Whats your opinion on this, Jow Forums?

A few weeks ago there was a thread where someone asked if it was possible to hide a picture in two other pictures. And there is: pic related. What do you think of it?
Also I'm not a faggot

Attached: img_eggsif_2671695469941832708.jpg (373x339, 18K)

Yet another image steganography thread

Explain to me how you intend to go about doing this without drawing a blatant outline of the actual image you're trying to hide.

the first image would still have the whole red channel of the secret image and ~94% of the green channel which is enough to tell what's in the secret image

GABR > RGBA

first you show me the outline

its way easier to just hide stuff in the metadata and hope the mods dont question your picture of bread that you posted in the loli thread being 4GB big

what if the pattern is reversed for every set of six pixels?

but these informations are hidden in ~6% of the red channel + ~6% of the green channel + ~6% of the blue channel so it wont be visible.
it could easily be done better by permutating the channels in a predefined way for example using the pixel coordinate to create the permutation (also this )

thats not how images work you dingus

>What do you think of it?
Where does each channel's high nybble go

so instead of speculating on this like you fags I decided to actually do it

Attached: out1.png (200x200, 51K)

it's pretty trivial to write a script to do this

better results would probably be achieved with a more chaotic set of images to hide the third image in

Attached: out2.png (200x200, 86K)

the big issue is that one of these (in this case out1) essentially contains the entire image, so if you have that one and simply reverse the nibbles, you'll have the original image.

Wouldn't it be better to use a YUV color space, that way the composite's colours just look like artwork?

what about hiding pictures inside huge pictures in little corners and shit, like imagine a nice huge picture of a generic shitty usa bridge and the noisy city behind, and you zoom in zoom in and tadah on the window of one of the dirty buildings is your favourite cuckold porn picture, or even better no picture, just inside the window your wife with jamal

Attached: large.jpg (500x479, 36K)

Someone's going to notice when your gigantic 100k by 100k jpgs contains an entire image in one of the corners.

Why not GRAB

why not use matrices instead?

Is RR supposed to represent upper and lower nibble of a 8-bit value? Why don't you be a little less vague next time.

yes of course it is. dont you know basic html color codes?

youtu.be/TWEXCYQKyDc

Some people aren't frontend toddlers

It does. The first image contains the most significant 4 bits of the green channel, which contains most of the channel's information. It obviously depends on the image, but blue generally has a lot lower information density and you're combining it with the residue from green, which is important to get the right colors but not the shape of the image.
Unless you're doing further permutations, this is worthless to hide something like an info graphic.

Attached: needful_split.png (400x1269, 784K)

For the latter however it might be useful to split up the brightness value of a grayscale image and distribute the least-significant nibble first.

Attached: doc_.png (1442x630, 59K)