if the teams/leagues decide to make money, this would make them a ton of money. money controls the scoreboards (after some point.)
>Pretty sure potus has stated it will not be mandatory
I've seen him do it twice in pressers.
comparing TinEye's image matching algorithm with Twitter's image matching algorithm is likely a waste of time.
>learning to defeat multiple different types of image sniffer algorithms is probablynota waste of time…
except that the one that matters right now is Twitter.
do what you want to do.
I've just spent 40 years doing advanced imaging. what do I know?
>"Enhance"
If I had a nickel for every time someone told me I could be rich by coming up with the algorithm that could enhance images like in the movies…
this bitch famous
this needs to be viral as fuck.
these idiots need to be shouted into the fucking ground.
>all of the social media sites
TinEye isn't a social media site.
yes, yes, I know. However, if you're using TinEye as a test, then you're not testing against social media sites.
>if we manage to create techniques that consistently mitigate the TinEye sniffer, we will have them ready to deploy when the socials (FB/Twatter/Insta/Reddit/etc etc etc) up their sniffer game
true, although image recognition algorithms like this can be (aren't necessarily) very, very feature dependent.
as long as you know that nearly all of that effort may prove useless, go for it.
personally, I'd focus on efforts that were more time relevant. but that's me.
>Twitter algo sniffer isalreadyproper fucked
I agree with the latest camouflage techniques.
Q said 2-4 days for the black squares in the corners (which is a really specific attack).
That said, were I the one working against the efforts here, I think I'd be able to overcome them in a week or two.
With the developers they have working on it? I give them a few months. ;)
Additional notes: They're going to use a shortcut on the overlay measures and tolerate a larger deviation from a match on an area bases, meaning that you'll need to reduce the transparency of the overlay. But that's not a proper response - it will be a short-term band-aid until they can figure out how to properly see through the overlays.
>dont mean to sound so defensive
digital communications (i.e.: chat) loses all of the human cues and tends to drive people into dissent that wouldn't normally occur in a face-to-face or even audio conversation.
no offense taken. i've been communicating this way since the 70's.
>In this cat and mouse game, the advantage is ours.
This is true.
The problem that Twitter et al have has more to do with being able to reduce an image down to as small of a dataset as possible (like a few bytes) and then be able to hash against a database of hundreds of millions or billions of known images to see if there's a match.
Their infrastructure was not designed to fight what we're doing.
This is the current advantage.
If they ever deploy 'real' imaging technology to recognize this kind of stuff, they'd be able to get around what we're doing very reliably.
But it would cost them and they can't do it quickly.
YOU ARE A FAG.
>I was wondering if there was any atributes in .png images that delineate "layers"
There are not.
However, if you want to cost them more money, use PNG images. They don't compress as well as JPEG (because they are non-lossy) and they take more computing resources to encode/decode.
>the wisdom is apreciated
I think you're going to see a festival of wisdom and technology squirting out of everywhere after we're beyond the watershed moment of this reveal.
The future world is looking very bright indeed.
Right now, I would imagine that most creative types are sitting back and waiting, much like I am. I've got some shit up my sleeve that would make cockroaches wretch.
>I think Twitter converts everything to jpeg.
They may for the very reason I mentioned.
They suck.
We spent all that time developing PNG to get away from the LZW patents and Unisys assholes just so they could ignore it.
Ho hum.
>Nope, nevermind. I just found a PNG.
They might convert to JPEG when the image is primarily continuous tone rather than high contrast. JPEG will artifact the hell out of high contrast edges (and the compression suffers anyway.) PNG is actually better at compressing high contrast stuff with very few colors. So it might be selective - that's the point I'm trying to make.
Jesus it is comfy in here.
Almost like the bad guys are about to get thrown in a box.
Bye bye, bad guys.
why doesn't green text work?
It needed to be done.
you have to push that little yellow button.
I'm ashamed to have thought to search for this.
Of course, it is there.
Note: 7,941,555 views.
Also note: Thank GOD it is in HD.
This is one of the most fucked up things I have ever seen.
Thank you for that anon.
Or fuck you.
Not sure which.
Unfortunately, I cannot find one.
>10 days from now a massive proportion of the votes will be already cast
Imagine how many Democrats are going to be upset when the October surprise reveals [part of] the criminal behavior of the Democrats.
kek
>If everyone votes early and is accounted for ahead of time it will be easier to weed out fuckery later.
Yes.
Also, is it possible Trump could deploy the military to postal facilities to help with the volume? (They'd actually be monitoring fuckery.)
also a hue shift
with some creative coding, you could color balance the colors chosen to generate an average gray in the result.
>I added random block size and random rotation to make it harder for their sniffer to look for color-block transition at predictable locations.
good idea.
if you want to do the color/area balancing, look into 'stochastic dithering'. You're basically generating randomness but then compensating for errors in nearby regions. The overall effect will minimize the color distortion as far as the human perception is concerned.
>That would require much smaller blocks.
No, the dithering itself could be done on the larger blocks. The reasoning is just that if you have a blue block, you want it close to a green block and a red block (or a purple block). For any larger area than the blocks, you want the average overlay color to be as close to gray as possible. The dithering of the block colors would achieve this.
It isn't a major enhancement, but it may make the results less circusy.