Starting at square one with image processing

[Wilhelm Sanke]

  James Hurley jhurley0305 at sbcglobal.net
  <mailto:use-revolution%40lists.runrev.com?Subject=Starting%20at%20square%20one%20with%20image%20processing&In-Reply-To=20070706134327.98049488F6E%40mail.runrev.com>

wrote:

>I've discovered that I have no idea what's going on with the  
>imagedata function.
>
>I thought that the imagedata was a pixel by pixel map of the image. I  
>tried the following handler to look at each pixel of my image:



Allow me to add some information to the recommendations you already got from this list.

The basic format for imagedata manipulation I use in my applications is like shown below.
Treating the image as a square enables you also to access and change pixels according to 
their positions, i.e. you could select single pixels or rectangular parts of the picture
 and manipulate only them. Changing the imagedata for an entire image would look like here:


 "put the imageData of image "x" into iData
 put the height of img "x" into theight
 put the width of img "x" into twidth
 put 4* twidth into ro  # "ro" for "row": There are 4 chars to each pixel in a row.
  
  repeat with i = 0 to theight - 1
     repeat with j = 0 to twidth - 1
#get the color values
      put (chartonum(char (i*ro + j* 4 + 2) of idata)) into tRed
      put (chartonum(char (i*ro + j* 4 + 3) of idata)) into tGreen
      put (chartonum(char (i*ro + j* 4 + 4) of idata)) into tBlue
#=============================
# now do whatever you want to change the values of tRed, TGreen, and tBlue
# and then put them back into the imagedata
#==================================
     put numtochar(tRed) into char  (i*ro + j* 4 + 2) of idata
     put numtochar(tGreen) into char (i*ro + j* 4 + 3) of idata
     put numtochar(tBlue) into char  (i*ro + j* 4 + 4) of idata       
    end repeat
  end repeat
  set the imageData of image "x" to iData"

I also use the "for each" format in my scripts, but this is not necessarily faster. 
I have even found cases where "for each" is slower than the double repeat loop.

To speed up script execution it is advisable to simplify the computation inside 
the loops - especially for larger images. Such a simplified script is not as 
readable as the basic one, but it is indeed faster.
Thus, the repeat loop from above could be changed to:

"repeat with i = 0 to theight - 1
    put i*ro into ti
     repeat with j = 0 to twidth - 1
      put ti + j*4 into tij
#get the color values
      put (chartonum(char tij+ 2) of idata)) into tRed
      put (chartonum(char (tij+ 3) of idata)) into tGreen
      put (chartonum(char (tij+ 4) of idata)) into tBlue
#=============================
# now do whatever you want to change the values of tRed, TGreen, and tBlue
# and then put them back into the imagedata
#==================================
     put numtochar(tRed) into char  (tij+ 2) of idata
     put numtochar(tGreen) into char (tij+ 3) of idata
     put numtochar(tBlue) into char  (tij+ 4) of idata       
    end repeat
  end repeat"

To produce a gray-scale image based on the red values (this looks much better
 than computing the average value of tRed, tGreen, tBlue) the inner part of 
the script examples above would be

" put (chartonum(char tij+ 2) of idata)) into tRed
  put numtochar(tRed) into char (tij+ 3) of idata
  put numtochar(tRed) into char  (tij+ 4) of idata"       

In my various "imagedata" stacks there are quite a number of accessible scripts
 that you could change and experiment with - about hundred in 
stack "Imagedata toolkit 2 preview".

<http://www.sanke.org/Software/ImagedataToolkitPreview3.zip>
<http://www.sanke.org/Software/SeamlessTiles.zip>

A new imagedata stack that can handle images of any size will be released soon.

Regards,

Wilhelm Sanke
<http://www.sanke.org/MetaMedia>