当前位置：文江博客话题详情

在 Pixel Bender for Flash/Flex 中模拟嵌套循环

发布于 2024-11-26 03:05:53 字数 558 浏览 1 评论 0 原文

我有两个向量，其点数很大，但任意（v1.length 不需要等于 v2.length），我想将它们成对相乘。因为这是 AS3 代码大循环中的主要瓶颈，所以我想我应该尝试提取代码并通过 Pixel Bender 异步运行它。现在，为了表示数据，我将有两个 image2 变量作为输入。

如何从一张图像中获取像素并用另一张图像的每个像素进行计算？我刚刚学习 Pixel Bender，所以我可能忽略了一些循环的解决方法。

编辑：也许我需要澄清一下。假设我有这样的东西

var v1:Vector.<Point> = ...;
var v2:Vector.<Point> = ...;
var result:Vector.<Point> = ...;
for (var i:int = 0; i < v1.length; ++i)
    for (var j:int = 0; j < v2.length; ++j)
        result[i] += v1[i] * v2[j];

这是一个嵌套循环 - 我如何在 Pixel Bender 中模拟它？

原文

I have two vectors of a large, but arbitrary (v1.length need not equal v2.length), number of points, and I want to pairwise multiply them. Because this is the main bottleneck in a large loop of AS3 code, I thought I would try to pull out the code and run it async via Pixel Bender. Now, to represent the data I would have two image2 variables as input.

How can I get a pixel from one image and do the calculations with each pixel of the other image? I am just learning Pixel Bender, so I may have overlooked some work around for looping.

EDIT: Perhaps I need to clarify. Let's say I have something like this

var v1:Vector.<Point> = ...;
var v2:Vector.<Point> = ...;
var result:Vector.<Point> = ...;
for (var i:int = 0; i < v1.length; ++i)
    for (var j:int = 0; j < v2.length; ++j)
        result[i] += v1[i] * v2[j];

This is a nested loop--how can I emulate it in Pixel Bender?

分享到QQ

分享到微博

如果你对这篇内容有疑问，欢迎到本站社区发帖提问参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。

发布评论

需要登录才能够评论，你可以免费注册一个本站的账号。

り繁华旳梦境 2024-12-03 03:05:53

闪光灯像素弯曲器中禁止循环和嵌套循环（内置循环除外）。

http://forums.adobe.com/thread/840318

http://blog.leeburrows.com/2011/02/pixelbender-filters-3/

我想例外的是 Pixelbender 3D，它处于测试阶段，设计用于与鼹鼠山一起使用蜜蜂。

编辑

以下是 Adobe 官方文档的链接，明确确认了上述内容：

http://wwwimages.adobe.com/www.adobe.com/content/dam/Adobe/en/devnet/pixelbender/pdfs/pixelbender_guide.pdf

从第 17 页：

Pixel Bender Flash 播放器不支持：
->除了 if 和 else 之外的循环或控制结构。

当我发现自己的情况时，这对我来说真是一个无赖，我认为这是 Adobe 的一个重大失败。

回复收藏 0 原文

零崎曲识 2024-12-03 03:05:53

像素弯曲器内核中的主要函数是一个循环，并且会针对内核评估的每个像素进行回调。以下是有关如何准确实现您想要的操作（使用多个输入）的教程的链接。

http://www.adobe.com/devnet/pixelbender/articles/creating_effects_pt09 .html#articlecontentAdobe_numberedheader

本质上，它只是定义两个输入：

<languageVersion : 1.0;>

kernel blendy
<   namespace : "com.adobe.devnet.pixelbender";
    vendor : "Kevin's Filters";
    version : 1;
    description : "mashes two inputs together";
> 
{
    input image4 src;  //Input image 1 as image4 (RGBA)
    input image4 src2; //Input image 2 as image4 (RGBA)
    output pixel4 dst; //Single pixel data type/represents single pixel value (RGBA)

    void evaluatePixel()
    {
       dst = sampleNearest(src,outCoord());
    }
}

请注意，sampleNearest 的两个参数是源图像，以及要采样的像素的坐标。我相信 outCoord() 只是循环内的当前像素。如前所述，evaludatePixel（据我理解）对于输入中存在的每个像素被调用一次。这是上述代码的修改版本（来自链接），它同时读取两个输入的值：

<languageVersion : 1.0;>

kernel blendy
<   namespace : "com.adobe.devnet.pixelbender";
    vendor : "Kevin's Filters";
    version : 1;
    description : "mashes two inputs together";
> 
{
    input image4 src;  //Input image 1 as image4 (RGBA)
    input image4 src2; //Input image 2 as image4 (RGBA)
    output pixel4 dst; //Output image

    void evaluatePixel()
    {
       dst = sampleNearest(src2,outCoord()) + sampleNearest(src, outCoord());
    }
}

这里有两个视频教程，将详细解释像素如何工作：

http://gotoandlearn.com/play.php?id=83

http://gotoandlearn.com/play.php?id=84

http://www.gotoandlearn.com/play.php?id=95

The main function inside a pixel bender kernel is a loop, and is called back for every single pixel being evaluated by the kernel. Here is a link to a tutorial on how do to exactly what you want (working with multiple inputs).

http://www.adobe.com/devnet/pixelbender/articles/creating_effects_pt09.html#articlecontentAdobe_numberedheader

Essentially it just comes down to defining two inputs:

<languageVersion : 1.0;>

kernel blendy
<   namespace : "com.adobe.devnet.pixelbender";
    vendor : "Kevin's Filters";
    version : 1;
    description : "mashes two inputs together";
> 
{
    input image4 src;  //Input image 1 as image4 (RGBA)
    input image4 src2; //Input image 2 as image4 (RGBA)
    output pixel4 dst; //Single pixel data type/represents single pixel value (RGBA)

    void evaluatePixel()
    {
       dst = sampleNearest(src,outCoord());
    }
}

Note that the two parameters of sampleNearest are the source image, and the coordinates of the pixel to sample. outCoord() I believe is simply the current pixel within the loop. As mentioned, evaludatePixel (to my understanding) is called once per pixel that exists in the input. Here is a modified version of the above code (from the link) that is reading the value of both inputs at the same time:

<languageVersion : 1.0;>

kernel blendy
<   namespace : "com.adobe.devnet.pixelbender";
    vendor : "Kevin's Filters";
    version : 1;
    description : "mashes two inputs together";
> 
{
    input image4 src;  //Input image 1 as image4 (RGBA)
    input image4 src2; //Input image 2 as image4 (RGBA)
    output pixel4 dst; //Output image

    void evaluatePixel()
    {
       dst = sampleNearest(src2,outCoord()) + sampleNearest(src, outCoord());
    }
}

Here are two video tutorials that will explain much more about how pixel works at length:

http://gotoandlearn.com/play.php?id=83

http://gotoandlearn.com/play.php?id=84

http://www.gotoandlearn.com/play.php?id=95

回复收藏 0 原文

~没有更多了~