Is this possible to code?
category: general [glöplog]
I love that he actually put that in his FAQ.
random + local maxima = artistic value!
If it produced optimal result, everyone would get same images out of it. Now everyone gets slightly different one, which is great!
If it produced optimal result, everyone would get same images out of it. Now everyone gets slightly different one, which is great!
LoneStarr: You and your zero-tree coded wavelets... ;)
Hehe - I know.. but really - I think this idea is mostly interesting for generating an artistic look, not good compression ratio. That said, I think it's great, and it is probably pretty fun to implement aswell.
Not to sound like Mensa Member or anything, but it'd have to be a bit more interesting than random polygons and pixel-by-pixel RGB distance to be "fun to implement". Wouldn't it?
it is an evolving algorithm, which at least I would find fun to implement since you can see the result getting better and better(if you got the criterias right).
Quote:
LOL. Look the word "acclaimed" up in a dictionary to realize what you've just said...adok> could you please remind me your acclaimed IQ score?
sounds very similar to this :
http://www.romancortes.com/blog/bush-css/
http://www.romancortes.com/blog/arcimbolder/
http://www.romancortes.com/blog/bush-css/
http://www.romancortes.com/blog/arcimbolder/
adok: it's called irony.
Huh? I looked up "acclaimed" and there's was no mention of irony.
Tigrou: and even more to this:
http://www.romancortes.com/blog/jpeg2css/
The idea for jpeg2css was mostly from my origins of the pig 2 128b.
The problem with color images is that it would be very slow... In CSS you can only do transparency and not adition, so calculating the best positions with a greedy algorithm would be slow.
By other hand, I believe doing it with an evolutive algorithm or whatever algorithm is slow too, at least if you do it at a decent resolution...
Maybe doing it with the gpu for the rendering... but it is a problem to calc the image distance... (you know, to sum all values requires to resize down the texture several times... it's not cool).
Maybe it can be done in a lower resolution, let say, 128x128 or 64x64 and when the result is good enough, change to a higher resolution and adjust the polygons or circles or whatever one by one...
In any case, it doesn't look like if there is a good method to do it very fast... I think that a program is not very useful if it takes more than few minutes to execute
http://www.romancortes.com/blog/jpeg2css/
The idea for jpeg2css was mostly from my origins of the pig 2 128b.
The problem with color images is that it would be very slow... In CSS you can only do transparency and not adition, so calculating the best positions with a greedy algorithm would be slow.
By other hand, I believe doing it with an evolutive algorithm or whatever algorithm is slow too, at least if you do it at a decent resolution...
Maybe doing it with the gpu for the rendering... but it is a problem to calc the image distance... (you know, to sum all values requires to resize down the texture several times... it's not cool).
Maybe it can be done in a lower resolution, let say, 128x128 or 64x64 and when the result is good enough, change to a higher resolution and adjust the polygons or circles or whatever one by one...
In any case, it doesn't look like if there is a good method to do it very fast... I think that a program is not very useful if it takes more than few minutes to execute
texel: Interesting program. I guess its algorithm finds out the positions of the largest circle first by pattern matching, or does it do it by random?
Quote:
LOL. Look the word "acclaimed" up in a dictionary to realize what you've just said...
LOL, you actually needed to use dictionary to understand what he said.
And yes, i realize that you said HE needs to use dictionary, but i can read between the lines.
rainmaker: And did you understand what he said?
Adok, fuck off lamer.
Adok: We all understood what he said, and we all laughed at your behalf. Now get off Pouët, you're not welcome here.
jmagic:marvellous