/qa/ - Question & Answer » Thread #2694111

[63 / 39 / 9]

420KiB, 723x1023, 1554956694716.jpg

View Same Google ImgOps iqdb SauceNAO

Anonymous Sat 13 Apr 2019 02:32:24 No.2694111 View View Reply Original Report

Quoted By: >>2694112 >>2694266 >>2694286 >>2694494

What color is this image when you open it in your browser?

Anonymous

Anonymous Sat 13 Apr 2019 02:34:19 No.2694112 Report

Quoted By:

>>2694111

>See thumbnail
THANKS DOC

>Open it

Whoa, dude!

Anonymous

View Same Google ImgOps iqdb SauceNAO crop.jpg, 631B, 8x8

Anonymous Sat 13 Apr 2019 04:34:47 No.2694229 Report

Quoted By: >>2694494

testing the top left corner

Anonymous

Anonymous Sat 13 Apr 2019 04:53:34 No.2694241 Report

Quoted By: >>2694243

dude wtf

this is different than the normal changing pictures because 1) it's not a png and 2) it works on my phone

Anonymous

Anonymous Sat 13 Apr 2019 04:56:19 No.2694243 Report

Quoted By: >>2694266 >>2694268

>>2694241
What color do you get? It might be different for different people.

The mechanism, I think, is differences in how out-of-range colors are handled in different JPEG decoding implementations.

Anonymous

Anonymous Sat 13 Apr 2019 05:26:52 No.2694266 Report

Quoted By: >>2694268 >>2694272 >>2694446

>>2694243 >>2694111
Not him, but it's blue for me in desktop firefox.

Anonymous

Anonymous Sat 13 Apr 2019 05:28:27 No.2694268 Report

Quoted By:

>>2694243
>>2694266
Also it's green in feh, sxiv, and gimp

Anonymous

View Same Google ImgOps iqdb SauceNAO 1555122744793.png, 899KiB, 723x1023

Anonymous Sat 13 Apr 2019 05:32:44 No.2694272 Report

Quoted By: >>2694278 >>2694285

>>2694266
A purplish blue like this?

Anonymous

Anonymous Sat 13 Apr 2019 05:43:11 No.2694278 Report

Quoted By: >>2694285

>>2694272
I get this in both desktop Firefox and GIMP. Also in Chromium, eog, ImageMagick's display, and gdk-pixbuf-thumbnailer. I have yet to find an application on my machine that displays it as green. My OS is Debian testing.

Anonymous

Anonymous Sat 13 Apr 2019 05:53:57 No.2694285 Report

Quoted By:

>>2694278
Wow, that's very strange. I would have thought it'd at least be the same as long as we're using the same application. I just tried it in my imagemagick too and it's also green there.
My current distro is GuixSD.
I suppose all these applications use the same library, so maybe we have slightly different versions or something. Aside from FF (actually I'm using icecat but that shouldn't affect the image rendering libraries)

>>2694272
Yes, just like that.

Anonymous

Anonymous Sat 13 Apr 2019 05:55:20 No.2694286 Report

Quoted By: >>2694299

>>2694111
Out of curiosity, what is the actual color that it's supposed to be?

Anonymous

Anonymous Sat 13 Apr 2019 06:08:34 No.2694299 Report

Quoted By:

>>2694286
I don't think there is an actual color. The values are outside the valid range.

Anonymous

View Same Google ImgOps iqdb SauceNAO table.jpg, 12KiB, 512x512

Anonymous Sat 13 Apr 2019 06:53:39 No.2694321 Report

Quoted By: >>2694332 >>2694335 >>2694494

let's see what this ends up looking like

Anonymous

Anonymous Sat 13 Apr 2019 07:14:22 No.2694332 Report

Quoted By:

>>2694321
...what the fuck...

Anonymous

Anonymous Sat 13 Apr 2019 07:21:13 No.2694335 Report

Quoted By:

>>2694321
In FF I see only 2 stripes instead of 4. Green on top and blue/purple at the bottom.
But again, when I open it in anything other than FF though it looks more like the thumbnail 4chan generates.

Anonymous

View Same Google ImgOps iqdb SauceNAO q16.jpg, 41KiB, 1024x1024

Anonymous Sat 13 Apr 2019 07:50:29 No.2694344 Report

Quoted By: >>2694355 >>2694370 >>2694494

Anonymous

View Same Google ImgOps iqdb SauceNAO q64.jpg, 41KiB, 1024x1024

Anonymous Sat 13 Apr 2019 07:51:31 No.2694346 Report

Quoted By: >>2694494

Anonymous

View Same Google ImgOps iqdb SauceNAO q256.jpg, 41KiB, 1024x1024

Anonymous Sat 13 Apr 2019 07:52:31 No.2694349 Report

Quoted By: >>2694353 >>2694355 >>2694370 >>2694494 >>2694511

Anonymous

View Same Google ImgOps iqdb SauceNAO ac.jpg, 119KiB, 1024x1024

Anonymous Sat 13 Apr 2019 07:57:15 No.2694352 Report

Quoted By: >>2694494

Anonymous

View Same Google ImgOps iqdb SauceNAO image.png, 2MiB, 1500x1488

Anonymous Sat 13 Apr 2019 07:59:01 No.2694353 Report

Quoted By: >>2694355

would it be possible to have an actual image in the thumbnail? Like the thumbnail looks like this but when you open it it's gray like >>2694349 or something

Anonymous

Anonymous Sat 13 Apr 2019 08:02:07 No.2694355 Report

Quoted By: >>2694370

>>2694353
>would it be possible to have an actual image in the thumbnail?
It should be possible.
>when you open it it's gray
So >>2694349 looks uniformly gray to you when opened?
What does >>2694344 look like?

Anonymous

View Same Google ImgOps iqdb SauceNAO image.jpg, 104KiB, 700x700

Anonymous Sat 13 Apr 2019 08:16:41 No.2694370 Report

Quoted By: >>2694373

>>2694355
>So >>2694349 # looks uniformly gray to you when opened?
Yes (iphone)
>What does >>2694344 # look like?
The same as the thumbnail except there are more squares like this

Anonymous

Anonymous Sat 13 Apr 2019 08:27:25 No.2694373 Report

Quoted By:

>>2694370
Interesting. On desktop Firefox I see fewer squares than are in the thumbnail. I see 4x4 whereas the thumbnail has 8x8 (counting the white, blue, yellow, and black parts as 2x2 squares). And it looks like you have 16x16.

There seems to be an integer overflow going on when decoding these images, but different software is hitting the overflow at different points.

Anonymous

View Same Google ImgOps iqdb SauceNAO miku2.jpg, 144KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:28:28 No.2694374 Report

Quoted By: >>2694381 >>2694390 >>2694393 >>2694394 >>2694397 >>2694459 >>2694494

test

Anonymous

View Same Google ImgOps iqdb SauceNAO Screenshot_2019-04-13 (0) qa - W (...).png, 162KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:37:12 No.2694381 Report

Quoted By: >>2694390 >>2694393

>>2694374
how it looks in Firefox on Debian testing

Anonymous

Anonymous Sat 13 Apr 2019 08:37:38 No.2694383 Report

Quoted By:

You guys, I think my puter is broke.

Anonymous

View Same Google ImgOps iqdb SauceNAO miku2.png, 72KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:44:22 No.2694390 Report

Quoted By: >>2694393 >>2694494

>>2694374
how it looks in ImageMagick display -- same person as >>2694381

Anonymous

View Same Google ImgOps iqdb SauceNAO miku2-gimp.png, 10KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:45:24 No.2694393 Report

Quoted By:

>>2694374
how it looks in GIMP -- same person as >>2694381 >>2694390

Anonymous

View Same Google ImgOps iqdb SauceNAO miku.jpg, 144KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:46:25 No.2694394 Report

Quoted By: >>2694396

>>2694374
The image before adding an offset of 512 (multiplied by the quantization table coefficient) to all the DC values.
Looking at this again, I see that the luminance DC quantization coefficient is 16 as intended but the coefficient for chroma is 17, which explains why the colors are off in the thumbnail. Going to try again with all coefficients set to 16.

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512.jpg, 277KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:51:27 No.2694396 Report

Quoted By: >>2694398 >>2694451 >>2694459 >>2694464 >>2694494

>>2694394
>Going to try again with all coefficients set to 16.

Anonymous

View Same Google ImgOps iqdb SauceNAO 0008.jpg, 43KiB, 319x310

Anonymous Sat 13 Apr 2019 08:51:37 No.2694397 Report

Quoted By:

>>2694374

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512ac1.jpg, 579KiB, 1500x1488

Anonymous Sat 13 Apr 2019 08:56:20 No.2694398 Report

Quoted By: >>2694399 >>2694432 >>2694436 >>2694459 >>2694494

>>2694396
And now let's try one of the AC coefficients.

Anonymous

Anonymous Sat 13 Apr 2019 08:57:45 No.2694399 Report

Quoted By:

>>2694398
this is nice ac coefficients

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512ac4.jpg, 649KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:09:33 No.2694410 Report

Quoted By: >>2694413 >>2694419 >>2694420 >>2694432 >>2694436 >>2694459 >>2694464 >>2694494

different AC coefficient

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o488ac4.jpg, 634KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:13:57 No.2694413 Report

Quoted By: >>2694459 >>2694466 >>2694494

same coefficient as >>2694410 but an offset of 488

Anonymous

Anonymous Sat 13 Apr 2019 09:19:59 No.2694417 Report

Quoted By:

Cursed images!!
Don't click, virus!!

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q32-o512ac4.jpg, 565KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:25:18 No.2694419 Report

Quoted By: >>2694427 >>2694459 >>2694467 >>2694494

like >>2694410 but using an image with quantization tables full of 32s rather than 16s

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q64-o512ac4.jpg, 508KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:26:21 No.2694420 Report

Quoted By: >>2694459 >>2694467 >>2694494

like >>2694410 but using an image with quantization tables full of 64s rather than 16s

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q32-o256ac4.jpg, 549KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:34:48 No.2694427 Report

Quoted By: >>2694459 >>2694464 >>2694494

like >>2694419 but with an offset of 256

convert miku.png miku.ppm
for i in {1..64}; do echo 32; done > qtable32.txt
cjpeg -qtables qtable32.txt -qslots 0 -sample 1x1 miku.ppm > miku-q32.jpg
dcttransform 'dct.map((x,i) => (i%64)==4 ? x+256 : x)' miku-q32.jpg > miku-q32-o256ac4.jpg

https://gitgud.io/dcttransform/dcttransform

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512ac2.jpg, 621KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:43:37 No.2694432 Report

Quoted By: >>2694434 >>2694436 >>2694459 >>2694468 >>2694494

like >>2694398 >>2694410 but the second coefficient

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512ac3.jpg, 625KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:46:10 No.2694434 Report

Quoted By: >>2694459 >>2694468 >>2694494

>>2694432
third

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512ac5.jpg, 662KiB, 1500x1488

Anonymous Sat 13 Apr 2019 09:47:27 No.2694436 Report

Quoted By: >>2694437 >>2694459 >>2694468 >>2694494

>>2694432
fifth

first was >>2694398
fourth was >>2694410

Anonymous

View Same Google ImgOps iqdb SauceNAO Dctjpeg.png, 15KiB, 584x584

Anonymous Sat 13 Apr 2019 09:49:22 No.2694437 Report

Quoted By: >>2694440 >>2694442

>>2694436

Anonymous

Anonymous Sat 13 Apr 2019 09:51:34 No.2694440 Report

Quoted By:

>>2694437
to be clear about how I've been numbering them
they're stored in a zig-zag order but that's not how I'm numbering

Anonymous

View Same Google ImgOps iqdb SauceNAO Screenshot_2019-04-13 JPEG - Wik (...).png, 6KiB, 517x58

Anonymous Sat 13 Apr 2019 10:06:08 No.2694442 Report

Quoted By: >>2694445

>>2694437
the numbers for coefficient 4 work out nice and round

0 [0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003]
1 [0.17337998066526844, 0.14698445030241986, 0.0982118697983878, 0.03448742241036789, -0.03448742241036787, -0.09821186979838774, -0.14698445030241986]
2 [0.16332037060954707, 0.06764951251827464, -0.06764951251827463, -0.16332037060954707, -0.1633203706095471, -0.06764951251827472, 0.06764951251827467]
3 [0.14698445030241986, -0.03448742241036787, -0.17337998066526844, -0.09821186979838777, 0.09821186979838771, 0.17337998066526844, 0.034487422410367966]
4 [0.12500000000000003, -0.125, -0.12500000000000003, 0.12499999999999997, 0.12500000000000003, -0.12499999999999986, -0.12499999999999994]
5 [0.0982118697983878, -0.17337998066526844, 0.03448742241036788, 0.14698445030241988, -0.14698445030241983, -0.034487422410367834, 0.17337998066526847]
6 [0.06764951251827464, -0.1633203706095471, 0.16332037060954704, -0.06764951251827465, -0.06764951251827477, 0.16332037060954707, -0.16332037060954702]
7 [0.03448742241036789, -0.09821186979838777, 0.14698445030241988, -0.17337998066526847, 0.17337998066526844, -0.1469844503024198, 0.09821186979838765]

Anonymous

Anonymous Sat 13 Apr 2019 10:07:16 No.2694445 Report

Quoted By:

>>2694442
oops, missed the last number

0 [0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003, 0.12500000000000003]
1 [0.17337998066526844, 0.14698445030241986, 0.0982118697983878, 0.03448742241036789, -0.03448742241036787, -0.09821186979838774, -0.14698445030241986, -0.17337998066526844]
2 [0.16332037060954707, 0.06764951251827464, -0.06764951251827463, -0.16332037060954707, -0.1633203706095471, -0.06764951251827472, 0.06764951251827467, 0.16332037060954704]
3 [0.14698445030241986, -0.03448742241036787, -0.17337998066526844, -0.09821186979838777, 0.09821186979838771, 0.17337998066526844, 0.034487422410367966, -0.14698445030241983]
4 [0.12500000000000003, -0.125, -0.12500000000000003, 0.12499999999999997, 0.12500000000000003, -0.12499999999999986, -0.12499999999999994, 0.12499999999999985]
5 [0.0982118697983878, -0.17337998066526844, 0.03448742241036788, 0.14698445030241988, -0.14698445030241983, -0.034487422410367834, 0.17337998066526847, -0.09821186979838765]
6 [0.06764951251827464, -0.1633203706095471, 0.16332037060954704, -0.06764951251827465, -0.06764951251827477, 0.16332037060954707, -0.16332037060954702, 0.06764951251827458]
7 [0.03448742241036789, -0.09821186979838777, 0.14698445030241988, -0.17337998066526847, 0.17337998066526844, -0.1469844503024198, 0.09821186979838765, -0.03448742241036793]

Anonymous

Anonymous Sat 13 Apr 2019 10:07:23 No.2694446 Report

Quoted By:

>>2694266
yellow on Android chrome

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o256.jpg, 277KiB, 1500x1488

Anonymous Sat 13 Apr 2019 10:15:10 No.2694451 Report

Quoted By: >>2694459 >>2694469 >>2694494

>>2694396
just as a check, this is with an offset of 256

Anonymous

Anonymous Sat 13 Apr 2019 10:26:24 No.2694459 Report

Quoted By:

So we can at least see that the thumbnail is normal when the (Y, Cb, Cr) values are off by a multiple of 1024.

>>2694374 - 512*(16 or 17)*(1/8) = 1024 or 1088 - a bit off
>>2694396 - 512*16*(1/8) = 1024 - normal
>>2694398 - 512*16*(irrational number) - very far off
>>2694410 - 512*16*(±1/8) = ±1024 - normal
>>2694413 - 488*16*(±1/8) = ±976 - a bit off
>>2694419 - 512*32*(±1/8) = ±2048 - normal
>>2694420 - 512*64*(±1/8) = ±4096 - normal
>>2694427 - 256*32*(±1/8) = ±1024 - normal
>>2694432 - 512*16*(irrational number) - very far off
>>2694434 - 512*16*(irrational number) - very far off
>>2694436 - 512*16*(irrational number) - very far off
>>2694451 - 256*16*(1/8) = 512 - very far off

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-q16-o512.png, 142KiB, 1500x1488

Anonymous Sat 13 Apr 2019 10:39:49 No.2694464 Report

Quoted By:

>>2694396 in Firefox on Debian testing
>>2694410 and >>2694427 are similar but with more solid color

Anonymous

View Same Google ImgOps iqdb SauceNAO 1555146837787.png, 96KiB, 1500x1488

Anonymous Sat 13 Apr 2019 10:40:51 No.2694466 Report

Quoted By:

>>2694413 in Firefox on Debian testing

Anonymous

View Same Google ImgOps iqdb SauceNAO 1555147518264.png, 162KiB, 1500x1488

Anonymous Sat 13 Apr 2019 10:41:53 No.2694467 Report

Quoted By:

>>2694419 in Firefox on Debian testing
>>2694420 is similar but with more banding in the gradients

Anonymous

View Same Google ImgOps iqdb SauceNAO 1555148617704.png, 89KiB, 1500x1488

Anonymous Sat 13 Apr 2019 10:42:55 No.2694468 Report

Quoted By:

>>2694432 in Firefox on Debian testing
>>2694434 is similar
>>2694436 is similar but with the edges of the original image more visible

Anonymous

View Same Google ImgOps iqdb SauceNAO 1555150510391.png, 12KiB, 1500x1488

Anonymous Sat 13 Apr 2019 10:43:58 No.2694469 Report

Quoted By:

>>2694451 in Firefox on Debian testing

Anonymous

Anonymous Sat 13 Apr 2019 11:53:33 No.2694494 Report

Quoted By:

testing on Android Firefox and Android Chrome, the results were the same between the two browsers
running Android 7.0 on a Moto X Pure

these look the same as they did in Firefox on Debian testing

>>2694111
purplish blue
>>2694229
light purplish pink
>>2694321
two bands, yellow over blue
>>2694344
4x4 (if thumbnail is counted as 8x8)
>>2694346
16x16
>>2694349
manyxmany
>>2694352
4x4 blue and yellow checker pattern with thin vertical stripes

>>2694374
looks more like >>2694390 except blockier
>>2694396
more solid colors, looks like >>2694410 does in desktop Firefox
>>2694398
like Firefox but with much more pronounced edges, like someone put it through a sharpening filter
>>2694410
very different from Firefox, more like how >>2694398 looked but without the pronounced edges
>>2694413
also very different, solid pattern of thin green and purple stripes with a few stray light green pixels
like the previous two without the edges of the original image
>>2694419
image looks completely normal
>>2694420
almost normal but with some banding
>>2694427
similar to >>2694410
>>2694432
>>2694434
>>2694436
also similar to >>2694410 and their appearance in desktop Firefox
edges more pronounced than in >>2694410 but less than in >>2694398

>>2694451
solid magenta, same as desktop Firefox

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-1bit-q16-o256.jpg, 3MiB, 1500x1488

Anonymous Sat 13 Apr 2019 12:14:42 No.2694501 Report

Quoted By: >>2694504

convert miku.png -ordered-dither o8x8,2 miku-1bit.ppm
for i in {1..64}; do echo 16; done > qtable16.txt
cjpeg -qtables qtable16.txt -qslots 0 -sample 1x1 miku-1bit.ppm > miku-1bit-q16.jpg
dcttransform 'dct.map((x,i) => (i%64)==0 ? x+256 : x)' miku-1bit-q16.jpg > miku-1bit-q16-o256.jpg

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-1bit-q16-inv-o256.jpg, 3MiB, 1500x1488

Anonymous Sat 13 Apr 2019 12:16:20 No.2694504 Report

Quoted By: >>2694621

>>2694501
dcttransform 'dct.map((x,i) => (i%64)==0 ? -x+256 : -x)' miku-1bit-q16.jpg > miku-1bit-q16-inv-o256.jpg

Anonymous

View Same Google ImgOps iqdb SauceNAO q256-gimp.png, 6KiB, 1024x1024

Anonymous Sat 13 Apr 2019 12:22:11 No.2694511 Report

Quoted By:

>>2694349 in GIMP on Debian testing
It doesn't seem to have the wraparound effect at all.

Anonymous

View Same Google ImgOps iqdb SauceNAO miku-1bit-q16-inv-o255.jpg, 3MiB, 1500x1488

Anonymous Sat 13 Apr 2019 13:17:58 No.2694621 Report

Quoted By: >>2694785

>>2694504
dcttransform 'dct.map((x,i) => (i%64)==0 ? -x+255 : -x)' miku-1bit-q16.jpg > miku-1bit-q16-inv-o255.jpg

Anonymous

View Same Google ImgOps iqdb SauceNAO image.jpg, 203KiB, 750x755

Anonymous Sat 13 Apr 2019 16:42:09 No.2694785 Report

Quoted By:

>>2694621
This one was weird. When it was loading it was a solid pink then when it fully loaded it turned into this.

Anonymous

Anonymous Sat 13 Apr 2019 18:31:33 No.2694785,1 Report Delete

Quoted By:

i can only imagine this was deleted for petty reasons stemming from a mod being scared of actually having to look at posts they delete

Anonymous

Anonymous Sun 14 Apr 2019 00:40:37 No.2694785,2 Report Delete

Quoted By:

we should make use of this then

Anonymous

Anonymous Sun 14 Apr 2019 18:53:56 No.2694785,3 Report Delete

Quoted By:

But how is it done?

Capcode	All Only User Posts Only Verified Posts Only Moderator Posts Only Manager Posts Only Admin Posts Only Developer Posts Only Founder Posts
Show Posts	All Only With Images Only Without Images Only Spoiler Images Only Non-Spoiler Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

On these archives

On these boards

Your latest searches