|
Ben_Hurr
|
 |
« Reply #180 on: March 22, 2013, 04:25:06 PM » |
|
< desire to make J-Snake style post >
Do it, I can then accuse you of selling Evan Oil.  But no really do it I love reading technical articles. |
|
|
|
|
Logged
|
|
|
|
|
epcc
|
|
« Reply #181 on: March 22, 2013, 06:34:31 PM » |
|
what are micro impulses? why are they so good?
|
|
|
|
|
Logged
|
|
|
|
|
Evan Balster
|
|
« Reply #182 on: March 22, 2013, 08:20:51 PM » |
|
Force vectors. Since they're relative values, we've established there's no benefit to representing them as fixedFloats rather than doubles.
|
|
|
|
|
Logged
|
|
|
|
|
J-Snake
|
|
« Reply #183 on: March 23, 2013, 02:20:09 AM » |
|
In order to compute a distance-vector you need to subtract positions. Now think about the demands in order to compute microscopic distance-vectors.
If you don't need resolution-spikes around full parts for cave story+ etc. Fine. But you don't need to repeat telling it over and over. I am the creator of the upcoming fixedPoint perfection called TrapThem.
|
|
|
|
|
Logged
|
Independent game developer with an elaborate focus on interesting gameplay, rewarding depth of play and technical quality. TrapThem |
|
|
|
Gimym TILBERT
|
|
« Reply #184 on: March 23, 2013, 09:22:38 AM » |
|
What's the effect of resolution spikes?
|
|
|
|
|
Logged
|
 ILLOGICAL, random guy on internet, do not trust |
|
|
|
Quarry
|
|
« Reply #185 on: March 23, 2013, 02:35:26 PM » |
|
So, I was working on a FixedFloat-like implementation in Java. However I noticed some flaw I had done; http://pastebin.com/XvGKyRsYoutput the following floatOne: -2.5000
floatTwo: 2.3499
floatOne + floatTwo: 0.1501
floatTwo + floatOne: 2.5000
because I didn't have the "right" of the number have the ability to be negative! |
|
|
|
|
Logged
|
|
|
|
|
rivon
|
|
« Reply #186 on: March 23, 2013, 04:03:39 PM » |
|
Have you even read the first post in this thread? FixedFloat uses int+float, not int+short.
|
|
|
|
|
Logged
|
|
|
|
|
Quarry
|
|
« Reply #187 on: March 23, 2013, 11:24:30 PM » |
|
FixedFloat-like I tried using integers for the float part for extra precision |
|
|
|
|
Logged
|
|
|
|
|
Crimsontide
|
|
« Reply #188 on: March 24, 2013, 02:24:02 AM » |
|
FixedFloat-like I tried using integers for the float part for extra precision So what you have is fixed point then? |
|
|
|
|
Logged
|
|
|
|
|
Quarry
|
|
« Reply #189 on: March 24, 2013, 02:31:28 AM » |
|
I suppose
|
|
|
|
|
Logged
|
|
|
|
|
nikki
|
|
« Reply #190 on: March 24, 2013, 02:38:10 AM » |
|
I would like to see an implementation to this number proposition too though I guess,
makes things alot clearer to me then talks of äquidistant and correlations and resolution spikes.
|
|
|
|
|
Logged
|
|
|
|
|
Quarry
|
|
« Reply #191 on: March 24, 2013, 09:24:47 AM » |
|
I went back on the drawing board and came up with a lot healthier solution, the code below; http://pastebin.com/e3wWfCzQand outputs Addition and Subtraction
floatOne: 2.5000
floatTwo: -1.8999
floatOne + floatTwo: 0.6001
floatTwo - floatOne: -2.5000
Multiplication
floatThree: 2.5000
floatFour: -5.7500
floatThree * floatFour: -14.3750
Division
floatFive: 16.7999
floatSix: -5.1999
floatFive / floatFive: -3.2308
It's again not exactly FixedFloat (I didn't want to call it IntFloat) but it has no precision errors (except for the ones of the floats in the constructor) EDIT: Disregard this post, I just unknowingly reinvented fixed point arithmetics |
|
|
|
« Last Edit: March 24, 2013, 09:39:05 AM by Quarry »
|
Logged
|
|
|
|
|
rivon
|
|
« Reply #192 on: March 24, 2013, 10:45:23 AM » |
|
floatTwo - floatOne is not -2.5...
|
|
|
|
|
Logged
|
|
|
|
|
Crimsontide
|
|
« Reply #193 on: March 24, 2013, 11:01:13 AM » |
|
EDIT: Disregard this post, I just unknowingly reinvented fixed point arithmetics
LOL, its ok, I've done that many times in other areas  Its how you learn. |
|
|
|
|
Logged
|
|
|
|
|
Quarry
|
|
« Reply #194 on: March 24, 2013, 11:04:24 AM » |
|
floatTwo - floatOne is not -2.5...
floatOne is 0.6 after the first equation. -1.9 - 0.6 = -2.5! |
|
|
|
|
Logged
|
|
|
|
|
Developer
