一篇关于图标处理的短文，蛮不错，翻译一下，共享 ^_^
[可能有些地方翻译得不好，请见谅 :( ]
[转载请说明出处，http://www.xhlv.com/blog ，翻译不容易啊]
以下是译文：
---------------------------------------------------------------------------------------------------------
在这个冠冕堂皇的标题之下，我想回答一个土方，有人在博客提出很久了。该土方很好的描述了很多图标设计师所面临的典型问题。
这里是那个问题：
"你好，我也是做图标的，我有一个问题：如何合适地缩小一个图标？我做的大图标清晰而且漂亮，但缩小的图标就变得肮脏和讨厌了，我很想纠正这种情形"
在缩小图标的过程中，至少有两点。首先是和人们所能感知的具体特点细节有关，另外是和屏幕分辨率有关。
让我们开始第一项。
对一个大图标（或其他图像） ，我们潜意识对物体的大小进行分类，选出大型，中型和小型。对较小的图标，我们可以更粗略地划分：大（涉及图标大小）和小的物体。小的图标只有很微小的局部，如果一个设计师没有作出特别努力来帮助用户区分它们，那么所有内容都会被混合成一个密集点。
以一个抽象图形为例。在大的版本中，我们可以分辨出3大对象（圆，三角形和正方形） ，然后几个中型物体和一些小件物品。
单纯地缩小图形，我们只得到一塌糊涂的图形。
为了修正这种情形，我们需要清理图形中所有琐碎细节，只留下基本形状（大中型物体） 。
我们已经除去了细小的东西，而现在我们面临着第二个挑战：恢复图标上的大中型物体间的从属关系。事实上，当降低图标大小，我们减少了大中型物体的差异，因此我们要故意地夸大来弥合这一差距。
对大画面，我们清楚地看到要素间的关系和从属：三绿圆被蓝色条纹相连接。
那些会在缩小图形时消失，因而我们会看到的只是一个三角形。
放大圆和缩小条纹，强调元素间的差异会恢复最初的元素层次。
所以，请记住：摆脱琐碎，并强调从属地位。
现在让我们谈谈电脑屏幕功能，这是一个小图标创造中困难工作的一个基本原因。像素格。
为什么清晰漂亮的图标变得混浊和乏味？这是因为像素格把图标隐藏在折痕中，失去了所有的细节，虽然你已经那么狂热得画了这些细节。所以图标缩小之后，就会像"犯罪周刊"电视节目里演得那样。
你怎么办？
你已经做了一件事：你摆脱了无论如何将陷入湮没的小细节。你已经做了另一件事：强调基本部分，因此现在这些部分获得了更多的像素空间。
第三：注意，你现在不仅仅是画一幅画，实际上是从方形像素框建立起来的。显然，你对水平和垂直线没有问题，也是方形的形状。但其他的图形就不会那么容易画了，你研究严格的几何越深，你就会发现你的困难也越艰巨。它会在正面投影中呈现出来，这正是为什么建议把图标以这种方式转换。并且你一定要调整的所有物体的边到像素格。
现在我将以一个活生生的例子证明上面所说的一切：
一个叫 Hedwig 的英俊猫头鹰同意协助我。
现在，我们有一个128 × 128大小的大的白猫头鹰，然后进行下面工作：缩小到48 × 48 ，然后到16 × 16 。什么是我们的常识忠告？ " 将它变换比例到所需最小值" -细语那些喜欢复杂而巧妙的短语。那些的熟用photoshop人说 "按下ctrl+T，并按住shift键缩小" 。 ?Seize and squeeze!? -对那些喜欢简单而快速的人巧辩。
那么，让我们跟随它的建议，并看看结果。
尽管事实上48 × 48的版本似乎是可以接受的，但是还有一定的改进余地。
甜美的。
但目前16 × 16版让人觉得是猫头鹰粪而非一个猫头鹰本身，因此把它重新涂画一遍。
这是由于它使我们相当不知羞耻地看到人的眼睛里去。
Hedwig 飞，飞！飞向第七部哈利波特，我不会告诉那里什么东西等着你，到时候你就知道了。
-------------------------------------------------------------------------------------------------------------------------
以下是原文：
Quoted from Unkown:
Dribs and drabs
Under this pompous title I'd like to answer a quiestion that was raised in the blog a long time ago. The quiestion is well formulated and describes the typical problem faced by many icon designers.
Here it is:
"Hello, I also do icons, and I have a question: how to reduce an icon competently? I do large ones clear and beautiful, but small ones are dirty and horrible, I would like very much to rectify the situation"
In the process of reducing icons there are at least two points. One of the them is concerned with specific characteristics of human perception, and the other has to do with screen resolution.
Let's begin with thr first one.
On a large icon (or other image), we subconsciously sort the sizes of objects, picking out large, medium and small ones. On a smaller icon, we can divide more roughly: large (in relation to the icon size) and small objects. Bur a tiny icon has tiny part only, and if a designer does not make special efforts to help users to classify them, then all will be mixed into one dense spot.
Take an abstract image. On a large version we can point out three large objects (a circle, a triangle and a square), then several medium-sized objects and some small items.
Simply reducing the image, we receive only mess.
To fix the situatuation, we should clear up all the trivial details from the picture and leave only basic shapes (large and medium objects).
We got rid of the small stuff, and now we are faced with a second challenge: to restore the subordination between large and medium-sized objects on the icon. Indeed, lowering the icon size, we reduce differences between them, therefore we have to close that gap with intentional hypertrophying
On a large picture we clearly see the relationship and subordination of elements: three green circles are connected with blue stripes.
Those accents disapper when reducing the image, so we're starting to see just a triangle.
It's possible to restore the initial hierarchy by stressing the difference between elements: enlarge the circles and reduce the stripes.
So, remember: get rid of trivia and emphasize subordination.
Now let's turn to the computer screen feature that is a basic reason of the most difficult works in the creation of a small icon. Pixel grid.
Why does the clear and beautiful icon becomes turbid and stupid? Because pixel grid, taking the icon tucked into its fold, eliminates all the details, you've drawn with such enthusiasm. So after that the icon is just like a "Crimes weekly" TV show.
What do you do?
You've already done one thing: you got rid of small details which would sink into oblivion anyway. You've already done another one: emphasized basic parts, so now they take more pixel space for themselves.
Third: note that you are not just painting a picture now, but actually building it from the square blocks. Obviously, you'll have no problems with horizontal and vertical lines as well as square shapes. But everything else won't be so easy to draw and the farther you go from the strict geometry, the more difficult you'll find your difficult difficulties. That's why it's recommended to turn the icon in such way it would present itself in a frontal projection. And surely you need to adjust all the object edges to the pixel grid.
Now I'll demonstrate everything said above on a living example:
A handsome owl named Hedwig agreed to assist me.
So, we have a large snowy owl in 128×128 size and a following task: reduce it into first 48×48 and then into 16×16. What does our common sense advises? "Transform it proportionally to the minimization of the required values" – it whispers to those who prefer complex and clever phrases. "Hit Ctrl-T and reduce with Shift key pressed", — it says to those who's friendly with Photoshop. ?Seize and squeeze!? — it declaims for those who likes it simple and fast.
Well, let's follow it's suggestions and see the result.
Despite the fact that the 48×48 version seems to be acceptable, there is a certain room for improvement.
Sweet.
But the current 16×16 version reminds more of owl droppings than of an owl itself, therefore draw it from a scratch.
This is the result which allows us look into people's eyes rather shamelessly.
Fly Hedwig, fly! Fly into the seventh part of Harry Potter, I won't tell what's there waiting for you, you'll learn it when time comes!
