I've resized physics objects before with no crashing

it didn't work when I tried it just now though

I guess it's pretty touchy

anyway, for now though if you don't mind the collision mask thing

you can have a non physics object follow it around

here's an example with some random emission stuffs

and growing particles with physics

http://dl.getdropbox.com/u/1013446/particlephys.cap

you can store the current color filter for each sprite in a private variable and you could do the color ramp as well

btw, if it runs super slow on your system, cut down the fade out time on the sprites, it should make it run better

btw, in case anyone was wondering

it isn't a significant amount of time to check lots of UIDs

I just made construct check the unique ids of 100 sprites 500 times a tick

which is a total of 50,000 UID checks per tick

and it was still running at 400fps

thanks again deadeye

this made my day

man oh man, wait till you guys get a load of what I'm working on

thanks deadeye.

> sometimes you need to be able to choose an object out of a pool of many

> without being able to pick it first

>

I don't know what you mean by this. Is there any significant difference between "choose" and "pick" in this context?

I meant just pick it without it having to trigger an event

[quote:25orbjsp]Also, you don't need a For Each Object loop to pick by UID, you just compare the UID. For Sprite, there is a specific condition just for that under Misc. For other objects you can do a Pick By Comparison and use obj.UID as an expression.

cool, I just realized the compare UID thing for sprites, before I came back to the forums. as for the other objects and pick by comparison, I guess something someone said at one point made me feel this would be inefficient in large doses, like it's cycling through each object, and if you had a few hundred things to pick from and you were picking more than one UID every tick, it would eventually start slowing things down. is this true? or is it cycling through every object to check a variable a negligible amount of cpu time that won't really add up? because if that's the case, it will simplify alot of other things on this project.

what about the other issue, with the rotation?

I think it's been mentioned by others as well as myself,

but then I learned alot and thought it was no longer necessary

and then I learned alot more and I realize

it really would be ridiculously useful

and allow alot of flexibility with functions and complex object interactions

sometimes you need to be able to choose an object out of a pool of many

without being able to pick it first

or you want to be able to remember a list of certain objects for later picking

without being able to repick some object that you're going to pair them to

I don't mean anything complex, just an event

pick object by UID

then you put the UID

and it will pick whatever object and only that object will be picked(unless there are other conditions in the event)

I understand it's possible to do a for every object and then a UID is equal to temp

but first, this would require a different event for every object type or family

and if you want to use this feature alot, it will always be looping through every object over and over

if it's possible with python, someone please explain where I can start looking and whether this feature is broken, or more generally, exactly how broken python is

I can't remember what the reason this UID picking idea is usually rejected for, or if it's just usually a request not responded to. is it harder than it seems?

Davo can say for sure, but I think he said he'd fixed some of the problems with Python in the next build.

Just curious about the status of the python stuff in general

especially any changes planned for the .99 build

in particular:

making it so you don't have to have the extra python file with your exe

whatever it is that makes everyone say "python is broken right now"

and also the object picking issue

wondering what the priority and/or progress is on those

I'll post it, I didn't mean the crash might be a feature

I meant the way canvas has private variables that are different from everything else

I thought that was a feature

it would help if you posted a cap file

there are many ways to do this, and it might be different depending on how everything else works.

but for starters

there is a system expression for random

like if you have 5 weapons, you can have them have a variable weapon held that equals random(4)+1 (a random number between 1 and 5) when they run over the weapon spawner

and then have each weapon have a number associated with it

and in the code for firing the weapon, the last step could change weapon held back to zero(which you can set to mean no weapons held)

as far as random paths for bots, that totally depends on how the game moves. I'm guessing it's top down, but without a cap file, or a more detailed explanation of how the game works, there's not much advice to be given on that, other than to read the other threads where people asked for help with path finding

I once saw ashley respond to a post by saying you shouldn't have sprites too much larger than 512 by 512. so I would consider that a reasonable size

first, if you rotate a canvas, and paste an object to it, it pastes it as if it were in the unrotated position when you did it

example

is this a bug?

also, canvas doesn't have normal private variables

it doesn't force them to be initiated, so you can choose any name at any time, like a hashtable, but it doesn't seem to work anyway. it crashed when I tried it. just making sure this isn't some intentional feature I just don't know how to use before I bug track it

it will be much cleaner when it's not like a prototype hack on thing where we're trying out different features, and leaving other ones halfway on. I don't think the tons of pickupable objects will be a problem, you could do the same thing with families and it shouldn't need any extra events as far as the picking up itself is concerned. as far as the spawning multiple physics ball glitch, it should be easy to fix once we've decided on a final method of doing the sticking.

As far as moving slower, yes, you're probably right, but for it to work, the katamari needs to be able to climb walls and do the superspin for it to work slower without limiting it too much. also, eventually it should have the thing where it bumps into something too big and pieces fall off

also, figured out while it gets blurrier and blurrier,

it's because of constructs linear sampling, when it copies the ball and trash at an angle, it's copying extra little semi transparent pixels here and there, that construct puts in to make games look smooth normally, but when added together over and over, gives it that blurry look. it's either antialiasing, or it's like antialiasing. there are two ways to fix it I can think of off hand. one is by using that original stick method from the first example, but only when you pick up a new object, so it's not keeping track of all those objects every frame, when you run over something, it can put them all back in place take a picture, and delete them all again. if that's not fast enough, you can use the stick method to rotate the new object around to where it it's rightside up with the canvas. The second one is far better I would think. I didn't mention this before, but the reason i did the copy back and forth between two canvases is because if you paste to a canvas that's at an angle, it pastes it on the canvas as if it were right side up, so it's stuck to the wrong part of the ball.

anyway

gotta go to bed for now

this is some fun stuff though

updated with zoom out feature, and large area to drop down and get alot of stickies to test out zooming out

http://dl.getdropbox.com/u/1013446/katama.cap

EDIT: there's a glitch where if you get alot of stickies at the same time your ball starts acting weird, if you press space, you can see it's because more than one ball spawned and the invisible balls are hitting eachother. fun stuff