Actually, it is possible with some products to get higher quality out of the same or lower bitrate when you're using a better encoder - just because they all output bitstreams that comply with the same spec doesn't mean they are all equal, even when given comparable parameters and the same input. Not knowing what encoders BBC was using, and which they switched to, it's hard to say, but there are certainly better and worse encoders, particularly in MPEG-2 (where there's a long history and lot of variety).

That said, I'd not expect a something as radical as a 50% bitrate reduction to result in better encodes unless the original encoder was the original iDVD - the differences are rarely that extreme.

Bah - I hate his show, have never found him funny, but he is well known for being friendly and down to earth in person, unlike many of his peers - he's a lucky bastard and he knows it. I met him briefly once (working in the automotive media industry in LA) and had a similar experience as the GP, and everyone I know who's had encounters with him has said the same.

Not all celebs are total douches, and making a ton of money doing something that is basically harmless and using it to enjoy your life is forgivable in my book.

Oh, and in case it wasn't clear - I agree with the overrated network whore with pedestrian humor part.

We have ourselves as evidence that intelligent, sentient, wholy natural beings can and do exist - it's not nearly the same leap of faith to assume other creatures like us might exist.

Of course, that's difference from *believing* they do exist - that's purely conjecture, perhaps with slightly stronger ground than belief in God but nonetheless just guessing.

And when there is absolutely no way one could repay what they have taken from others? Do you think Bernie Maddof should be working off his debts, and if so how does one propose he is given the financial power to earn back those billions lost without granting him the trust he has already demonstrated he is unworthy of?

Ever noticed that most people actually remove their foot from the accelerator before applying the brakes? Ever tried stomping the brakes *while* the accelerator is floored?

There was a recent case in Southern California of an off-duty sherrifs deputy, I believe, dying (with several members of his family) in a Lexus whose accelerator was stuck (floor mats, apparently)... applying the brakes did nothing but disintegrate his brakes, enough so that witnesses said flames were licking out of the wheels. At 120mph, with the accelerator stuck, your brakes are not going to help much.

Can you please provide any example where the input of other countries was ignored by ICANN, leading to a negative outcome? Just wondering.

Lions need air.

So if I'm otherwise a male - have a penis, for instance - but cannot produce sperm at all due to a physical abnormality, I'm not a male?

Keep in mind, the Warcraft universe has a highly developed lore - while the story isn't always a major focus in WoW, it's there, it's been explored in novels, comics and other media - I'd expect the movie to be more like this, a telling of the underlying Warcraft story, than an attempt to translate the game into cinema.

The image toolkit TinEye is based on (Piximilar) is far more powerful even than TinEye. Awesome stuff, one of the best commercial CBIR engines I've seen.

If you just want to group near-identical images, which vary only by minor processing - resolution, minor color correction - there are simple, low-end tools that can do this easily. imgseek is open source and works pretty well; I also use the Windows-based VSDIF, which isn't bad for finding duplicates in various formats, scales, and color spaces (I use it for deduplicating image libraries - the corporate edition has a command line interface). Both of these tools have limits when it comes to cropping, non-right-angle rotations, whereas Piximilar and some of its competitors can handle pretty radically modified images, or recognize individual components of larger images.

Definitely not a reliable test - it'll vary significantly by the image content.

There are many cases where the version compressed initially with a lossy encoder and then recompressed with a lossless encoder will be larger than the version just compressed with the lossless encoder. For instance, a simple image of horizontal, solid color lines will compress very well with any lossless encoder; when you run it through a JPEG encoder first (at just about any quality level) it'll add a lot of noise that'll bloat the size once you compress it with a lossless encoder.

Alternately, a noisy source image will have it's noise levels softened by heavy JPEG compression, likely resulting in a *smaller* file when encoded losslessly than the source. But a clean, low-noise source image will behave differently, at least at some JPEG compression levels. Too many variables here for this to be a useful test.

There's no easy solution to this problem, at least not without making certain assumptions about the image source - for instance, if you know all compressed versions came from the same source file, with no additional processing, and were encoded with the same jpeg library then the test can be very simple.

Lossless JPEG and lossless JPEG2000 are both exactly that - lossless. Not perceptually lossless, which is what people often use to refer to high-quality, lossy JPEG/JPEG2000, or JPEG-LS. Lossless JPEG uses a PCM-like encoder, not DCT, AFAIR. Lossless JPEG and lossless JPEG2000 are, in fact, lossless, at least with regards to image data in supported color spaces. This is in part a result of *not* converting to YCrCb, since that conversion is lossy, of course. Not all Lossless JPEGs are 8bit YCrCb.

Accusoft, for one, has a toolkit for building lossless JPEG applications which supports 16bit RGB and greyscale lossless JPEG modes.

The near-lossless JPEG you're thinking of is JPEG-LS, which is perceptually lossless, and guarantees a maximum error rate that is generally neglible for almost all applications. This format gets better compression ratios than Lossless JPEG, of course.

Neither the lossless or near-lossless JPEG modes are common though, outside of niche apps. Lossless JPEG2000 is, however, since almost all JPEG2000 libraries support it alongside the lossy modes.

Just one minor correction - Vivendi is no longer Blizzard's parent; Activision Blizzard is.

Wait - I wasn't responding to any of those statements - I was responding to you saying "Apple doesn't seem to want to update Quicktime" by pointing out that Quicktime is on the verge of its biggest update in a number of years.

As for the rest - I don't use RTP/RTSP with QT anymore; it is certainly stable playing MP4 profiles it supports; Quicktime documentation is generally pretty good, I don't know where you get the idea they don't document their APIs; more stable on Mac vs. PC may be true, but I don't see the relevance of that, or why it would be surprising.

Anyway, I wasn't responding to any of those statements when I said you were wrong; you were simply wrong about the fact of updates. Now, whether the updates satisfy all parties - no, historically Apple has dropped support for less mainstream parts of Quicktime (certain codecs, sprite and Flash support, VR hotspot shenanigans, etc) without so much as a heads up to its content creator community, leaving many Quicktime content producers who used to push the limits of the format on constant edge whenever an update is pushed, as it often means an overnight scramble to fix or re-author content that no longer works thanks to dropped features. I'm no Quicktime partisan anymore, but it's certainly not suffering from lack of development right now - Quicktime X is just around the corner, no doubt offering a better consumer experience and, for Quicktime content producers who do more than audio+video, a swift kick in the nuts.

Absolutely - the notion of "submarine patents" rising up, should Theora take off, is not a new idea, and not specific to Apple. By mandating Theora in HTML5, you'd be risking the years of negotiations on the spec on the bet that there are no such patents - a bet I'd be surprised if any good Slashdot reader would take.

As others have pointed out, HTML has never mandated a specific image format reference in an IMG tag; a type of plugin referenced in OBJECT or EMBED; or the type of resource referenced in an A tag; it's outside it's scope. Let the standard focus on its scope, and let the market hash out the rest - it's not the end of the world to not have a single, mandated codec - in fact, I'd argue that having such a thing would unnecessarily limit our options - Theora is, to be kind, not the most efficient codec on the market; and the situation will likely only get worse. Don't hamstring HTML5 by hitching it to any particular codec.