There are only so many such images available for use, and the image library could fairly easily be exhausted and all of the images correctly identified at which point a bot could be used with near-100% accuracy.
OT, but I found a way to make RECAPTCHA entertaining. With two words given, I always just type one of the words, and put "fuck" for the other. The accuracy falls below 50%, but the giggles make it all worthwhile.
Below 50%? I probably average ~90%... the key is in figuring out which word you have to get correct. There’s always the button to get a different captcha if you can’t tell on the one it gave you...
Best rickroll I’ve seen was written in assembly code and instructed you to paste it into DEBUG, resulting in a never-ending loop playing the first stanza or two. I ran it in DOSBox just to be on the safe side...
Best goatse was a black PNG with the image stored in the alpha channel.
This was meant to be in reference to Carl Sagan's Arecibo Message*, which used a similar encoding scheme, but with 1679 bits forming a 2d image if plotted as rows & columns based on its prime factors (23*73)
Ah, I didn’t catch that – I thought you were just referring to the same thing that this guy was. (What with the pretty pattern and all. Did you check the gif sequences I posted?)
You don’t take the tedious but accurate route to narrow down 100+ applicants. You take the quick and inaccurate way, narrow that down to a dozen or so, and hopefully you didn’t get rid of the person who was really the right one for the job... then you have a manageable task: finding out the skills of whoever made the initial cut and selecting the right one for the job.
If a prospective employer googled her name and found a bunch of porn and spam, it would be entirely possible that they’d conclude either (a) she’s personally involved in porn or spam or more likely (b) she gave them a fake name.
At least now they’ll see that she’s a real person, that’s her real name, and although googling it might have gave porn and spam (at one point) it at least didn’t have anything to do with her.
You forgot its other 2 factors:1 & 33. You can plot the bits as a 2d image in one other, final, pattern: 1 row of 33.
If you’re interpreting it as a repeating tile pattern you might as well be complete, and the 1x tile is a non-trivial solution since you can align it in 33 different ways relative to its neighbouring tiles. And it does make someinterestingpatterns, but I still doubt that it’s meant to be interpreted that way...
Buffalo (n), the city of Buffalo, in New York; Buffalo (n), the animal, also known by the name bison; Buffalo (v), to bully, confuse, deceive, or intimidate.
The full sentence means: “Buffalo bison (which) Buffalo bison bully (themselves) bully Buffalo bison.” Or, more succinctly, Buffalo bison bully each other.
The word “irregardless” probably originated from a bastardized combination of the words “irrespective” and “regardless”. Regardless of the fairly-widespread use of the word, it is improper and should not be used.
In terrorism, you generally have a loosely-defined cause and you wage haphazard strikes against civilians and/or targets unrelated to the cause you claim to fight for. I.e., your goal is to cause widespread terror in everyone.
In a war, even a guerrilla war, you have a well-defined target or entity and you wage calculated and deliberate strikes against that target or targets directly related to the entity you’re at war with. Naturally your target should be reasonably concerned but the attacks should not cause widespread terror.
I admitted there’s a noticeable difference, but what I did claim was that the one saved with higher compression wouldn’t be noticeably bad-looking to most people. Also take into consideration, though, that I greatly exaggerated the compression to show just how far you can go and still get an image which is basically good enough.
Maybe an analogy would help to make my point.
They say a picture is worth a thousand words. Okay. Now, admittedly, the second picture lost something from the original. How many words should we agree that it’s worth? It’s still the same story, but it has less detail. Maybe it’s only 900 words... but it’s also less than 1/7th the size. Of course your authors are going to raise bloody hell if you tell them to cut 100 words out of their masterpiece, but if you can deliver 7x as much content by making that tradeoff, you might decide it’s worth it. Or you might not. It is a tradeoff, admittedly.
I’m actually a bit disappointed. I halfway expected to see demos of real WebP image files, complete with a Javascript decompressor to render onto canvas... this is Google we’re talking about, after all.
Maybe I should have said certain types of gradients. I was trying to put it in layman’s terms, though...
More specifically, it causes both ringing and blur. This, for example, is compressed all to hell to show the artifacts: http://ompldr.org/vNXAzbQ/Untitled.jpg
Check the originals... the lights are clearly visible in the JPEG original, but what’s mystifying (to me) is that for this image in particular the JPEG is much darker than the WebP version, particularly evident in the dark areas of the image. None of the other sample images have a corresponding brightness discrepancy. I actually think they might have brightened it and forgot to tell us... (and yes, I know about the possibility that the browser is doing funky things with the PNG colour profile – I’m using Firefox 3.6.9, though, and Windows Preview and GIMP both show the same thing)
Slashdot Mirror
There are only so many such images available for use, and the image library could fairly easily be exhausted and all of the images correctly identified at which point a bot could be used with near-100% accuracy.
OT, but I found a way to make RECAPTCHA entertaining. With two words given, I always just type one of the words, and put "fuck" for the other. The accuracy falls below 50%, but the giggles make it all worthwhile.
Below 50%? I probably average ~90% ... the key is in figuring out which word you have to get correct. There’s always the button to get a different captcha if you can’t tell on the one it gave you...
Eh, a few.
Best rickroll I’ve seen was written in assembly code and instructed you to paste it into DEBUG, resulting in a never-ending loop playing the first stanza or two. I ran it in DOSBox just to be on the safe side...
Best goatse was a black PNG with the image stored in the alpha channel.
Not always when they’re in the summary...
Sure, I probably should, but still...
2nd link is a PDF. Thanks for the warning...
Employers have much more to do than simply review resumes and interview candidates all day long, believe it or not.
Time & money she apparently had, and was willing to part with...
This was meant to be in reference to Carl Sagan's Arecibo Message*, which used a similar encoding scheme, but with 1679 bits forming a 2d image if plotted as rows & columns based on its prime factors (23*73)
Ah, I didn’t catch that – I thought you were just referring to the same thing that this guy was. (What with the pretty pattern and all. Did you check the gif sequences I posted?)
You don’t take the tedious but accurate route to narrow down 100+ applicants. You take the quick and inaccurate way, narrow that down to a dozen or so, and hopefully you didn’t get rid of the person who was really the right one for the job... then you have a manageable task: finding out the skills of whoever made the initial cut and selecting the right one for the job.
Look at it from her perspective...
If a prospective employer googled her name and found a bunch of porn and spam, it would be entirely possible that they’d conclude either (a) she’s personally involved in porn or spam or more likely (b) she gave them a fake name.
At least now they’ll see that she’s a real person, that’s her real name, and although googling it might have gave porn and spam (at one point) it at least didn’t have anything to do with her.
Maybe she prefers it this way...
For that matter, how do you know this wasn’t her plan all along...?
She said: "That tattoo doesn't say 'Tiny'. It says 'Ticonderoga New York'.
Oh, I get it... they abbreviated it because it didn’t fit. (Ti NY)
You forgot its other 2 factors:1 & 33. You can plot the bits as a 2d image in one other, final, pattern: 1 row of 33.
If you’re interpreting it as a repeating tile pattern you might as well be complete, and the 1x tile is a non-trivial solution since you can align it in 33 different ways relative to its neighbouring tiles. And it does make some interesting patterns, but I still doubt that it’s meant to be interpreted that way...
That string of words has no meaning.
It most certainly has a meaning.
Buffalo (n), the city of Buffalo, in New York;
Buffalo (n), the animal, also known by the name bison;
Buffalo (v), to bully, confuse, deceive, or intimidate.
The full sentence means: “Buffalo bison (which) Buffalo bison bully (themselves) bully Buffalo bison.” Or, more succinctly, Buffalo bison bully each other.
The word “irregardless” probably originated from a bastardized combination of the words “irrespective” and “regardless”. Regardless of the fairly-widespread use of the word, it is improper and should not be used.
Terrorism != act of war.
In terrorism, you generally have a loosely-defined cause and you wage haphazard strikes against civilians and/or targets unrelated to the cause you claim to fight for. I.e., your goal is to cause widespread terror in everyone.
In a war, even a guerrilla war, you have a well-defined target or entity and you wage calculated and deliberate strikes against that target or targets directly related to the entity you’re at war with. Naturally your target should be reasonably concerned but the attacks should not cause widespread terror.
I admitted there’s a noticeable difference, but what I did claim was that the one saved with higher compression wouldn’t be noticeably bad-looking to most people. Also take into consideration, though, that I greatly exaggerated the compression to show just how far you can go and still get an image which is basically good enough.
Maybe an analogy would help to make my point.
They say a picture is worth a thousand words. Okay. Now, admittedly, the second picture lost something from the original. How many words should we agree that it’s worth? It’s still the same story, but it has less detail. Maybe it’s only 900 words... but it’s also less than 1/7th the size. Of course your authors are going to raise bloody hell if you tell them to cut 100 words out of their masterpiece, but if you can deliver 7x as much content by making that tradeoff, you might decide it’s worth it. Or you might not. It is a tradeoff, admittedly.
I’ll accept that argument, but from a practical perspective there’s not much difference.
And you’re delusional if you expect me to know important details of your problem that you didn’t mention in your original post...
They’re not any darker in IE (just checked)... his mind must have been playing tricks on him.
I’m actually a bit disappointed. I halfway expected to see demos of real WebP image files, complete with a Javascript decompressor to render onto canvas... this is Google we’re talking about, after all.
I've just had the (dis)pleasure of programmatically converting JPEG2000 images to JPEG and bitmaps
Would using IrfanView to batch-convert them be too easy, or something?
Well, the original source image was JPEG, but that in the left column ain’t it... they resampled both for their blog...
E.g. the real images, not the scaled-down ones, for the football player (Cato June) are:
2_original.jpg (677,662 bytes)
2_webp.png (164,910 bytes, in the WebP image... 877,967 bytes as a PNG)
Maybe I should have said certain types of gradients. I was trying to put it in layman’s terms, though...
More specifically, it causes both ringing and blur. This, for example, is compressed all to hell to show the artifacts: http://ompldr.org/vNXAzbQ/Untitled.jpg
Check the originals... the lights are clearly visible in the JPEG original, but what’s mystifying (to me) is that for this image in particular the JPEG is much darker than the WebP version, particularly evident in the dark areas of the image. None of the other sample images have a corresponding brightness discrepancy. I actually think they might have brightened it and forgot to tell us... (and yes, I know about the possibility that the browser is doing funky things with the PNG colour profile – I’m using Firefox 3.6.9, though, and Windows Preview and GIMP both show the same thing)
(warning – they’re rather huge)
http://ompldr.org/vNXAzZA/7_original.jpg (3.5 MB)
http://ompldr.org/vNXAzZQ/7_webp.png (13 MB)