Official Off-Topic Thread

[tocky]

i just spent ages puzzling out this cheap way to do it:



you can count the diagonals on the display and then work backwards to pinpoint the cell

[Indigo]

really confused what you're talking about, tocky.  Can you explain?

[tocky]

um, maybe.

in isometric your original x and y coordinates are skewed and don't follow the x and y dimensions of the display. any given x-gridline (on the grid data structure) takes up several different y values on the screen (and vice versa for y-lines and x values). but the diagonals (on the data structure) map to constant x and y positions (on the display). so you can measure the distance in diagonals between any given point and the origin (on the display), and that'll tell you which cell you're addressing.

for each diagonal in one direction, the sum of the x and y adresses (on the data structure) for all cells on that line is constant. in the other direction, the difference between x and y is constant. so if you have both of those values, you've found the cell - but these are the same values you get from counting the x- and y-distance from the origin on the isometric display.

in the example, the cell you're after is one diagonal east and three diagonals south of the origin. so you find the cell in the target area where x-y=1 and x+y = 3, and that's your guy.

[surt]

Some simple matrix transformation strikes me as the most elegant solution (axonometrically general to boot).

Using a 2x2 matrix derived from your world-to-screen x and y vectors:

xx	xy
yx	yy

Multiply the matrix by your world coords to get screen coords.
Multiply the inverse of the matrix by your screen coords to get world coords.

[Indigo]

that IS a good way to go about it.  I didn't think about that at all.  Although my current solution works - this seems much more versitile.  Gah.  I may rewrite my function now.  Thanks surt

[junkboy]

Some information (only in 日本語) about the creation of the new KOF sprites, along with galleries and stuff. 100% DOT action!
http://kofaniv.snkplaymore.co.jp/info/15th_anniv/2d_dot/index.php

So far my favorite bit about the new game (aside from the sexy banding on the characters) is the somewhat disturbing backgrounds, especially the one with the french pig women.

[ptoing]

100% DOT action!

Edited from cartoonshaded 3D models. Fair enough tho, the sprites are huge and lots of frames.

EDIT: Just looked at the example animation. I have to say I have mixed feelings. Because stuff is made from 3D models the consistency is extremely high, but there is not much flair. Even for really fast attacks they do not use any blurring or stretching, which is some cases looks a bit robotic.

Also it is very formulaic and they added buffershades for the main body but when dropshadows which are of course all from the 3D source overlap they did not apply the buffershades, which looks a bit odd too, but probbaly wont be all that noticable in motion.

[junkboy]

Yeah, the lack of buffer shades in the "self shadows" is a bit strange, but I'm sure the style will be copied by pixelers of all ages for the foreseeable future. It'd be nice to read a translation of the page, especially since the jump from 2d to 3d to 2d is a bit fuzzy picture wise.

[Helm]

BANDING HELL

[Panda]

Uhmm let's see...

It takes them 3 days to make the rough sprite.
Then they make a 3d model + animations, based on it so they can trace the frames later on.
And after that, it takes them a week to make a finished frame.

A whole character (with all the frames and stuff) takes them roughly 6 months and a half.