http://www.blog.lessrain.com/?page_id=573
I sort of get it, but it's beyond my ken. Impressive nonetheless.
yer looks pretty cool suspect the code just loops through the pixels and then makes a judgement based on surrounding which way to draw the line? What is cunning though is not the details of how it is done, but getting the concept/idea in the first place that is clever.
Cool little app, with some very neat results.
p.s. I don't get Daniel Johnston. Sorry.
Originally posted by: JLM yer looks pretty cool suspect the code just loops through the pixels and then makes a judgement based on surrounding which way to draw the line? What is cunning though is not the details of how it is done, but getting the concept/idea in the first place that is clever. I'm guessing it's something like that, with length and rotation values based on rgb so you get the brush depth look. I still haven't figured out how to recreate a bmp looping through a pixel grid but have played around with bitmapdata enough to do thergb grab for variables at least.... I can post some code if anyone's interested. Or wants to explain to me how to make a pixel grid the same size as the original bmp with looping...
Not up to speed on all the latest flash enough to answer that, but from what you just said maybe you missed a trick here...
Brief Suggestion Take a small sample calculate direction and draw a lineTo then cashe as bitmap and repeat. Use onEnterFrame to limit the number of iterations per frame this will stop the processing from ever crashing.
Reasons Small sample processing (my term) means that you are only ever working with small samples of data, like fractals this is very efficient as the computer only gets the data it needs to do one iteration. LineTo would be very quick to render since it is done at byte level by flash rather than at actionscript level if that makes sense. Calculating the pixels and drawing them per pixel would be slow. Using cashe as bitmap means you can throw away the vector information which can be much heavier than the pixel equivelent.
More Indepth 1) If you make a bitmapData object of just a part of the picture say 16x16 square (Make 16x16 bitmap data and then draw image to the object with correct position/starting offset).
2) Loop around the edge of the 16x16 square and find the nearest color match to the centre (break the RGB's into 3 numbers and calculate the average of the sum of differences push into an array of {difference and position}, and sort on the objects smallest difference). Use the 16x16 position of the smallest differences in a lookup object that contains offset distances for drawing the lineTo eg: lineTo(center.x,center._y, centerX+offset[ difference[0] ]._x,center._y+offset[ difference[0] ]._y) Rather than manually calculating these offsets, they can be calculated prior to running the main algrorithm.
3) Draw the line on a black movie and repeat step one at a new position.
Goes off to code such - realises on red wine - relaxes back into seat wishes was dancing salsa than home alone without a tele.
As for code yer post anything you have, googling under 'Flash bitmapData Array' seems quite good.
Thought I would revive this, but the link ari posted is 404. Any examples of something similar?
what was the original link doing? What were you looking to do baron?
I found the thread too late in the game to know what the link was, but from the discussion it sounds like it was interesting. I'm not looking for anything in particular - simply digging the idea of playing with pixels. 
Found the original post though not the full post here:http://www.blog.lessrain.com/blurbs-and-blurbs-bits-and-bits-pieces-and-pieces/#comments