Data Realms Fan Forums
http://45.55.195.193/

Half a pixel; another look at offset decimals
http://45.55.195.193/viewtopic.php?f=1&t=10949		 Page 1 of 1
Author:  Azukki [ Mon Apr 14, 2008 9:53 am ]
Post subject:  Half a pixel; another look at offset decimals
"There is no such thing as half a pixel."
That seems reasonable, but I'll test against it anyways.

I'll test it with the balancing of a rocket. For different tests, I'll move the offset of the main thruster by increments of .2 pixels. I'll test each adjustment until any patterns are clear. Tests are in a controlled environment, the rockets being ordered from the tradestar and with no other thrusters, and not getting hit by MOs. After some fall, thrusters will be activated. I'll observe and note anything I find noteworthy, but really it only matter which way it veers.

Control Testing, no adjustment.
Veers right, considerably and consistently

Test 2, main thruster offset moved .2 pixels towards balance
Veers right, but sometimes initially veers a tiny bit to the left

Test 3, main thruster offset moved another .2 pixels
Complete inconsistency; sometimes a considerable veer, sometimes three wobbles before veering

Test 4, main thruster offset moved yet another .2 pixels
Veers left somewhat more than right, still inconsistent

Test 5, main thruster offset moved yet another .2 pixels
Veers left consistently
Author:  Metal Chao [ Mon Apr 14, 2008 4:12 pm ]
Post subject:  Re: Half a pixel; another look at offset decimals
There is no such thing as half a pixel
Author:  robolee [ Mon Apr 14, 2008 4:18 pm ]
Post subject:  Re: Half a pixel; another look at offset decimals
hahaha, omg, I can't believe you would even post that.

http://en.wikipedia.org/wiki/Pixel
Author:  Azukki [ Mon Apr 14, 2008 9:57 pm ]
Post subject:  Re: Half a pixel; another look at offset decimals
I meant physically in the offsets of CC.
Not in images.

Did you read my post... or even the entire title?
Author:  Electroclan [ Mon Apr 14, 2008 10:19 pm ]
Post subject:  Re: Half a pixel; another look at offset decimals
robolee wrote:
hahaha, omg, I can't believe you would even post that.

http://en.wikipedia.org/wiki/Pixel


Quote:
There is no such thing as half a pixel


http://en.wikipedia.org/wiki/Game_engine

Have you guys modded recently? Anything more in-line with offsets?
Author:  robolee [ Tue Apr 15, 2008 12:40 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
by "that" I meant:
Azukki wrote:
"There is no such thing as half a pixel."
That seems reasonable, but I'll test against it anyways.

which suggests that you think you can have fractions of a pixel.

I think your finds are slightly strange. Sorry for my unclear post.
Author:  Grif [ Tue Apr 15, 2008 12:55 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
He obviously does NOT think you can make images have a half a pixel. He's not RETARDED. Though your comprehension of what he's saying might lead one to think you are.

Placebo effect? Are you seriously that dumb?

PUT ROCKET ON GROUND. PRESS UP. SEE IF IT GOES UP.

With repeated testing, variables are eliminated. Tell me, do you even KNOW what the placebo effect IS?

I cannot begin to see how you feel it applies here. Cortex Command.exe is not going to expect anything to happen, and deliver those results. Data has not yet created a SENTIENT PROGRAM.
Author:  Lord Tim [ Tue Apr 15, 2008 12:58 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
Perhaps Tradestar Midas, in believing sales would go up over the controversy, placed random emitters on either side of the rockets.
Author:  BlueThen [ Tue Apr 15, 2008 1:02 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
...

uh... 0.2 isn't a half.
0.2 is a fifth...
Author:  robolee [ Tue Apr 15, 2008 1:05 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
Grif wrote:
blah blah blah


CC uses physics, with random particles shooting out the back of an engine at random speeds, there is in theory unlimited things that could happen, however these random particles are limited to a certain range, but there is still a lot of possibilities, MEANING IT COULD GO IN ONE DIRECTION without even changing the code at all.

Quote:
may be a placebo
yes, I feel slightly stupid now, I thought a placebo wasn't limited to just healing, but it is still the same thing, if you constantly tell yourself you are the best, you will think you are doing better than everyone else, by thinking something is going to happen you may trick your mind into thinking it IS happening. (in this case you see that it is veering, which the rocket does normally anyways IIRC, but you tested it only once and now you think that adding .2 to the code will always do that)

also note I said "may", it all depends on how many times he tested it to how accurate the results are.

EDIT:
upon multiple testing myself, I found that the rocket varied in the amount of veering, sometimes a lot, sometimes a little, mostly to the right, but soetimes to the left... without changing anything at all.
EDIT2:
also I think CC rounds up/down soby moving it .8 pixels, it is moving it over to the right, putting more force on the right hand side and less on the left, meaning that it would veer to the left slightly, so these results are not entirely unprdictable. whereas there is no such thing as an actual half-pixel that does not mean that the cc engine cannot calculate the physics by fractions of a pixel, but I never said that.
EDIT3: (lol) CC may not round up/down for calculations, but I think we all agree that it does on the output, so moving a SPRITE/attachable by 0.x will get rounded for visual output.
Author:  Lord Tim [ Tue Apr 15, 2008 1:55 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
CC allows fractions (floats or longs or something) for things like mass, but doesn't allow fractions (ints) for stuff like FrameCount. I'm assuming that since the compiler doesn't give you an error for using a decimal value in the offsets, it probably doesn't round them either. That'd be an extra unneccesary step.

Meaning that, it does count the decimal parts of offsets without visually showing it (since it can't (pixels)), but that the randomness of emissions means your experimentation method is flawed.
Author:  Grif [ Wed Apr 16, 2008 2:12 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
Robolee, you're retarded.

Placebo != psychosomatic.

Quote:
When referring to medicines, a placebo is a preparation which is pharmacologically inert but which may have a therapeutic effect based solely on the power of suggestion. It may be administered in any of the ways in which pharmaceutical products are administered.


A "placebo" is used in double-blind trials to counter the variables in treatment. It is a false drug, it DOES NOT affect the patient in any real way. You give out a placebo to make SURE that your drug actually does something. A "placebo effect" has no bearing on Cortex Command.

What you're confusing it with is a PSYCHOSOMATIC effect.

Quote:
Psychosomatic medicine is an interdisciplinary medical field studying psychosomatic illness, now more commonly referred to as psychophysiologic illness, disorders whose symptoms are caused by mental processes of the sufferer rather than immediate physiological causes.


A psychsomatic effect is when the patient convinces themselves they have something they don't. Basically, exactly what you're describing, EXCEPT FOR THE TINY BIT WHERE YOU'RE TOTALLY WRONG.

Also, Robolee, UNLESS YOU CORROBORATED WITH DATA ON MAKING THE GAME, stop assuming things about how the engine works. No physics engine has truly random values. Most likely, it uses an array of "random" values that it cycles between. True randomization cannot be produced without access to a non-artificial randomization agent. Radio static is one of the more common ones.

Also also, your second edit is self contradicting. "I never said that CC doesn't calculate by fractions of a pixel, but CC rounds fractions of a pixel."

BlueThen: Azukki is referring to the fact that some GibOffsets the game editors spit out feature .5 pixel offsets.
Author:  Winterous [ Wed Apr 16, 2008 5:21 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
O_O
That's a lot of yelling.

Personally I think you should take into consideration that the emitter might not be throwing the particles out all at the same angle, because if one goes to the side then it'd throw the whole thing off balance.

Nice idea though.
Author:  Thanatos [ Wed Apr 16, 2008 6:22 am ]
Post subject:  Re: Half a pixel; another look at offset decimals
<3 Grif.

And back when I was testing this, I beleive I noticed a slight difference with the way my reactors were landing, based on the decimal of the pixel.
So I'm behind the idea that decimals of pixels do affect things in cc.
Author:  robolee [ Wed Apr 16, 2008 5:33 pm ]
Post subject:  Re: Half a pixel; another look at offset decimals
how am I contradicting myself? I said there is no such thing as half a pixel... and there isn't, you can however calculate half a pixel, and things such as spread based off a half pixel calculation. I still stand by the fact that there is no such thing as half a pixel. EDIT: (or any fraction of a pixel)

also I did not "corroborate" ("to make more certain; confirm". never heard of the word before) with data on the engine and I did not collaborate with him on the engine either, I did not make assumptions on how it works, I suggested a way on how it might, but how the hell would I EVER know unless he release the source code. (however I am pretty sure data does use a method to create "random" variables)

I also accepted that I was wrong about the placebo thing, look I made made a mistake, and so did you (corroborate), and I said I was wrong, but you brought it back up anyway... why? I would never have pointed out the whole "corroborate" thing if you weren't jumping on me for a mistake I made.

also your "psychosomatic" does not fit in with the effect either as it is still limited to the medicinal field, and I check out what a placebo really was shortly after you told me that it was wrong, which I then replied to notifying that, "yes, I feel slightly stupid now".
Page 1 of 1 All times are UTC [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/