subjective experience of sound — differences which are conventionally thought should not be there. Called ‘Why We Believe‘, using very interesting measurements, he suggests that differences may be largely due to slight differences in head position. He shows pretty conclusively that large differences in frequency response — certainly quite audible — result from moving the listening point by just a few centimetres.

Truth be told, I think he is too kind. I myself have experienced audible differences which, it has soon become apparent, are entirely the result of my expectations. Because I am essentially a chicken — I would be very embarrassed if I couldn’t reliably distinguish a phenomenon which I may note in an article under test conditions — I refuse to rush into print until I’ve checked multiple times. When it comes to cables and amplifiers and the like, I find that the phenomenon has disappeared, or the thing I thought I heard was there all along, just previously unnoticed with other equipment.

One more pointer. Mr Winer links to an excellent article ‘Science and Subjectivism in Audio‘. I highly recommend this article if you think there may indeed be something in the more mystical claims of some in the audiophile community.

* I’m not suggesting you go cheap and nasty for your cabling needs. Just be sensible. $5 for a set of interconnects suggests that they might be a light for quality control, and even lack proper shielding. $20 should provide excellent quality. $200 is a waste.

For speaker cables, I use good thick stuff which sells for about $5 a metre. Just get the cable impedance as low as reasonably possible for speakers. Remember, the cable resistance for interconnects is totally insignificant compared to the typically 47,000 ohm input to which it’s connected. For speakers, a cable’s 0.1 ohms is still a significant proportion of the three to thirty ohms of resistance that an amplifier my face across a loudspeaker’s operating frequency range.

Posted in Audio, Cables, Mysticism, Testing | Leave a comment

Slavish copying into picture quality disaster

Top Gear illustrationThe Australian version of the BBC TV show Top Gear is growing on me. Unfortunately, it copies the original so slavishly that this extends to some of the more horrible photographic choices. In particular, it follows a practice about which I earlier wrote in ‘Insane artistic choices‘. It uses — sometimes, and not consistently — a picture processing technique of drop field deinterlacing. That is, the second field is simply omitted, and replaced by a repeat of the first field. This has the intended effect of adding a slight film-like jerkiness to motion. It has the unpleasant side effect (I’m assuming that this isn’t actually desired!) of reducing the vertical resolution by half. Instead of getting 576 lines, you are only betting 288.

The shot to the right is from a segment within the studio, where there isn’t even the excuse of motion. Note all the jaggies. Note, also, that I have not retouched this in any significant way. I grabbed a frame digitally from the video stream and saved it. I cropped the picture in Photoshop to the size you see, and then recompressed it into JPEG. The re-compressed image is indistiguishable to the eye from the original.

I have not scaled or deinterlaced this in any way. This was a section of the picture actually broadcast on Australian TV in 2009!

Posted in DTV, Video | Leave a comment

New Blu-ray vs DVD comparisons

Here are some new side-by-side Blu-ray vs PAL DVD comparisons that I’ve just put up:

Some really stand out, such as this shot of a newspaper in Sin City:

Sin City comparison

It’s not just the readabily of the text, but also the visibility of the dots used to print photos.

The Blu-ray of Sleeping Beauty looks pretty good too. Note that the comparison DVD was the remastered one:

Comparison 3

I’ve also added a video bitrate graph and a PIP one as well, plus some stats, to my earlier posting for Batman Begins.

Posted in Blu-ray, Disc details, DVD | Leave a comment

Widescreen ‘pan and scan’

A few weeks ago I recorded The Da Vinci Code from high definition TV, since I was too busy to watch it during broadcast, and in any case I wanted to test out the Panasonic Blu-ray recorder I was reviewing. The recorder had to go back to Panasonic, so I used it to burn the movie to a BD-RE disc (ie. re-writable). Then, before I had a chance to watch it, the formal Blu-ray release of the movie (on a massive two disc special edition) was sent to me. So yesterday I thought I’d compare a couple of frames to see if there is any marked difference between a 12-ish Mbps MPEG2 free to air HD broadcast and a 24.44Mbps MPEG4 AVC Blu-ray version. I ripped a few thousand frames from the latter in preparation, and then started to try finding matching frames from the former.

It wasn’t to be. The first thing I noticed was that the resolution of the TV version was 1,440 by 1,080 pixels. I had forgotten that WIN TV, which was the broadcasting station for the movie, still uses the lower resolution (full HD resolution is, of course, 1,920 by 1,080 pixels). Not really comparable in that case.

And then I compared a frame from each. The Blu-ray was 2.35:1 while the TV version was 1.78:1. Here is the same frame from both (Blu-ray top, TV bottom):

Da Vinci Code framing

Note the reformatting. The TV frame doesn’t just have the edges lopped off to make it fill up the screen. It has them somewhat lopped off, but more of the top and bottom of the frame is presented as well.

So, even with HDTV, you may well not be getting the original aspect ratio.

Posted in Blu-ray, DTV, HDTV | Leave a comment

Blu-ray video inputs and outputs

Blu-ray players have to cope with many more signal standards than DVD. On the audio side, the latter deal with PCM, Dolby Digital, DTS and (for non-US players) MPEG2, whereas Blu-ray players must deal with all of those, plus two new varieties from Dolby and DTS, plus multichannel PCM. For video, DVD had only MPEG2, whereas Blu-ray adds VC1 and MPEG4 AVC.

It is with resolutions that things really go berserk. DVD had to accept 352 x 240, 352 x 288, 352 x 480, 352 x 576, 704 x 480, 702 x 576, 720 x 480 or 720 x 576 pixel video streams at either 60 (for all the 240 or 480 pixel tall formats) or 50 hertz (for the rest). On the output side, things were simpler. The output had to be either 480i60 or 576i50. No matter which the original resolution was, it was scaled to match one or other of those output formats. (Progressive scan conversion and upscaling to high resolution are simply later enhancements, not requirements.)

Blu-ray players have to cope with all that and more. I shall ignore all those super low res formats and just deal with the actual video formats available on Blu-ray discs. These are: 480i60, 576i50, 720p60, 720p50, 1080i60, 1080i50 and 1080p24. The way things are developing, the two 720p formats seem to be very rare, so I won’t bother mentioning them any further. As it happens, the 1080i formats can have physical resolutions of either 1,920 or 1,440 by 1,080 pixels, but I’m not going to worry too much about that here either.

The output formats typically available from a Blu-ray player are: 480i60, 576i50, 480p60, 576p50, 720p60, 720p50, 1080i60, 1080i50, 1080p60, 1080p50 and 1080p24. That’s a lot of options! But it isn’t as complicated as all that because some of these output resolutions are better fitted to the modern home entertainment display, and some are natural fits for the source.

Since most new high quality displays offer a physical resolution of 1920 by 1080 pixels, and accept 1080p signals at 50, 60 and 24 hertz, 1080p is usually the best choice. The frequency should depend on the frequency of the source signal.

Blu-ray players typically allow you to select 1080p24 as an output separately from the others, because the first generation of 1080p-capable displays did not necessarily also support 1080p24. So what you normally find is a resolution setting (you can choose from 480/576p, 720p, 1080i or 1080p), with a separate setting to 1080p24.

If you select 1080p as the resolution, and switch on 1080p24, then any 480i60 or 1080i60 content will be output at as 1080p60, and any 1080p24 content will be output at 1080p24. The 1080p24 will have no processing applied to it at all. The 480i60 will be deinterlaced (ie. converted to progressive scan) and scaled up, and 1080i60 will merely be deinterlaced.

Similar processes happen for 576i50/1080i50 material, except that the output format in those cases is 1080p50.

Now here is what one Blu-ray player — which shall remain nameless — does. It is pre-fixed to Australian television standards. If you select ‘1080p’ as the output resolution, everything is converted to 1080p50. Everything. If you play a 1080p24 Blu-ray disc, it gets output at 1080p50. Because of the mismatch in frame rates, some of the film frames have to be repeated more frequently than others, so camera pans are jerky. Right now I have the Australian version of ‘Citizen Kane’ DVD playing. This is NTSC format, despite being released here. It, also, is showing jerkiness during camera pans, since the player is showing it at 1080p50.

Likewise, if you select ‘1080p24’, then everything gets converted to 1080p24. Even PAL and 1080i50 material. Despite that meaning that instead of frames being repeated to adjust the timing, frames are dropped.

So it’s worth paying attention to these things when you’re selecting your Blu-ray player.

Posted in Blu-ray, Equipment | Leave a comment

Blu-ray Audio

Over at the ‘Unofficial Blu-ray Audio and Video Specifications Thread‘ on the AVS Forum, a Russian contributor has put up specifications for a couple of Blu-ray audio discs. Note, there’s no particular format difference between Blu-ray audio and regular Blu-ray. It’s just that the former has little or no video (just puts up a static picture or whatever), so as to allow more bits for the audio.

The second disc — DIVERTIMENTI – TrondheimSolistene — is incredible. The audio is so dense that the whole 69 minute program is repeated in two *.m2ts files to cover the five different formats in which it is presented. You get it in 24 bit, 192kHz, 5.1 channels in three ways: losslessly compressed DTS-HD Master Audio (12990kbps), losslessly compressed Dolby TrueHD (13122kbps) and uncompressed LPCM (27648kbps). It is also presented in stereo 24 bit, 192kHz LPCM (9216kbps) and regular 5.1 channel Dolby Digital (640kbps). Actually, you get a sixth version as well: the DTS-HD Master Audio track contains a standard (albeit nominally 24 bit) DTS version at 1,536kbps.

This seems like an ideal disc to use to compare audio standards.

Posted in Audio, Blu-ray, Compression | Leave a comment

Time to stop using ZoneAlarm for my protection, it seems

Well, that’s irritating. For quite a few years I’ve been using ZoneAlarm Pro for my computer security needs. I’ve currently got three licences for three computers. It provides a firewall, antivirus, and a few odds and sods to keep my computer safe.

Or so I thought.

Today I received an email from ZoneLabs warning me of a security risk called ‘Gumblar’, in part as follows:

Gumblar is currently targeting users of IE and Google search, delivering malware through compromised sites that infects a user’s PC and subsequently intercepts traffic between the user and the visited sites. This means that once infected, anything the victim types could be monitored and used to commit identity theft, such as stealing credit card numbers, passwords or other sensitive data. Visitors encountering the compromised website also risk having their subsequent search results replaced with links that point to other malicious websites. The malware can also steal FTP credentials from the victim’s computer and use them to infect more sites, thus increasing the spread of this threat.Who is at risk?
Users of Internet Explorer and Google’s search engine.

Okay. But I’ve got the ZoneAlarm Security Suite which is supposed to be all I need to secure my computer. In fact, I pay for something I don’t really need, since I could rely on the Windows XP firewall. All I really need to add is antivirus.

But:

How do I protect myself?
If you are running ZoneAlarm® ForceField™ browser security technology you are already protected. If you are running ZoneAlarm Extreme Security, you must turn ON ForceField virtualization (Open ZoneAlarm Extreme and go to ‘browser security’, ‘settings’, ‘advanced’ and click ‘enable virtualization’).If you are only running the ZoneAlarm Pro or ZoneAlarm Security Suite, you could be at risk, since they do not include virtualization technology which protects from these types of stealth Web browser attacks. [colour in original]

Apparently, all I have to do is pay $US29.95 to save myself from Gumblar. Oh, is that for one computer or all three? And what about next week when some new form of attack develops. And the week after with a different type. Am I going to have to keep paying in dribs and drabs forever, on top of my ZoneAlarm licence?

When my licences run out I shall be going back to Windows Firewall and looking into some other antivirus.

Posted in Admin, Computer | Leave a comment

Sound, Branching and Disc Real Estate

My Blu-ray disc data gathering project continues, in which I try to determine as many technical and standards details as possible about Blu-ray discs. Of the nearly 300 Blu-ray discs in my database, 98 now have a ‘Completion’ score of ’12’, the highest I presently give. With DVDs the highest score I gave was ‘5’, which shows how much more complicated BDs are.

To get ’12’, my database record identifies — by my own exploration — amongst the more common stuff things like region code, size of the disc, the number of layers, the Titles on the disc, the special features, the video and audio formats of the main feature, and the video and audio formats of every identifiable special extra on the disc. Oh, and the *.m2ts file names for each title, and the audio and video bitrates.

That can be quite a bit of work.

I am presently going back over the database entries I have with a score of ‘8’ (these lack the items mentioned in the last sentence of the list above) and promoting them by running them through BDInfo. Brilliant tool! Right now I am working on The Simpsons Movie. It has some interesting characteristics.

Those familiar with the disc will know that there are two ways of playing it back. One is just to watch the movie in the usual way, with a range of different languages and a ‘Director’s Commentary’ available. The other way is to watch it with ‘Commentary’ enabled. This is different to the other commentary, and invoking it sends to the movie back to the start.

This version uses a different playlist and is 101:28 long vs 86:44. The difference is due to three points at which the commentary team ‘pause’ the movie to allow themselves time to expand on an issue. What you see on the screen is the image pausing, and then fading to grey scale. At the end, the colour ramps up again in the moment before movement is resumed. Obviously this ‘paused’ moment demands far fewer bits of data than movements, so you can see these three sections clearly on this bitrate graph:

Pauses in the Simpsons

All this is managed through playlists and seamless branching. The main movie consists of six files playing one after the other, while the commentary version has nine: the original six plus three others.

While trying to confirm this I played the three files for the ‘pause’ sections — 00120.m2ts, 00122.m2ts and 00121.m2ts, in order — and noticed that all three files had the main audio with which the movie opens: the 20th Century fanfare, followed by Ralph’s explanation as he appears on their logo, and so on. Further exploration revealed that all the other language soundtracks were also present, along with the other commentary track, and all took their sound from the start of the movie. The exception was the actual commentary track which was supposed to be there. This had the proper comments relating to that section of the movie.

When you’re playing the disc normally, only that commentary section is available. All the other tracks are invisible.

So why are these other audio tracks present on those insert files when they will never be played? My guess is that the requirements of seamless branching include having compatible audio tracks. All those other languages exist in the main part of the movie, so presumably they must exist for that seamless matching together in the pause sections. I could be wrong here, but let us grant that that is indeed the case.

Some space is still wasted, though. All the audio tracks are constant bitrate, so their content makes no difference to their size. Except for the main English audio track, which is DTS-HD Master Audio, 24 bits, 48kHz. It averages 3619kbps in the movie. If this track were empty, then presumably the amount of data required by it could fall back to the ‘core’ 1,509kbps DTS track. So by including sound in these three files, the disc wastes over 213MB by my calculation.

No criticism here: the movie scores a massive MPEG4 AVC video bitrate averaging over 33Mbps, and still only uses 31.5GB of the available 50GB. But I do find it interesting.

One other point on this disc. In the playlists for both versions of the movie the file 00127.m2ts is listed by BDinfo, and immediately followed by ‘00129.m2ts(1)’. This would suggest that they are alternatives, and indeed they turn out to be. Both are 19 seconds long and contain the Fox promo banner gag running across the screen while Bart is shooting Homer with an air rifle. The 00129.m2ts has this in French rather than English.

So it seems that the WALL-E branching technique is not necessarily required to present multiple language sections of the movie:

The Simpsons in French

Posted in Blu-ray, Disc details | Leave a comment

Warner Bros, Roadshow Entertainment and Gran Torino

Most Warner Bros Blu-ray discs are fairly basic in their organisation. They lack BD-Java, BD-Live, BonusView and so on. I’m not complaining, because Warner Bros has been pretty aggressive at getting its back catalogue out there in the new format. And its discs don’t always lack these things. For example, Batman Begins has BonusView and Bookmarks.

Gran Torino, Clint Eastwood’s most recent movie as both director and actor, is BD-Java enabled and has BD-Live. The Australian version is distributed by Roadshow Entertainment, but the disc appears to be identical to the US Warner Bros version.

Now there is something interesting about a lot of Warner Bros Blu-ray discs, including Gran Torino and many others that quite lack BD-Java: they often carry Japanese language audio and subtitles … but you won’t see them on either the disc’s setup menu, nor if you use the ‘Audio’ key on your remote control. It seems that even the BD-Java-less ones can query the ‘Menu language’ setting on the player and invoke different menu and access systems according to whether or not the player is set to Japanese.

Hardware players sold in Australia will not, as far as I’ve explored so far, allow you to set Japanese as their language. But you can in PowerDVD. So I was able to confirm this.

The Gran Torino Blu-ray when played with Japanese selected omits all the languages available except English, and adds Japanese, both for subtitles and audio. In fact it gives two different Japanese subtitles (I believe that there are two forms of writing in Japanese). It also starts up with two Japanese language text screens before the Warner Bros logo (with English it goes directly into the logo). These are the screens:

Opening screen 1 to 'Gran Torino'

Opening screen 2 to 'Gran Torino'

I wonder what they are saying.

Posted in Blu-ray, Disc details | Leave a comment