I want to implement various effects like Sepia, GrayScale, Posterization etc. on BitmapData in ActionScript-3.
How to do this and What is most efficient way to do this ?
Thanks in advance!!
The most efficient way to do image processing would be to use Adobe Pixel Bender. This is basically a pixel shader like language that can be used to create filters for Photoshop, After Effects and Flash. There is also an exchange where developers share filters.
Here is a tutorial on how to load pixel bender filter into flash.
There is more information available if you search Google for "pixel bender flash".
This post explains how to do the GrayScale:
AS3:How to change a colored Bitmap's BitmapData to black and white?
Sepia is a combination of grayscale and tinting. Tinting is typically accomplished by multiplying one of the colour channels by a certain amount to boost that colour.
The techniques illustrated in that post are the way you should proceed. You'll have to look up the per-pixel filtering algorithms, but many of them exist on wikipedia or math sites.
Related
Is it possible to add a glow to the outline of an Image? I know I can do this by adding an additional "glowed" version of the image and switching to it when desired, but since there are many different shapes in my game, I would prefer the glow to be done programmatically. How can I do this?
As #Metaphore mentioned, shaders are indeed the best option I found. I've succeeded in adding an outline to any desired image in my game by following this article and by getting crucial information from asking a follow up question.
The only way to do this is to use your own pixel shader when drawing this particular image.
You can find a lot of glow effect shaders on the net and there are may tutorials how to use them as well. You may check official LibGDX article on this topic https://github.com/libgdx/libgdx/wiki/Shaders
However, I'd not recommend you follow that way, because using different shaders for drawing single images will make you render cycle code much more complicated and less optimal. So either you will compromise on it, or just find easier way to achieve such behavior without shaders (I mean draw it statically somehow).
I need a library which, fed with a bitmap, returns me an array of rectangles with coordinates and dimensions of the different areas found in the image.
I'll give a graphic example:
From this:
I want this:
Or from this:
I want this:
Is there such a library?
If I want to write one on my own where can I start to inform myself about it?
To my knowledge, the best you'll find are image filters, and color conversion methods, but not the kind of complicated edge detection you're looking for.
Of course, your query supersedes the canny edge detection, and is focused on image boundaries, but I've found no material on that even beyond AS3.
I'm making a Stage3D space game: www.sugarspook.com/darkmatters/demo.html,
and I'd like to add 'glows' to some objects.
Is there a way to do this solely in the AGAL shaders, or do I need to pre-render the glowing objects and 'blur' them?
Yeah, you will need to prerender to a texture first. Here's what i would do:
Prerender an object to texture.
Downsample multiple times to a desired size (more downsamples = bigger glow).
Apply some kind of blur (I prefer Gaussian).
Tint blurred texture with some color.
Upsample.
Render.
You can take a look at the example I posted in my blog. It's for Alternativa3D but may be useful as the principle is the same everywhere. Here's github link to whole lib.
I'm working with a few programming buddies to create an AS interface to the kinect and one problem we're running into is image differencing. We need to be able to throw out image data that doesn't change from image to image so we can pin-point only things that are moving(i.e. people).
Anyone have any experience with this or a direction we can go?
I would consider creating a pixel bender shader to find the difference and also do any other math or tracking. Pixel bender gets its own thread outside of the normal flash player so you can get some more horse power for your setup. Pixel Bender shaders can be applied to bitmaps, vectors, or video so I think it is perfect for this project. Good Luck!
http://www.adobe.com/devnet/flash/articles/pixel_bender_basics.html
And is is a full collection of shaders including difference
Take a look at the threshold method on BitmapData.
It'll allow you to do this stuff. Their docs have a simple example so check that out.
It might be a long shot, and this is just me rambling, but in sound theory (strange how I'm relating it to image cancellation, but here goes...) the concept of cancellation is when you take a wave sample and add its inverse. It's how you make acapellas from instrumentals + originals or instrumentals from acapellas + originals.
Perhaps you can invert the new image and "normalize" the two to get your offsets? I.e. the first image is 'black on white' and the second image is 'white on black', and then detect the differences for the bitmap data. I know I did a similar method of finding collisions with AS3 a few years back. This would, in theory, cancel out any 'repeating' pixels and leave you with just the changes from the last frame.
With BitmapData your values are going to be from 0 to 255, so if you can implement a cancellation (because a lot of parts of the image are going to stay the same from frame t frame) then you can easily find the changes from the previous frame.
Just a thought! Whatever your solution is it's going to have to be quick in order to beat the flash runtimes' slow speeds. Your Kinect read FPS rate will be greatly hindered with bad code.
Here is some frame differencing code I wrote awhile back. It uses bitmapData: http://actionsnippet.com/?p=2820
I also used this to capture moving colors in a video feed: http://actionsnippet.com/?p=2736
I am a junior developer I can't understand how canvas tag is beneficial for us?
I read lot of articles on that but I can't get the root benefit getting from the canvas tag.
Think of the difference between canvas and svg as the difference betwee Photoshop and Illustrator (or Gimp and Inkscape for you OSS folks). One deals with bitmaps and the other vector art.
With canvas, since you are drawing in bitmap, you can smudge, blur, burn, dodge your images easily. But since it's bitmap you can't easily draw a line and then decide to reposition the line. You need to delete the old line and then draw a new line.
With svg, since you are drawing vectors, you can easily move, scale, rotate, reposition, flip your drawings. But since it's vectors you can't easily blur the edges according to line thickness or seamlessly meld a red circle into a blue square. You need to simulate blurring by drawing intermediate polygons between objects.
Sometimes their use case overlaps. Like a lot of people use canvas to do simple line drawings and keep track of the objects as data structures in javascript. But really, they both serve different purposes. If you try to implement general purpose vector drawing in pure javascript on top of canvas I doubt you'd be faster than using svg which is most likely implemented in C.
Basically, thanks to canvas, we can now draw/render 2D shapes using HTML5 and the canvas API.
As an example of what's possible now with canvas, see this
Some possible uses for Canvas:
Image drawing program
Photo editing/manipulation
2D Games
Advanced image viewing such as Microsoft's Deep Zoom
If you can't understand how it's beneficial, then maybe it isn't from your point of view at least. Don't think that because it's there I have to use it somehow, pick and choose what technologies work for you based on what you're trying to build, an Accounting web app probably wouldn't need a canvas for instance.
The canvas will enable you to draw pixel perfect graphics.
The cool projects that came to mind for me are:
Visualize gps data. GPS data is just an XML list of coordinates. You could easily build something in canvas to "connect the dots".
An mobile app where the user can actual sign a document with her finger - canvas allows you to export out the rendered canvas drawing to PNG where it can be saved on the server.
In a game where you have avatars, you can allow the user to actual draw on the avatar. Moustaches anyone?
Other stuff:
In iOS / Android using lots of CSS3
effects like box-shadow can lead to
poor performance, especially when
animating. You can do a lot of these
graphics in a single canvas tag,
here's an example:
http://everytimezone.com/. This thing is flawless on an ipad.
Cool background effects. For example try going to Paul Irish's
site and move your cursor around the
background: http://paulirish.com/
In this HTML5 book sponsored by Google a lot of the effects are using
canvas:
http://www.20thingsilearned.com/ -
particularly the "page flip"
animations.
my personal take on canvas (and when I actually found a use case for canvas) is the ability to color pick and color change per pixel in a canvas element - actually moving the image from something we don't have any information about what is happening inside it to an element like all other DOM elements (and yes, I know about the current problems with canvas and DOM - sure this would be taken care of in the future)
sure - canvas made some sort of animation easier and pluginless, but that we could do before (mostly with flash) - I think the real importance is the ability to know what is happening on the page.