Page 437 of 2684

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:21 pm UTC
by Pikrass
NoMouse wrote:Well, we're all cheating anyway. We are storing the old newpixes and writing apps that allow to watch them over and over... That's not how the One True Comic was intended to be read.

I'm actually pretty sure Randall knew we will do such things.

Well, Randall keeps all the images on the server. He could as well have deleted all the past ones, or made them inaccessible.
But we can still access them all provided we have their hash, and he even changes them after their Time was over.

In fact, now that I think about It, this simple fact is the biggest evidence in favor of the loop theory.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:26 pm UTC
by KarMann
mscha wrote:The last few fades:
#676767 => #6e6e6e => #757575 => #7b7b7b. That's +7, +7, +6.
(Raw data, not mangled by browsers or overpriced Adobe cr*pware.)

I've been wondering whether the curve would decelerate after it passed #7F7F7F or #808080. Of course, this +6 isn't specific, definitive evidence that it's decelerating yet; it could just be cruising along at about 6 2/3, and happened to come up short this particular NewPix. Looking forward to analyzing the next few frames, except Outside may call me away for a Time. :(

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:28 pm UTC
by BytEfLUSh
Pikrass wrote:he even changes them after their Time was over.


I probably missed something; which Frames were changed and what was changed?
Also, does anyone actually have a running script that monitors the Past for changes?

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:33 pm UTC
by spamjam
Pikrass wrote:Well, Randall keeps all the images on the server. He could as well have deleted all the past ones, or made them inaccessible.
But we can still access them all provided we have their hash, and he even changes them after their Time was over.

In fact, now that I think about It, this simple fact is the biggest evidence in favor of the loop theory.

I see what you did there.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:35 pm UTC
by Goggalor
Another suggestion for the soundtrack of Time. Towers by Bon Iver. It might be a bit of a stretch, but it's a beautiful song and it's about equating love1 with an object (such as a tower2), but at the end of the day everything will fall apart and you have to move on.3
1or really any other strong emotion.
2or sandcastle.
3I'm not really that good at analysis or wording things, so take this as you like

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:36 pm UTC
by Latent22
The problem with the colour differences seem to me to be something like this. At a low level the image stores the colour and brightness of the pixels (PNG's support a few methods of storing this data). Now lets say a pixel is around 90% black. When it is displayed it has to be turned into amounts of red, green and blue to work on our RGB monitors. You would think that the easy way to do this is just use 25,25,25 which is 10% intensity. However it has been found that many images when converted to RGB like this look faded out and horrible. This is to do with how we humans see gradients of light and how the monitors we use work etc. When we see 10% RGB we see it as quite a noticible grey colour. Many of the pictures and videos have dark shadow areas and to us it looks a lot more real if the blacks are really black which makes the bright colours stand out more etc. So to make things look more 'Real' we have colour profiles and Gamma levels. The PNG has a default gamma curve for it somewhere and when we display the 90% black pixel instead of using 25/255,25/255,25/255 RGB the values first pass though a standard gamma correction step and become something like 05,05,05 instead (i'm just making this number up as a guess). Note that as it gets lighter it gets closer and closer to the straight linear conversion you would expect.

Now we get to browsers like chrome and they have the gamma correction that most people take for granted turned off for PNG file it seems. This means many images will display with washed out shadow colours. Firefox also choose a different colour profile it seems which changes things a bit as well. Also note it seems file formats like JPG don't have the same built in colour correction problems and display the same no matter what browser. PNG is quite a variable format and it is probably possible to get it to display the same in most browsers if you choose the right encoding/profile settings when generating the file.

Also of note is that different graphics editing programs also do the PNG colour converstion differently by default. So Photoshop can give different results than GIMP by default. This makes it all very confusing.


BytEfLUSh wrote:
Pikrass wrote:he even changes them after their Time was over.


I probably missed something; which Frames were changed and what was changed?
Also, does anyone actually have a running script that monitors the Past for changes?


The only frame we have proof he edited after was frame 20. The original had Cueballs head facing one way and there was a problem with it as there was a gap between Cueball's neck and his head. He Redrew the head and made it face towards Megan and update the frame 15 hours after it Donged.

up until I posted the below graph on the 22nd he left the modified time stamps in the files so we could actually see when he changed them. But someone must have told him and a few hours later he hid all the timestamp data with a fixed value. But we still have the original data and can see when he changed many of the frames. Like the "River is Small - Sea is Big" frame was changed 2 minutes before it was posted! But now it's all hidden and I don't know of anyone doing any checking now.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:41 pm UTC
by NoMouse
Has anyone already come up with a theory that this is actually a very bright light caused by nuclear explosion and at the end the castle will be destroyed in an epic way by a shock wave? If not, this is my theory now.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:47 pm UTC
by ChronosDragon
NoMouse wrote:Has anyone already come up with a theory that this is actually a very bright light caused by nuclear explosion and at the end the castle will be destroyed in an epic way by a shock wave? If not, this is my theory now.


Well in that case, I guess Megan and Cueball are just trying to be Cool Guys. Cool Guys Don't Look at Explosions, after all. Also, radiation poisoning sucks, even for stick figures.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:48 pm UTC
by buffygirl
NoMouse wrote:Has anyone already come up with a theory that this is actually a very bright light caused by nuclear explosion and at the end the castle will be destroyed in an epic way by a shock wave? If not, this is my theory now.

Yes! That's what I meant by lighting change or glare or sunrise or some such thing! It's definitely a nuclear explosion! We're saved!

hey, wait a sec ....

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:49 pm UTC
by AluisioASG
NoMouse wrote:Has anyone already come up with a theory that this is actually a very bright light caused by nuclear explosion and at the end the castle will be destroyed in an epic way by a shock wave? If not, this is my theory now.

I like that.
But it's going to take some time, right? We still haven't reached the “MY EYES! THEY BURN! AHHHHH!” stage yet.

EDIT: BlitzGirl, you're late!
EDIT2: Now that I think about it, it was a good thing to put on those sunglasses a couple pages ago.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:52 pm UTC
by treadman
John Keats wrote:...
That I might drink, and leave the world unseen,
And with thee fade away
...

However this turns out, it has truly been a pleasure immersing with all of you in the OTC. Cheers, everyone!

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:53 pm UTC
by doeanethum
buffygirl wrote:
Eliram wrote:- BuffyGirl, is there still room in your queue for a hatless spider?


Okies, you're added to the Pre-Hat-Queue-Mob-Mob (which is the holding spot for any new hats until I complete those in the hat-queue-mob).


So, if one was eligible for cardinality, would one get assigned to some other queue if one was to request a hat?a

And yay, my post count is not 1 anymore, this will make me jump 380 places on the post count list!

a: I actually did already, but since that was my first post it got stuck in time and was probably not read at all...

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:53 pm UTC
by Pikrass
BytEfLUSh wrote:
Pikrass wrote:he even changes them after their Time was over.


I probably missed something; which Frames were changed and what was changed?
Also, does anyone actually have a running script that monitors the Past for changes?

There. Go back in Time.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:55 pm UTC
by fhorn
I don't think we need to say our goodbyes yet.... I think we can Wait for it.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:56 pm UTC
by descor
fhorn wrote:I don't think we need to say our goodbyes yet.... I think we can Wait for it.

I'm not saying goodbye until the bitter end and I know that the comic won't loop or fade back in or any other possible ending.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 9:58 pm UTC
by NoMouse
buffygirl wrote:
NoMouse wrote:Has anyone already come up with a theory that this is actually a very bright light caused by nuclear explosion and at the end the castle will be destroyed in an epic way by a shock wave? If not, this is my theory now.

Yes! That's what I meant by lighting change or glare or sunrise or some such thing! It's definitely a nuclear explosion! We're saved!

hey, wait a sec ....
Or, it could be this. That's not so fatal and has similar effect.

Spoiler:
I hope no one will say "Hey, that's a spoiler!" now. I seriously doubt there's someone who is now watching first or second season of Lost...

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:01 pm UTC
by Bytes
Been away for a bit, where is blitzgirl up to?

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:01 pm UTC
by Valarya
Goodbyong.

Image

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:04 pm UTC
by slinches
Still fading #838383*

*by GIMP reckoning. a.k.a. value 131

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:07 pm UTC
by Pikrass
I hope LaPetite will come and take the flag before it gets stolen by the sea.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:08 pm UTC
by ChronosDragon
The sea is almost up to the same level as the sand-like substance on the left side of the frame.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:13 pm UTC
by mittfh
HSV 0 / 0 / 51
RGB grayscale Hex 83, Dec 131, Oct 461, Quar 2003, Bin 100110001 (since I was doing hex and dec anyway, thought I'd spice things up with a few other bases).

And the tower's collapsed. If Megan and / or Cueball don't return within the next few newpix, they might lose one of their flags...

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:13 pm UTC
by pelrigg
Okay.. so here's another noob de-cloaking (err..) delurking here at the End Of Time {no, say it ain't so}. I started looking in from time - to - Time at 1190 while it was still in the "what the" phase when I finally realized it was changing (been reading xkcd for over an outside year). I've started checking into this thread ever since and must say I've enjoyed the whole thing.
The various people have become familiar... AluisioASG; buffygirl; blitzgirl (go B-girl, go); the posters of charts and graphs and phlip (who's posts helped me gather the newpixs in the beginning before I looked at explain-wiki). The semi-off-course discussions, the haikus, the attempts to figure out the hashes. It has been fun (and thanks for the fish).
I will add that OTC has become quite a big part (not as big as the sea*) of my on-line fun (factoid: I have several more places to {haunt} saved now) and I've saved the entire run of the OTC, although I'm not sure yet what I'll do with it (them?).

Perhaps I could mention a song to go with the "Fade-out"???? (It goes with my OF** status) --- Simon & Garfunkel: "Old Friends" ---

Old Friends
Sat on their park bench
like bookends.

....

The old men
lost in their overcoats,
waiting for the sunset.

...

Can you imagine us
years from today,
sharing a park bench quietly?


----- Actually, thinking about it, after rummaging through the lyrics, pretty much the whole "Bookends" album (okay: disc) fits in with the Time motif: "Bookends Theme"; "Overs"; "Fakin' It"; "A Hazy Shade Of Winter" (excuse please there's something in my eye and I need to put something on the jukebox)


signing off


* which is bigger
** Old Fart

ps This is only the third time I've ever been on ANY b-board. So that's how inspiring this has been.......

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:24 pm UTC
by cmyk
Rick_LANL wrote:
Spoiler:
cmyk wrote:
Rick_LANL wrote:
cmyk wrote:
Tass wrote:[spoiler]
cmyk wrote:Still on track for around frame 962/963.



Are you using chrome values? Because it is not at 50% yet.


Nuh-uh.

I'm downloading the original png from the hash, opening in Photoshop and making sure not to apply any color management (and its gamma set to 1.0 — linear).

Seems other image editors are opening the png as Index Color (with some default color lookup table), which, despite its 256 swatch values, is a bit less accurate than true 8-bit grayscale, so it rounds the value to the closest swatch.

Then, I sample the brightness level of the sand, using the HSB (Hue, Saturation, Brightness) color model, which gives me a percentage from black to white (0 - 100%) under "B".

With that, I just simply plug that percent into its appropriate frame for the graph data in Numbers, and fit an exponential curve to it.

Either way, the color values sampled are absolute, despite any "corrective" gamma function curves being applied to make it appear darker/lighter.


The .png file encodes a grayscale image, with integer values from 0 to 255 at each pixel. The value of the formerly-black pixels (all of which are identical, except near the edges where antialiasing has been applied) seems to me to be a much better measure of whiteness than converting 1-D intensity values to a 3-D colorspace model and then taking one of the three values. Right now it's 110, or about 43.14%.


Sciscitor wrote:
cmyk wrote:Nuh-uh.

I'm downloading the original png from the hash, opening in Photoshop and making sure not to apply any color management (and its gamma set to 1.0 — linear).

Seems other image editors are opening the png as Index Color (with some default color lookup table), which, despite its 256 swatch values, is a bit less accurate than true 8-bit grayscale, so it rounds the value to the closest swatch.


I am anything but an expert on colour spaces and icc profiles, but I took a look at the raw thing.
According to the output
[spoiler]

Code: Select all

\x00\x00\x03\x90ADBE\x02\x10\x00\x00prtrGRAYXYZ \x07\xcf\x00\x06\x00\x03\x00\x00\x00\x00\x00\x00acspAPPL\x00\x00\x00\x00none\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\xf6\xd6\x00\x01\x00\x00\x00\x00\xd3-ADBE\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x05cprt\x00\x00\x00\xc0\x00\x00\x002desc\x00\x00\x00\xf4\x00\x00\x00gwtpt\x00\x00\x01\\\x00\x00\x00\x14bkpt\x00\x00\x01p\x00\x00\x00\x14kTRC\x00\x00\x01\x84\x00\x00\x02\x0ctext\x00\x00\x00\x00Copyright 1999 Adobe Systems Incorporated\x00\x00\x00desc\x00\x00\x00\x00\x00\x00\x00\rDot Gain 20%\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00XYZ \x00\x00\x00\x00\x00\x00\xf6\xd6\x00\x01\x00\x00\x00\x00\xd3-XYZ \x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00curv\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x00\x10\x00 \x000\x00@\x00P\x00a\x00\x7f\x00\xa0\x00\xc5\x00\xec\x01\x17\x01D\x01u\x01\xa8\x01\xde\x02\x16\x02R\x02\x90\x02\xd0\x03\x13\x03Y\x03\xa1\x03\xec\x049\x04\x88\x04\xda\x05.\x05\x85\x05\xde\x069\x06\x96\x06\xf6\x07W\x07\xbb\x08"\x08\x8a\x08\xf4\ta\t\xd0\nA\n\xb4\x0b)\x0b\xa0\x0c\x1a\x0c\x95\r\x12\r\x92\x0e\x13\x0e\x96\x0f\x1c\x0f\xa3\x10,\x10\xb8\x11E\x11\xd4\x12e\x12\xf8\x13\x8d\x14$\x14\xbd\x15W\x15\xf4\x16\x92\x172\x17\xd4\x18x\x19\x1e\x19\xc6\x1ao\x1b\x1b\x1b\xc8\x1cv\x1d\'\x1d\xda\x1e\x8e\x1fD\x1f\xfc \xb5!q"."\xed#\xad$p%4%\xf9&\xc1\'\x8a(U)")\xf0*\xc0+\x92,e-:.\x11.\xea/\xc40\xa01}2\\3=4\x1f5\x035\xe96\xd07\xb98\xa49\x90:~;m<^=Q>E?;@3A,B&C"D E\x1fF G#H\'I-J4K<LGMSN`OoP\x7fQ\x91R\xa5S\xbaT\xd1U\xe9W\x02X\x1eY:ZX[x\\\x99]\xbc^\xe0`\x06a-bVc\x80d\xace\xd9g\x08h8iij\x9dk\xd1m\x07n?oxp\xb2q\xees+tju\xaav\xecx/ytz\xba|\x01}J~\x95\x7f\xe1\x81.\x82|\x83\xcd\x85\x1e\x86q\x87\xc5\x89\x1b\x8ar\x8b\xcb\x8d%\x8e\x81\x8f\xdd\x91<\x92\x9b\x93\xfd\x95_\x96\xc3\x98(\x99\x8f\x9a\xf7\x9c`\x9d\xcb\x9f7\xa0\xa5\xa2\x14\xa3\x85\xa4\xf6\xa6i\xa7\xde\xa9T\xaa\xcb\xacD\xad\xbe\xaf9\xb0\xb6\xb24\xb3\xb4\xb54\xb6\xb7\xb8:\xb9\xbf\xbbE\xbc\xcd\xbeV\xbf\xe0\xc1l\xc2\xf9\xc4\x87\xc6\x17\xc7\xa8\xc9;\xca\xce\xccc\xcd\xfa\xcf\x92\xd1+\xd2\xc5\xd4a\xd5\xfe\xd7\x9c\xd9<\xda\xdd\xdc\x7f\xde#\xdf\xc8\xe1n\xe3\x16\xe4\xbf\xe6i\xe8\x14\xe9\xc1\xebo\xed\x1f\xee\xd0\xf0\x82\xf25\xf3\xea\xf5\xa0\xf7W\xf9\x10\xfa\xca\xfc\x85\xfeA\xff\xff


It is an Adobe ICC-Profile with a LUT (Look Up Table) for printing. So: No idea what all this means, but I am pretty sure it's not just an ordinary grayscale image.


Well, there's grayscale, where each pixel can be a intensity value from 0-255 (8 bits per pixel).

Then there's Index, where each pixel is assigned a number that pertains to one of the 256 color swatches, preordained in the color lookup table:
Image
Image

The important difference, is that a true 8-bit grayscale can have an absolute intensity value assigned, per pixel, so if you're employing color management (such as the ubiquitous sRGB gamma curve correction), the colors will be displayed via depending on your monitor gamma, and/or your color workspace gamma. The values can also be manipulated using Curves, or Levels, etc.

Not so with Index Color. Which is fine, accept, do the swatches for a grayscale proceed in a linear scale, or is there already some gamma curve "baked" into the swatches? Can't say for sure, which makes it harder to analyze for absolute values (unless you load your own CLUT, with a linear scale, 1.0 gamma).

(you can see, if you count from 0 to the 12th swatch (see eyedropper sample #1 - Idx:12) for the Mac OS System CLUT palette, you'll see a pink is assigned for the sand/sea color.)

ALLL that said, not sure if it really makes that much of a difference for this purpose! :mrgreen:


When I open the .png in Matlab, I get a 395 x 553 array of 1-byte integers. When I open color .pngs, I would get an M x N x 3 3-D array of 1-byte integers, the RGB values at each pixel. While I can't speak to the low-level details of how Matlab opens a .png file, I am postulating that It is saved by Him as a grayscale image.

Now, the question if what exactly a grayscale value of 117 *means* is more interesting. 117/255 as many photons as pure white? Maybe. And how that gets interpreted by the visual cortex as a shade of gray is almost certainly organism-dependent, even leaving out display hardware (and software) differences. So there's ambiguity if one chooses to see it.


Right on. Yeh, for an 8bpp grayscale image each pixel (which in this color model, can be assigned an intensity value 0-255), obviously, records 1 byte = 8 bits per... Hence the designation. Of course, if converted to an 8-bit RGB, you'll have an 8bpp capacity per color channel, so, 256 * 3 channels / 24 bpp, total.

If a grayscale image is converted to an Index Color Table, the intensity value for that pixel, will need to be mapped to the closest swatch square in regards to its luminance. Technically, you can easily create your own grayscale CLUTs, if the default isn't adequate enough. I've done my own, when getting fancy back in my print/design days.

This, of of course, saves on files size for images that are suitable for using a max of 256 colors, as the pixel (or cluster of pixels) only need to know to which swatch they're referring to. So, in Index, if it's 117, it's going to assume the color of square #117, no matter what's there. So, essentially, you can create color images that are a third, or less in size that 8-bit rgb images (albeit, usually with a quality trade-off, dithering and all that).

With 8bpp grayscale, you can be sure that stripping all color corrections, and gamma curves will display a linear scale (or vice versa, as conversion between spaces is trivial). You just need to make sure your monitor is calibrated, your working space is set appropriately, are assigning or embedding the right profile for the job, and not wearing those kickass blue blockers.

Some graphics, CG doods, like myself, like to mix/adjust colors in linear, and output as sRGB, as the colors combine far more intuitively, and the conversion to sRGB ensures everyone else will see, approximately, what I see.

And we all slappeth ourselves with the Holy Rainbow Trout of the Gamut Sea, and we all sayeth... Amen.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:27 pm UTC
by BytEfLUSh
Latent22 wrote:The only frame we have proof he edited after was frame 20. The original had Cueballs head facing one way and there was a problem with it as there was a gap between Cueball's neck and his head. He Redrew the head and made it face towards Megan and update the frame 15 hours after it Donged.

up until I posted the below graph on the 22nd he left the modified time stamps in the files so we could actually see when he changed them. But someone must have told him and a few hours later he hid all the timestamp data with a fixed value. But we still have the original data and can see when he changed many of the frames. Like the "River is Small - Sea is Big" frame was changed 2 minutes before it was posted! But now it's all hidden and I don't know of anyone doing any checking now.



Thank you!

I think we should set up a script that GETs every frame (with a 5-30 sec pause between each GET; optimize at will), store the md5 hash of each image in a DB and then keep GETting those PNGs over and over again and comparing the results. If there is a change, an alert should sound off. I don't think it would overload the server, but I also don't believe I could write that script easily. :)

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:32 pm UTC
by htom
cmyk wrote:And we all slappeth ourselves with the Holy Rainbow Trout of the Gamut Sea, and we all sayeth... Amen.


Aha! That's what I'm doing wrong. I was taught to slap the output device with the trout!

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:36 pm UTC
by cmyk
htom wrote:
cmyk wrote:And we all slappeth ourselves with the Holy Rainbow Trout of the Gamut Sea, and we all sayeth... Amen.


Aha! That's what I'm doing wrong. I was taught to slap the output device with the trout!


We've all been there. It ain't pretty.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:41 pm UTC
by buffygirl
doeanethum wrote:
buffygirl wrote:Okies, you're added to the Pre-Hat-Queue-Mob-Mob (which is the holding spot for any new hats until I complete those in the hat-queue-mob).

So, if one was eligible for cardinality, would one get assigned to some other queue if one was to request a hat?a

no. but you are, indeed, added to the phqmm.

doeanethum wrote:a: I actually did already, but since that was my first post it got stuck in time and was probably not read at all...

oops, sorry, missed that!

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:43 pm UTC
by Eliram
BytEfLUSh wrote:
Latent22 wrote:The only frame we have proof he edited after was frame 20. The original had Cueballs head facing one way and there was a problem with it as there was a gap between Cueball's neck and his head. He Redrew the head and made it face towards Megan and update the frame 15 hours after it Donged.

up until I posted the below graph on the 22nd he left the modified time stamps in the files so we could actually see when he changed them. But someone must have told him and a few hours later he hid all the timestamp data with a fixed value. But we still have the original data and can see when he changed many of the frames. Like the "River is Small - Sea is Big" frame was changed 2 minutes before it was posted! But now it's all hidden and I don't know of anyone doing any checking now.



Thank you!

I think we should set up a script that GETs every frame (with a 5-30 sec pause between each GET; optimize at will), store the md5 hash of each image in a DB and then keep GETting those PNGs over and over again and comparing the results. If there is a change, an alert should sound off. I don't think it would overload the server, but I also don't believe I could write that script easily. :)


No need for updating the entire graph, I'm interested if the timestamp of the recent LaPetite image is different from the timestamps of the images before/after.
So I went and checked:
Frame 948 - 2013-04-22 01:07:25 (-4)
Frame 949 - 2013-04-22 01:07:25 (-4)
Frame 950 - 2013-04-22 01:07:25 (-4) La Petite
Frame 951 - 2013-04-22 01:07:25 (-4)
Frame 952 - 2013-04-22 01:07:25 (-4)

No change. Hmm.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:45 pm UTC
by cmyk
descor wrote:
cmyk wrote:
Spoiler:
Rick_LANL wrote:
cmyk wrote:
Tass wrote:
cmyk wrote:Still on track for around frame 962/963.



Are you using chrome values? Because it is not at 50% yet.


Nuh-uh.

I'm downloading the original png from the hash, opening in Photoshop and making sure not to apply any color management (and its gamma set to 1.0 — linear).

Seems other image editors are opening the png as Index Color (with some default color lookup table), which, despite its 256 swatch values, is a bit less accurate than true 8-bit grayscale, so it rounds the value to the closest swatch.

Then, I sample the brightness level of the sand, using the HSB (Hue, Saturation, Brightness) color model, which gives me a percentage from black to white (0 - 100%) under "B".

With that, I just simply plug that percent into its appropriate frame for the graph data in Numbers, and fit an exponential curve to it.

Either way, the color values sampled are absolute, despite any "corrective" gamma function curves being applied to make it appear darker/lighter.


The .png file encodes a grayscale image, with integer values from 0 to 255 at each pixel. The value of the formerly-black pixels (all of which are identical, except near the edges where antialiasing has been applied) seems to me to be a much better measure of whiteness than converting 1-D intensity values to a 3-D colorspace model and then taking one of the three values. Right now it's 110, or about 43.14%.


Sciscitor wrote:
cmyk wrote:Nuh-uh.

I'm downloading the original png from the hash, opening in Photoshop and making sure not to apply any color management (and its gamma set to 1.0 — linear).

Seems other image editors are opening the png as Index Color (with some default color lookup table), which, despite its 256 swatch values, is a bit less accurate than true 8-bit grayscale, so it rounds the value to the closest swatch.


I am anything but an expert on colour spaces and icc profiles, but I took a look at the raw thing.
According to the output

Code: Select all

\x00\x00\x03\x90ADBE\x02\x10\x00\x00prtrGRAYXYZ \x07\xcf\x00\x06\x00\x03\x00\x00\x00\x00\x00\x00acspAPPL\x00\x00\x00\x00none\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\xf6\xd6\x00\x01\x00\x00\x00\x00\xd3-ADBE\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x05cprt\x00\x00\x00\xc0\x00\x00\x002desc\x00\x00\x00\xf4\x00\x00\x00gwtpt\x00\x00\x01\\\x00\x00\x00\x14bkpt\x00\x00\x01p\x00\x00\x00\x14kTRC\x00\x00\x01\x84\x00\x00\x02\x0ctext\x00\x00\x00\x00Copyright 1999 Adobe Systems Incorporated\x00\x00\x00desc\x00\x00\x00\x00\x00\x00\x00\rDot Gain 20%\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00XYZ \x00\x00\x00\x00\x00\x00\xf6\xd6\x00\x01\x00\x00\x00\x00\xd3-XYZ \x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00curv\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x00\x10\x00 \x000\x00@\x00P\x00a\x00\x7f\x00\xa0\x00\xc5\x00\xec\x01\x17\x01D\x01u\x01\xa8\x01\xde\x02\x16\x02R\x02\x90\x02\xd0\x03\x13\x03Y\x03\xa1\x03\xec\x049\x04\x88\x04\xda\x05.\x05\x85\x05\xde\x069\x06\x96\x06\xf6\x07W\x07\xbb\x08"\x08\x8a\x08\xf4\ta\t\xd0\nA\n\xb4\x0b)\x0b\xa0\x0c\x1a\x0c\x95\r\x12\r\x92\x0e\x13\x0e\x96\x0f\x1c\x0f\xa3\x10,\x10\xb8\x11E\x11\xd4\x12e\x12\xf8\x13\x8d\x14$\x14\xbd\x15W\x15\xf4\x16\x92\x172\x17\xd4\x18x\x19\x1e\x19\xc6\x1ao\x1b\x1b\x1b\xc8\x1cv\x1d\'\x1d\xda\x1e\x8e\x1fD\x1f\xfc \xb5!q"."\xed#\xad$p%4%\xf9&\xc1\'\x8a(U)")\xf0*\xc0+\x92,e-:.\x11.\xea/\xc40\xa01}2\\3=4\x1f5\x035\xe96\xd07\xb98\xa49\x90:~;m<^=Q>E?;@3A,B&C"D E\x1fF G#H\'I-J4K<LGMSN`OoP\x7fQ\x91R\xa5S\xbaT\xd1U\xe9W\x02X\x1eY:ZX[x\\\x99]\xbc^\xe0`\x06a-bVc\x80d\xace\xd9g\x08h8iij\x9dk\xd1m\x07n?oxp\xb2q\xees+tju\xaav\xecx/ytz\xba|\x01}J~\x95\x7f\xe1\x81.\x82|\x83\xcd\x85\x1e\x86q\x87\xc5\x89\x1b\x8ar\x8b\xcb\x8d%\x8e\x81\x8f\xdd\x91<\x92\x9b\x93\xfd\x95_\x96\xc3\x98(\x99\x8f\x9a\xf7\x9c`\x9d\xcb\x9f7\xa0\xa5\xa2\x14\xa3\x85\xa4\xf6\xa6i\xa7\xde\xa9T\xaa\xcb\xacD\xad\xbe\xaf9\xb0\xb6\xb24\xb3\xb4\xb54\xb6\xb7\xb8:\xb9\xbf\xbbE\xbc\xcd\xbeV\xbf\xe0\xc1l\xc2\xf9\xc4\x87\xc6\x17\xc7\xa8\xc9;\xca\xce\xccc\xcd\xfa\xcf\x92\xd1+\xd2\xc5\xd4a\xd5\xfe\xd7\x9c\xd9<\xda\xdd\xdc\x7f\xde#\xdf\xc8\xe1n\xe3\x16\xe4\xbf\xe6i\xe8\x14\xe9\xc1\xebo\xed\x1f\xee\xd0\xf0\x82\xf25\xf3\xea\xf5\xa0\xf7W\xf9\x10\xfa\xca\xfc\x85\xfeA\xff\xff


It is an Adobe ICC-Profile with a LUT (Look Up Table) for printing. So: No idea what all this means, but I am pretty sure it's not just an ordinary grayscale image.


Well, there's grayscale, where each pixel can be a intensity value from 0-255 (8 bits per pixel).

Then there's Index, where each pixel is assigned a number that pertains to one of the 256 color swatches, preordained in the color lookup table:
Image
Image

The important difference, is that a true 8-bit grayscale can have an absolute intensity value assigned, per pixel, so if you're employing color management (such as the ubiquitous sRGB gamma curve correction), the colors will be displayed via depending on your monitor gamma, and/or your color workspace gamma. The values can also be manipulated using Curves, or Levels, etc.

Not so with Index Color. Which is fine, accept, do the swatches for a grayscale proceed in a linear scale, or is there already some gamma curve "baked" into the swatches? Can't say for sure, which makes it harder to analyze for absolute values (unless you load your own CLUT, with a linear scale, 1.0 gamma).

(you can see, if you count from 0 to the 12th swatch (see eyedropper sample #1 - Idx:12) for the Mac OS System CLUT palette, you'll see a pink is assigned for the sand/sea color.)

ALLL that said, not sure if it really makes that much of a difference for this purpose! :mrgreen:

If there is one thing this Needle-pulled thing has taught me, it's that you do not question cmyk when it comes to graphics!


Shit, see what happens when someone brings up colometrics and color spaces, of which I'm no expert? Don't bring up color theory either, I could digress into a nerd-spiral, a technicolor abomination not even hippies on LSD would find pleasant. Okay, maybe they would, god bless 'em.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:45 pm UTC
by Jonas79
buffygirl wrote:For Jonas79

no, this is not sacrilege! And remember, I'm in a hurry!


The hat is beautiful and I'll wear it with pride when I get to a computer.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:46 pm UTC
by buffygirl
NoMouse wrote:
buffygirl wrote:
NoMouse wrote:Has anyone already come up with a theory that this is actually a very bright light caused by nuclear explosion and at the end the castle will be destroyed in an epic way by a shock wave? If not, this is my theory now.

Yes! That's what I meant by lighting change or glare or sunrise or some such thing! It's definitely a nuclear explosion! We're saved!

hey, wait a sec ....
Or, it could be this. That's not so fatal and has similar effect.

Spoiler:
I hope no one will say "Hey, that's a spoiler!" now. I seriously doubt there's someone who is now watching first or second season of Lost...

uhhhh ..... Lost is still on my to-do list (yes, I missed it first time round! I'm sorry!) So thank you for the spoiler heads-up, cuz it meant that I didn't follow the link! Woot!

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:47 pm UTC
by katakombe
descor wrote:I'm not saying goodbye until the bitter end and I know that the comic won't loop or fade back in or any other possible ending.


i'm not saying goodbye either, after all i just said hello.
so, while i agree with you that the story is most likely done, i'm still not entirely convinced that the OTC won't loop back. if it doesn't, people who will be new to XKCD and go through the archives will find a white, empty frame, nothing else. they'll have no chance to Wait for it. so, in order to provide future readers with a comic that makes sense, Randall might still let it loop.
i guess we will have to Wait and see (sea?).

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 10:59 pm UTC
by Angelastic
buffygirl wrote:uhhhh ..... Lost is still on my to-do list (yes, I missed it first time round! I'm sorry!) So thank you for the spoiler heads-up, cuz it meant that I didn't follow the link! Woot!

I get the impression that it's yet another of those shows whose entire premise is a tantalising overarching mystery but that got cancelled before the mystery was solved, only this one is worse than usual since it went on for six seasons first. So you can save yourself the time and frustration, and take it off the to-do list. Congratulations, you just gained nearly 85 timeframes of free Time!

Signed,
The girl who doesn't have a TV and doesn't start watching any such series until she knows it's been properly ended (ever since Dark Angel) and who therefore might be less prepared to deal with the end of Time than the rest of you.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 11:02 pm UTC
by Eliram
buffygirl wrote:uhhhh ..... Lost is still on my to-do list (yes, I missed it first time round! I'm sorry!) So thank you for the spoiler heads-up, cuz it meant that I didn't follow the link! Woot!

I think it won't be a spoiler to say that "Lost", just like the OTC (as we perceive it at this time), may not give full satisfaction with its ending, but the journey is well worth it. :-)

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 11:03 pm UTC
by buffygirl
For Eternal Density

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 11:04 pm UTC
by AK49BWL
What, no GONG?

Image

0x8B, I think...

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 11:11 pm UTC
by buffygirl
Jonas79 wrote:
buffygirl wrote:For Jonas79

no, this is not sacrilege! And remember, I'm in a hurry!


The hat is beautiful and I'll wear it with pride when I get to a computer.

Hmmm .... not sure if I like it anymore ... too .... blue. Might need to fix that.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 11:11 pm UTC
by descor
AK49BWL wrote:What, no GONG?
Spoiler:
Image

0x8B, I think...

Well, I'ma coma. Hope there's still some OTC in the morning. Night all y'all.

Re: 1190: "Time"

Posted: Sun Apr 28, 2013 11:15 pm UTC
by AluisioASG
Eliram wrote:
buffygirl wrote:uhhhh ..... Lost is still on my to-do list (yes, I missed it first time round! I'm sorry!) So thank you for the spoiler heads-up, cuz it meant that I didn't follow the link! Woot!

I think it won't be a spoiler to say that "Lost", just like the OTC (as we perceive it at this time), may not give full satisfaction with its ending, but the journey is well worth it. :-)

Well said. Don't watch it, and you'll miss lots of cultural references later.

Hmm, what's this I wrote on my hand?