Post PNG files, I will attempt to compress them
>>46788402
>>46788402
>awesome.png (979 KB, 500x500)
Wow... do you fucking suck!
>>46788419
I will compress it losslessly and I guarantee it will be smaller or your money back.
>>46788457
Is this KIOAASOFC?
Kraken.IO As A Service Over Four Chan
>>46788429
>>46788481
No, Kraken is lossy. bpg.js is better than Kraken
>>46788437
>>46788552
I ran optipng -o7
how are you doing that?
I know for sure that this one is a PITA to do.
>>46788495
>>46788456
>>46788582
Here you go friend :^)
>>46788664
Seems GIMP did a pretty good job on that one.
>>46788675
kek
>>46788627
ok friend
>>46788574
>>46788695
I don't have GIMP installed, I use Photoshop for my image editing.
>>46788582
Here's your lossless image. I will use tripcode for purpose of this thread
Here you go OP
>>46788402
optimize mai waifu
>>46788604
>>46788457
get fucked I guess?
>>46788809
wow 4chan says it's 848 rounding up.. windows says its 847.something
don't get fucked I guess...
>>46788627
>>46788819
Can this website's code be any shittier
>>46788790
>>46788841
P. nice gains, try this one
>>46788765
>>46788457 looks better than >>46788809
>>46788865
Hovering this pic caused Firefox to crash on 35.0.1, good to see 36.0 fixed it
>>46788923
not OP but here you go
>>46788860
>>46788442
http://a.pomf.se/mosvow.png
>>46788873
Mine is 100% lossless, you can verify by converting to bmp and comparing md5
>>46788950
Bring it!
>>46788765
This was a tough one
>>46788964
>>46789018
just use optipng guys
>>46789051
>>46789064
optipng is large compared to properly encoded png files
God, I really need to replace my Daman screenshots.
>>46789166
Does he fight with those baby arms
>>46789166
>>46789082
>Duplicate image
Congratulations on being first person
>>46789193
So what exactly are you doing to there images, anyway, anon? Why isn't your method of optimising png used universally already?
Also, here's one for you to do.
>>46789177
Don't know. Can't say I follow the series.
From my encode.
>>46789081
What are you doing to properly encode pngs? Teach me the ways of the bandwidth miser, ye who are so wise in bytes.
OP, are using compressing by dictionary , like LZW, or LZ77, LZ78 ?
>>46789258
>>46789241
>Why isn't your method of optimising png used universally already?
It's very CPU intensive
OP, you are fucking awesome.
>>46789279
answer my question OP
>>46789241
try PNGGauntlet.. it's pretty "universal" for you..
>>46789166
>>46789193
you call this optimized? fucking plebs.
>>46789387
a literal god
>>46789336
>>46789279
>>46789345
Yes.
>>46789378
My images will beat png gauntlet
>>46789387
It has to actually share md5 with bmps
>>46789387
is you optimization lossless ?
>>46789387
significant quality degradation
for i in *.png; do optipng -o7 $i; done
You're all faggots.
>>46789413
u forgot to add "in"
as in "insignificant"
>>46789424
>optipng
whats that ?
>>46789424optipng -o7 *.png
>>46788573
PNG is botnet, btw
>>46789413
>significant quality degradation
way? I looked at them side by side and flipped them and they look almost identical.
HUGE savings for minimal to no loss of any detail.
>>46789361
>>46789406
>>46789413
It's from pngquant. As the name implies, it uses a quantizer.
>>46789424
Post an image from optipng and I'll make it smaller.
>>46789426
Nobody likes dithering unless it's used in a DOS game
>>46788675
Saaaweeeeet. Now take some colors outta that dress.
>>46789463
That is subjective, it's still lossy compression. You're comparing apples and oranges.
Try me faget
>>46789490
>>46789490
so hard
>>46788419
like holy shit and it only took ONE SECOND
>>46789455
>>46789426
>>46789463
Just how fucking blind are you people?
>>46789490
Believe I did gap to this as a child, my good man
>>46789522
>-fs8
Pngquant looks like garbage
>>46789522
significant quality degradation
>>46789555
So basically you are a hacker
>>46789455
953k to 219k. over 4x reduction with no visual loss.
>>46789554
those are just the worst settings... the one posted earlier was fucking awesome at 1/3rd the size
>>46789595
better than that
>>46788865
came here to post this. Please do this next
>>46789490
377k to 145k
Go ahead, try to beat me on quality & size. Look at them side-by side. Do a fucking A/B test. You will not tell the difference.
I wrote this software myself for a PhD thesis.
>>46789636
>>46789595
I'm genuinely curious, but are you people really this fucking blind?
The differences are obvious to me. Then again, I work with digital image processing a lot.
>>46789595
>>46789633
>banding
>"no visual loss"
Stop using shit LCDs.
>>46789662
few pixel diff for 4x reduction. good luck convincing me this is not a good tradeoff.
>>46789651
>>46789662
LOL
the differences are obvious to me if I first 500x with a magnifying tool! silly plebs!
>>46789652
>82 KB
are u using compression by dictionary i suppose ?
lol of course you are
>>46789662
do you like tryhard to find this shit or a program?
>>46789652
I can ABX them pretty easily. Look are her lips for fuck's sake, it's ridiculously obvious even at 1:1
>>46789676
>>46789677
>>46789692
No, I'm comparing them at 1:1 size. I'm just using a tool to zoom in on the parts that I noticed right away, so you can see what I'm talking about.
Like I said, I work with digital image processing. I need to be able to find these kinds of image defects quickly.
>>46789636
>>46789692
Everyone with a monitor made in the past 5 years can see the difference. And now that you've been called out, anyone can see the difference for the reason that it's apparent it's actually there.
>>46789662
>lossy png
wtf is this shit?
>>46789738
Maybe it's my shit laptop
>>46789741
When you simplify an image by fucking up the colors, the lossless compression will make it smaller.
>>46789659
>jpg
Well I have to say I can't abx the difference.
>>46789636
1793kb to 385kb. Look at them side by side. Do an AB test. You will not be able to tell them apart.
It's even smaller than this jpeg >>46789659 LMAO
>>46789741
He's compressing the image using a lossy technique, then storing this as a (lossless) PNG result.
To be more precise: He's reducing the effective bit depth per sample, eg. by converting it to an indexed palette. This reduces the overall entropy of the image.
>>46789743
This is quite likely. It's probably running at 6 bits and also completely uncalibrated.
>>46789769
>This is quite likely. It's probably running at 6 bits and also completely uncalibrated.
Surely having brightness on the lowest possible is also a good thing.
>>46789762
>It's even smaller than this jpeg
what kid of wizardry is this?
>>46789762
Except for all the loss of grain.
>>46789762
Yours looks worse than the jpg. The colors are all fucked.
pngquant rapes the colors, while jpg can modify the shapes and general image
>>46789776
Massive color reduction.
>>46789777
>>46789778
what are u guys on about? I'm watching these images on a brand new lenovo thinkpad and I don't see any of this shit. to me they look indistinguishable from each other. no, I'm not him.
>>46789805
If you can't see the difference, convert both to bmp and look at the md5. Anything can be infinitely small if you rape the image
>>46789762 this one looks quite shitty
>>46789636
>>46789659
are virtually identical, I can't ABX any difference.
You'll notice that this is because the PNG file already contains extreme JPEG artifacts, so there's essentially nothing to be lost by converting it to JPEG again.
>>46789790
>Massive
nigger please. color loss is minimal at best. I'm now certain you guys are tolling me. it's like with that bullshit dress. my lenovo laptop is color corrected and IPS. I can't barely tell some diff when I switch between them but I can't say any of them is better than the other.
>>46789813
I can actually see a difference in the thumbnails, but not the images themselves
COMPRESS THIS FAGGOT
>>46789835
You're the faggot
>>46789829
Nobody cares about your chinkpad.
>>46789846
What the fuck happened here
>>46789846
AHAHAHA
Thanks for compressing my hacker encoded png n00b!
>>46789834
>I can actually see a difference in the thumbnails, but not the images themselves
Well, that's because thumbnails are JPEGS generated by moot's shitty library.
>>46789834
That's easy to explain: The PNG version has embedded colorspace information (ones that violate the spec, too), which the JPEG version stripped.
4chan is known for completely ignoring all colorspace tags, which is why those “the image changes if you expand it” trick thumbnail images work (they use very low gamma values, which your browser respects but 4chan ignores).
>>46789811
>if your eyes can't see it, COMPUTERS WILL!
you are getting btfo png compresser-kun
>>46789846
The colorcode I posted was #01FF00
Yours is #00FF00
Fuck off with your shitty lossy compression jpegfag
>>46789878
It must have had SRGB color information outside of the actual bitmap, any image viewer could ignore it
>>46789878
>The colorcode I posted was #01FF00
No, it was pure 001 green
>>46789875
>you are getting btfo png compresser-kun
this! that other guy is blowing him the fuck out and he's now bullshitting about some "massive" differences that only he can see on his super-computer.
Nigga plz... my IPS display is fucking amazing (it's made in Japan by Japan Display Inc) and I can't tell the difference.
His viewer probably lacks color correction so he can tell the differences.
4chan won't even let me post this
http://i.srsfckn.biz/eB.png
get rekt
>>46789875
>>46789912
>COMPUTERS WILL
Tell me what you're using right now.
>>46789919
>>46789018
>>46789054
None of these are lossless. Look at the colors, it's lightening the image
>>46789935
>>46789935
it's just a 1x1 png. Smallest image you can post is 10x10
>>46789919
i tried a 1x1 69 byte pic too
[hurr durr ur file contain embedded cp :^)][
t-thanks 4skinchan
>>46789968
>>46789955
Use a real browser like Firefox
>>46789968
i stand corrected
>>46789955
It doesn't matter if the source image is lossless for the purposes of this demonstration
>>46789973
>>46789974
Do you mind explaining how you made this white dot loss 40b
>>46789979
>>46788402
Here you go.
>>46789979
>>46789994
Maybe the original wasn't compressed at all
>>46789018
2692kb to 192kb over 14x reduction!
Now look at them side-by-side and tell me if you think the 14x reduction is size is worth it.
>>46789974
>>46788865
>>46789651
You won't find it any smaller than this
>>46790012
this is fucking crazy! again, I'm looking at them on an IPS display and I cannot tell them apart.
I save the original as JPEG and it was bigger than this guy's PNG!
wtf
I think he's saving as jpg and then just renaming to PNG since web browsers don't care about extension.
>>46789992
lmao
>>46790012
I can visibly see a difference with the hearts.
>>46790016
>>46789018
>>46790012
The second image is 0.04% different compared to the first.
http://huddle.github.io/Resemble.js/
yeah
>>46790012
>>46790040
Even if you claim to not see a difference, it doesn't change the fact that the data contained in the files are different.
>>46790042
It's the eyes for me, huge difference there
>>46790040
>>46790012
Some changes to the blue over her head and at the lower right, but pretty good.
>>46790041
ITT OP is saving to JPG at 90 quality then back to PNG
>>46790088
This will make files larger, not smaller.
>>46790007
>>46790012
Alright now tell me you did this. This could save a ton of bandwidth for my webservers and plebs aren't going to notice a difference.
>>46790085
>one fucking byte
why even live?
>>46790126
the 'software' we wrote is closed source
>>46790127
mad he lost
>>46790127
I'm drinking your tears
>>46790064
>huge difference
huge? what the fuck are you smoking! I've been staring at it and the diff is minimal. just few pixels changed. all the gradients are correct throughout the whole image. i don't know how this witchery works but it's bretty gud! and like he said, these few pixels changed is more than worth 14 times reduction.
>>46790138
well can i buy it?
>>46790126
http://pngquant.org
If you want to save bandwidth for your webservers, use JPG.
>>46790159
>use JPG
How so, you have done lower size PNGs than JPGs
>>46790154
It's still no replacement for JPG.
>>46790173
>It's still no replacement for JPG.
huh? his pngs are smaller than fucking JPEGs in this thread and no one can tell which of the two PNGs are better if you didn't know what the original was.
>>46790154
>all the gradients are correct throughout the whole image.
No, this is factually incorrect. The lossy image has very visible banding in the eyes.
>>46788402
>>46790170
In general, PNG can be smaller than JPG if the colors are uniform. Desktop screenshots, vector images will be smaler as png and look better.
I can't make JPG photographs smaller unless I use BPG
>>46790173
its open source you compression software?
>>46790192
>and no one can tell which of the two PNGs are better if you didn't know what the original was.
But that's a lie. If you're this delusional, we can set up an ABX challenge.
Suggest me a decent ABX software and set me up with your contact details. I will run an ABX trial with 100 comparisons.
If I get a single mistake, I'll pay you $100 in bitcoins - if I get all 100 right, you'll pay me $100 in bitcoins.
Interested?
>>46790192
A JPG at the same size as a pngquant image will look better most of the time.
>>46790199
>>46790222
do this one
>>46790032
from .png to .jpg
>>46790222
you used vector quantization algorithms ?
Do any browsers but IE support JPEG XR yet?
>>46790236
t-thanks.
>>46790239
>>46790245
I never used pngquant,, everything I posted has been lossless.
>>46788402
Checkmate
>>46790259
Checkmate? You're up against a grandmaster here.
>>46790256
I asked for a .png to .jpg why are you replying to me
>>46790266
ayy lmao
>>46790268
you can do that in paint yourself..
just save as .jpg
lets see you do this faglord
>>46790308
>not using imagemagick for this
pleb
>>46790266
shit
>>46790329
not the same image
>>46790338
>>46790349
>>46790329
UT OH
THE GRANDMASTERS REIGN IS OVER
>>46790329
failed miserably,try agn plox
>>46790314
>>46790391
>Deleting you´re failed attempts
Absolutely disgusting.
>>46790391
BetterPNGSmallerGuy
B
T
F
O
>>46790447
Don't bully.
He came through and is still da bes
>>46790391
please do this
>>46790447
he lost the battle but won the war >>46790391
>>46790221
Any takers? There were more than a few people who insisted there was no perceptible quality loss.
I wouldn't mind a quick buck.
>>46790391
ayy
>>46790474
>>46790490
BOO
YOU SUCK
>>46790490
well fuck.
>>46790494
Not the same image
>>46790490
>>46790524
>>46790524
>>46790541
>>46790490
>>46790549
Compressed your lossy image.
>>46790541
same quality i swer :^)
>>46790549
>all the color gone
and >>46790572 still gets it better
ayy lmao
>>46790579
>>46790603
0 KB denied
How are you guys doing this?
Is there some sort of script/program that will compress my multiple folders of png/jpg images losslessly.
>>46790623
not in a thousand hours..
>>46790623find . -name '*.png' -exec optipng {}\;
Or replace optipng by any other program to compress PNG files. The PNG Smaller guy uses a custom script that's a collection of multiple techniques.
code or gtfo
>>46790623
If you want to compress jpg losslessly, use BPG.
>>46790642
>>46790702
Nice dithering. Look at the eyes. It looks like a 256 color terminal.
Whooptipng -o25
Here?
>>46790664
>>46790682
oh for fucks sake use jpeg for this
>>46790682
>>46790664
Why the fuck do interlaced PNGs even exist. They're shit
>>46790664
>>46790642
why are you dumb nigger saving shit like screenshots and photos as png. learn to use a fucking computer.
>>46790040
>obvious color banding
>can't see the difference
I think your IPS display is 6-bit my friend. I can easily see the difference on my TN panel.
>>46790810
>>46790837
Because we can enjoy the pixel perfect quality from our lossy-encoded,laosian cartoons,which we then save in a lossless format and then apply lossy compression again
Makes sense amirite :^)
>>46790727
>that banding
Get a better monitor if you can't see that.
>>46790879
fuck you! you stupid cunt i've installed every image editor that has ever been made. my experience in clicking the moose and pushing buttons on my button board made me very rich! You fucking piece of cheap electronic trash. I am the real tech master and you cant even computer your mom. I wrote more code than there is in the mainframe of NASA + NSA.
>>46790881
>>46790911
Calm down, you can't properly handle the image formats you use.
>>46790727
ridiculous number of artifacts on this
don't you see all that HF color noise around the edges?
it's also blurry as shit
>>46790911
>>46790838