R0J0hound's Forum Posts

  • irina

    From the C2 manual:

    https://www.scirra.com/manual/126/system-expressions

    [quote:3dttumq0]CanvasToLayerX(layer)

    CanvasToLayerY(layer)

    Calculate the layout co-ordinates underneath a position in canvas co-ordinates for a given layer.

    LayerToCanvasX(layer)

    LayerToCanvasY(layer)

    Calculate the canvas co-ordinates above a position in layout co-ordinates for a given layer.

  • irina

    For the first capx the issue is the layer scales aren't considered currently.

    The second capx is the same. So moving the objects is a solution or you can maybe look into the system expressions to convert layer positions to screen and vise versa.

  • zixofranic

    There isn't really any documentation but the are descriptions in the editor when adding expressions, actions or conditions.

    For a coloring book you want a paint bucket fill like mspaint right? Look at the "floodfill canvas with color" action to do that.

    However the only limitation with that is Just make sure the canvas is completely opaque, aka has no transparency. Transparency can cause the flood fill to hang the game currently.

  • So it's like a triangle formation. If you disregard centering you can visualize the placement of the nth follower with:

     1  3  6 10 15
 2  5  9 14
 4  8 13
 7 12
11[/code:12i2572a]

I don't see any good patterns that we can make a formula with but I can see a pattern with the follower number and it's column:
[code:12i2572a]num:  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15
col: (1)(1  2)(1  2  3)(1  2  3  4)(1  2  3  4  5)[/code:12i2572a]
So we can do it with a loop:

number n=1
number column=1
for each follower
--- follower: set col to column
--- add 1 to column
------- if column >  n
---------- add 1 to n
---------- set column to 1

So with that you can set the followers x position with leader.x + 64*follower.col.

Then if you loop through each column of followers you can set the y position and even center them vertically.  See example for detail:
[url=https://dl.dropboxusercontent.com/u/5426011/examples22/squad.capx]https://dl.dropboxusercontent.com/u/542 ... squad.capx[/url] 

The simplest way to position relative to the leaders angle is to use "move at angle" instead of set x and y.  Again see example for detail.

  • Here's the equations for that.

    X=distance*cos(angle)

    Y=distance*sin(angle)

    But this is an angle and distance from the origin (0,0), usually you want the position (x,y) an angle and distance from another position (x0, y0).

    X=distance*cos(angle) +x0

    Y=distance*sin(angle) +y0

  • Snowmane

    You can set the clipboard with the nodewebkit object if you export or preview with nodewebkit. There isn't a way to set the clipboard otherwise. You could look through the platform specific plugins and see if any others have clipboard actions but I imagine not. The clipboard is mainly a win/linux/mac thing.

  • Here's an instantaneous way to plot the path:

    It doesn't take into account collisions though. For that your current method is the only solution atm.

  • One idea is to use a text list and use and remove one item at a time.

    global text frameList=",0,1,2,3,4,5,6,7,8,9,10,11,"

    global number randFrame=0

    Start of layout

    For each sprite

    --- set randFrame to tokenat(frameList, random(1, tokencount(frameList, ",")), ",")

    --- set frameList to replace(frameList, ","&randFrame&",", ",")

    --- sprite set animation frame to randFrame

    Another idea that equalizes eventually is:

    repeat 12 times

    Sprite animation frame is loopindex

    system compare sprite.pickedcount > 1

    --- sprite set animation frame to int(random(12))

    Or if you want to make it work in one tick you can do this:

    global number unique = 1

    while

    unique=1

    --- set unique to 0

    ------ repeat 12 times

    ------- Sprite animation frame is loopindex

    ------- system compare sprite.pickedcount > 1

    ---------- sprite set animation frame to int(random(12))

  • Maybe use a loop?

    global number radius=1000

    global number num =100

    Start of layout

    Repeat num times

    --- create object planetoids at (sun.x+radius*cos(loopindex*360/num), sun.y+radius*sin(loopindex*360/num))

  • The value you get from the bullet's angle of motion expression is in the range of -180 to 180, which still is a range of 360 degrees and still works with most all angle ACEs. If you want to convert it to the range of 0-360 do this:

    (ang + 360)%360[/code:3ofxclp9]

  • Ashley It only occurs when not using webgl. In the sprite plugin it has the function "preloadCanvas2D" that draws all the animation images to the canvas on load. You can see it in any project if you set "Enable WebGL" to off, "Clear Background" to no and make the bottom layer transparent. So when the load is quick you will never see them since it will be overdrawn.

  • You could try the canvas.rgbaAt(x,y) expession... or use the blueAt(), redAt() and greenAt() expressions to get the color components to build a color string like "rgb(0,255,255)".

  • The flood fill action seems to hang when there is transparency in the canvas being filled. Also the other color names work as I recall in Firefox. But in chrome they don't seem to be, but using color strings like "rgb(0,255,255)" do work.

  • One brute force way it could be done is have the computer simulate every possible game and build a tree graph of moves to different states of the board. Then once built each move could be rated by how close it is to a check mate. Then the cpu could just pick the move that would result in a win the quickest. A drawback of this is it's not really possible to do this as building the tree graph would be very time consuming and it would take much more memory than is available.

    You could instead look n moves ahead instead and choose moves that would lead to a win in n moves or less, but you'd run into the case where there is no win that close so you'd have to use FragFather's idea to have the cpu do better than just a random move. You wouldn't be able to look too far ahead though as the amount of board states to keep track of increases exponentially. Consider as a simplification that there are only 10 moves possible in any given board state. Then with 0 moves you'd have 1 board state to keep track of. With 1 moves 11 states, 2 moves 111 states... So at 9 moves ahead you'd have 1,111,111,111 states to keep track of or a least check for a win. That is over a billion moves and if will were able to store each board state in 64 bytes you'd be using over 64 gigs of memory. So the problem becomes finding some kind of shortcut to make the cpu make a descent decision within the limits of memory and cpu speed. Few would want to play against a computer that took event 10 seconds to make a move. There are shortcuts and simplifications that can be done such as only looking at the better moves per step instead of all of them. That would allow looking more moves away whist using less memory. Also less memory could be used if you only keep the board states as long as it's needed, and mainly keep a move list.

    One method that is used to pick a better move is called Minmax. Where the idea roughly is to minimize the player's score and maximize the cpu's score. For this we need to know if one board state is better than another. It's relatively easy to identify a checkmate as a win or lose, but for all the other states we need some kind of rating. One rating could be to sum up the values of the pieces of the board where say the black pieces are: pawn=1, ..king=5, and the white pieces are negative: pawn=-1... king=-5. So after a move if the score goes down then it's better for white and if it's positive it's better for black. For a minimum you'll want at least 2 moves so the opponent's response can be considered as well, and a move can be picked with the best end score.

    Another method is the Monte Carlo way. AKA a random set of moves is done and the end score is checked. Usually this is done many times and then only the best result is picked as a set of moves to go with. I imagine this will be haphazard at times but if a high number of iterations (times) is used the results will be better.

    Probably a mixture of the methods will be needed to make a good ai that doesn't take forever to decide on a move.

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads

  • Poking around the source it seems to use the Titanium SDK as a base. Which was slow with C2 games last I heard: