# pouët.net

Go to bottom

## The pixel dies in five years.

category: offtopic [glöplog]
What kb_ said.
added on the 2012-12-17 00:25:51 by trc_wm
This paper state that color are vectors, so how does this help your argument that RGB pixels are not vectors since they hold color information?

To restate, R, G, B components are the magnitude of the corresponding axis in a 3D space.
All vector & matrix math is directly applicable taking pixel as vector.

example: Magnitude of the vector (RGB Pixel) represent its brightness.
Normalizing of the vector will do a projection on the luminescence plane.
You can visualize easily all those operations in 3D space if think of RGB as 3d vectors. Do you see your red, green, blue little arrows now ?
Take an RGB color (a vector), you can directly build a rotation matrix from it. Doing this with a 'gray' vector will result in a hue rotation matrix.
etc..

But I guess. nonono those are not vectors.

added on the 2012-12-17 01:28:17 by T21
added on the 2012-12-17 01:40:01 by trc_wm

added on the 2012-12-17 01:54:58 by kb_
'us' == You and kb_ , I haven't seen any other Poueter confused on this subject.

You both seem smart enough to figure this out on your own.
Plenty of publications on how RGB vector map to color processing.

You guys can really act like stupid jerks sometimes.
added on the 2012-12-17 01:59:40 by T21
T21: They are just trolling. Wake up!

Of course, vectors cannot kill pixels as every vectorfield is obviously an isomorphism of a colormap. As anyone can see.

That, and what Kb_ said and Gargaj will reply.
added on the 2012-12-17 02:05:54 by ham
Apart from that you really should scale the color space according to the relative brightnesses of R,G and B before measuring the magnitude or normalizing the vector. That's eg. why HSV isn't luminance preserving when you modify the hue, and neither are any of the usual YUV variants, but LAB is.

Also, the abovementioned paper should have told you that viewing colors as mere 3-vectors is a bit too simplified and will get you nowhere as soon as you do some nontrivial compositing. The homogenous 4-vector representation really helps.
added on the 2012-12-17 02:12:14 by kb_
lol
added on the 2012-12-17 05:11:53 by iq
kb_ yes and yes.

My statement was that pixels hold color vector, you kill pixels you kill vectors.
This being a joke on the university of Bath stating that pixel will be replaced by vectors within 5 years.

I wasn't doing a lecture on color space, not going to waste my time on trolls :)

added on the 2012-12-17 05:46:05 by T21
My statement was that pixels hold color vector, you kill pixels you kill vectors.

Yeah, just like cars hold people, so if you get rid of cars, there won't be humans on this planet anymore. I mean, really, just imagine a human being outside of a car, that's, like, ridiculous. Same with horse carriages back then; they had to completely rebuild humanity when the carriages vanished.

Also if you keep insisting on mixing the concepts of vector in general and specialized location/direction vectors in R3 I'm slowly wondering WHO is trolling here.
added on the 2012-12-17 10:28:11 by kb_
This thread is so pointless that it needs it's Axolotl.

added on the 2012-12-17 11:47:18 by Optimonk
Oh I like that one! Is it made out of pixels?
added on the 2012-12-17 16:41:29 by numtek
I hope Macromedia Flash will die before pixels.
added on the 2012-12-17 18:40:01 by rez
it won't, because it has vectors.
added on the 2012-12-17 19:09:12 by skrebbel
wait a second.... i do have a height, a weight and an age. i can even write me = (heigh, weight, age). i'm a vector!

(just as a note, almost anything you can think can form a vector space, so what. that doesn't mean a pixel IS a vector, a pixel is a pixel, and that's why "vector" and "pixel" are two different words)
added on the 2012-12-17 21:22:45 by iq
Quote:
wait a second.... i do have a height, a weight and an age. i can even write me = (heigh, weight, age). i'm a vector!

that's... i... wow.
added on the 2012-12-17 21:29:04 by Gargaj
Wait, wait!

Just remember that our color representation is based on the trichromaticism of the primate retina. There are other posible primary colors that could be mapped in R4 or R5 vector spaces.

I see too much anthropocentrism in this thread!
added on the 2012-12-17 21:37:27 by ham
the fug. i don't get it. what's that vectors doing better. the demos of it are lame. sure. on low detail density they can save space defining spaces with vectors and filling up with gradients. that zooming is neither impressive. can be done with a coarse edge detection algorithm and rendering a vectorised representation with polygons and the pixel colors. it won't add more detail tho if you don't add more data in a sub level or layer. so...

;)
added on the 2012-12-17 21:48:05 by yumeji
This is just dense.

But ok, I'll byte one last time on this fresh new bait.

(heigh, weight, age) doesn't fully represent a person,
but a 3 dimensional vector (or point) does fully express the RGB color space.
(If anyone dare to go pedantic on this, I swear, I will release the ponies!)
added on the 2012-12-17 22:31:48 by T21
T21: but if a pixel is merely a 3 component vector, and 3d meshes are made up of the same kind of 3 component vectors, 3d objects are made entirely of pixels no?

(I won't comment on iq, since we know he's made of pixels, not vectors.)
added on the 2012-12-17 22:36:59 by psonice
i'm sure that if put all the dimensions and reduce them with PCA, the three main eigenvectors will be height, weight and age :D
added on the 2012-12-17 22:51:51 by iq
Humm...

So its might be possible to get a person importance from their vector length.
But I guess we need to take into account what kb said, that not all axis are of equal weight, so some axis scaling is needed.

importance (how 'big' of a person you are) = length(height*.2, weight*.1, age*.7)

Would this be then correct ? you as a dwarf = normalized(yourself)

Could you also build a rotation matrix around a person of great importance and do a 180 on a point cloud of people?

Fascinating...
added on the 2012-12-17 23:25:35 by T21
I am a vector too! I am a linear combination:

(Oxygen, Carbon, Hydrogen, Nitrogen) x (0.65, 0.18, 0.1, 0.03)
added on the 2012-12-17 23:47:36 by fizzer
fizzer: You should take in count other elements but your simplification is OK and could be mapped in a pretty tetrachromatic color space suitable for the eyes of most birds and arachnids.
added on the 2012-12-18 00:13:07 by ham
if you spread diseases you're a vector
added on the 2012-12-18 02:22:42 by xernobyl