GetPixelColor -> RGB

[Sephta]

My problem ist to check, if a symbol shows up at a specific place in the monitored window. The litte problem is: it has a changing background and it is 20% transparent, so the image recognition fails to check as well as the check for the exact color of certain pixels.

My idea was to check every pixel and calculate a 'range' of color, that matches, but this is difficult, as the getpixelcolor function returns a high range of numbers. To guess my color with the range i'd like to get to this format:

R: X1 +/- x
G: X2 +/- x
B: X3 +/- x

if X1-X3 are integer or hex, i don't mind...

I think the function FindImagePos with the color range does quite a familiar thing, but i can't use this function due to the great chances in the background. Any suggestions?

[Rain]

How about getting the highest and lowest range? You can then do
GetPixelColor>X,Y,COLOR
IF>{(%COLOR%>[Low Range])or(%COLOR% Greater than Low Range
< Less than High Range

Just a though.

[Rain]

Sorry it's AND not OR

Let’s say your high range is 15467 and your low range 13678 what you have to do is add 20 or so to the high range and subtract 20 or so from the low range.

The code would look like this:
GetPixelColor>X,Y,COLOR
IF>{(%COLOR%>13658) AND (%COLOR% Greater than Low Range
< Less than High Range

[Sephta]

Thanks for the hint, thougt about that as well - but:

The range can be very high, so i can tell nothing about a hit for a pixel. This happens every time, the first value (red) changes a bit or the numer of digits for one of the folling blue/green) changes.

And that's exactly my problem to pinpoint the right color range.

Try to get the RGB-values from this color:

color: 1213132

012 - 131 - 032 ?
121 - 031 - 032 ?
012 - 013 - 132 ?

You see, splitting the colors in RGB is quite difficult - for this example, i can't tell if it's mainly blue, green or red. But this can happen in my case.

IF I could split the value from GetPixelColot in it's RGB-Values, i could check every color channel for a small change. But the calculating from the given value is tricky - it's not a exact computable array of variables with the given information (leading zeroes would be fine )

[Me_again]

This not very elegant code converts a macroscheduler color number (px) to RGB. (I'm sure Marcus will now post a single line solution )

GetPixelColor>%xpos%,%ypos%,px
Let>px2={%px% MOD 16}
Let>px3={(%px%-%px2%)/16}
Let>px4={%px3% MOD 16}
Let>red={%px2%+(%px4%*16)}
Let>px5={(%px3%-%px4%)/16}
Let>px6={%px5% MOD 16}
Let>px7={(%px5%-%px6%)/16}
Let>px8={%px7% MOD 16}
Let>green={%px6%+(%px8%*16)}
Let>px9={(%px7%-%px8%)/16}
Let>px10={%px9% MOD 16}
Let>px11={(%px9%-%px10%)/16}
Let>px12={%px11% MOD 16}
Let>blue={%px10%+(%px12%*16)}

mdl>%red%-%green%-%blue%

[Sephta]

Thanks for the hint.

So the number I get for the pixel is truely a decimal conversion for the three hex values written in a row? If yes, i will get a algorythm, thanks

[Marcus Tettmar]

Just remember that Macro Scheduler returns the values in BGR sequence (not RGB).

[Me_again]

The code as posted has been tested against several color picking tools such as colorpic and returns the same Red Green Blue values.

[Aaron]

http://www.mjtnet.com/forum/viewtopic.p ... eft+corner

[Me_again]

Huh?